cli: Implement help command

Signed-off-by: Vijay Bellur <vijay@gluster.com> Signed-off-by: Anand V. Avati <avati@dev.gluster.com> BUG: 1229 () URL: http://bugs.gluster.com/cgi-bin/bugzilla3/show_bug.cgi?id=1229
author: Vijay Bellur <vijay@gluster.com> 2010-08-17 03:37:46 +0000
committer: Anand V. Avati <avati@dev.gluster.com> 2010-08-17 02:23:59 -0700
commit: e5737546071bd1d38430503bb64b9998bd7870dd (patch)
tree: c5a9fe82f2474ff6c8792381c1b77384c9f82983 /cli/src/cli-cmd.h
parent: ead92558ee8ec71442b2fa2da1a0df197ee78467 (diff)
1 files changed, 1 insertions, 0 deletions
diff --git a/cli/src/cli-cmd.h b/cli/src/cli-cmd.h
index d62309de23c..b325cfdfccc 100644
--- a/cli/src/cli-cmd.h
+++ b/cli/src/cli-cmd.h
@@ -30,6 +30,7 @@
 struct cli_cmd {
         const char     *pattern;
         cli_cmd_cbk_t  *cbk;
+        const char     *desc;
 };
 
 int cli_cmd_volume_register (struct cli_state *state);
diff --git a/xlators/Makefile.am b/xlators/Makefile.am
index 29549db724e..ef20cbb64fa 100644
--- a/xlators/Makefile.am
+++ b/xlators/Makefile.am
@@ -2,10 +2,10 @@ if BUILD_GNFS
   GNFS_DIR = nfs
 endif
 
-DIST_SUBDIRS = cluster storage protocol performance debug features encryption \
+DIST_SUBDIRS = cluster storage protocol performance debug features \
           mount nfs mgmt system playground meta
 
-SUBDIRS = cluster storage protocol performance debug features encryption \
+SUBDIRS = cluster storage protocol performance debug features \
           mount ${GNFS_DIR} mgmt system playground meta
 
 EXTRA_DIST = xlator.sym
diff --git a/xlators/cluster/afr/src/afr-common.c b/xlators/cluster/afr/src/afr-common.c
index 231a7970928..032ab5c8001 100644
--- a/xlators/cluster/afr/src/afr-common.c
+++ b/xlators/cluster/afr/src/afr-common.c
@@ -15,22 +15,20 @@
 #include <stdlib.h>
 #include <signal.h>
 
-#include "glusterfs.h"
+#include <glusterfs/glusterfs.h>
 #include "afr.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
-#include "byte-order.h"
-#include "statedump.h"
-#include "events.h"
-#include "upcall-utils.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/hashfn.h>
+#include <glusterfs/list.h>
+#include <glusterfs/call-stub.h>
+#include <glusterfs/defaults.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/statedump.h>
+#include <glusterfs/events.h>
+#include <glusterfs/upcall-utils.h>
 
 #include "afr-inode-read.h"
 #include "afr-inode-write.h"
@@ -47,6 +45,56 @@ afr_quorum_errno(afr_private_t *priv)
     return ENOTCONN;
 }
 
+gf_boolean_t
+afr_is_private_directory(afr_private_t *priv, uuid_t pargfid, const char *name,
+                         pid_t pid)
+{
+    if (!__is_root_gfid(pargfid)) {
+        return _gf_false;
+    }
+
+    if (strcmp(name, GF_REPLICATE_TRASH_DIR) == 0) {
+        /*For backward compatibility /.landfill is private*/
+        return _gf_true;
+    }
+
+    if (pid == GF_CLIENT_PID_GSYNCD) {
+        /*geo-rep needs to create/sync private directory on slave because
+         * it appears in changelog*/
+        return _gf_false;
+    }
+
+    if (pid == GF_CLIENT_PID_GLFS_HEAL || pid == GF_CLIENT_PID_SELF_HEALD) {
+        if (strcmp(name, priv->anon_inode_name) == 0) {
+            /* anonymous-inode dir is private*/
+            return _gf_true;
+        }
+    } else {
+        if (strncmp(name, AFR_ANON_DIR_PREFIX, strlen(AFR_ANON_DIR_PREFIX)) ==
+            0) {
+            /* anonymous-inode dir prefix is private for geo-rep to work*/
+            return _gf_true;
+        }
+    }
+
+    return _gf_false;
+}
+
+void
+afr_fill_success_replies(afr_local_t *local, afr_private_t *priv,
+                         unsigned char *replies)
+{
+    int i = 0;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (local->replies[i].valid && local->replies[i].op_ret == 0) {
+            replies[i] = 1;
+        } else {
+            replies[i] = 0;
+        }
+    }
+}
+
 int
 afr_fav_child_reset_sink_xattrs(void *opaque);
 
@@ -56,6 +104,581 @@ afr_fav_child_reset_sink_xattrs_cbk(int ret, call_frame_t *frame, void *opaque);
 static void
 afr_discover_done(call_frame_t *frame, xlator_t *this);
 
+int
+afr_dom_lock_acquire_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                         int op_ret, int op_errno, dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    afr_private_t *priv = this->private;
+    int i = (long)cookie;
+
+    local->cont.lk.dom_lock_op_ret[i] = op_ret;
+    local->cont.lk.dom_lock_op_errno[i] = op_errno;
+    if (op_ret < 0) {
+        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_LK_HEAL_DOM,
+               "%s: Failed to acquire %s on %s",
+               uuid_utoa(local->fd->inode->gfid), AFR_LK_HEAL_DOM,
+               priv->children[i]->name);
+    } else {
+        local->cont.lk.dom_locked_nodes[i] = 1;
+    }
+
+    syncbarrier_wake(&local->barrier);
+
+    return 0;
+}
+
+int
+afr_dom_lock_acquire(call_frame_t *frame)
+{
+    afr_local_t *local = NULL;
+    afr_private_t *priv = NULL;
+    struct gf_flock flock = {
+        0,
+    };
+    int i = 0;
+
+    priv = frame->this->private;
+    local = frame->local;
+    local->cont.lk.dom_locked_nodes = GF_CALLOC(
+        priv->child_count, sizeof(*local->cont.lk.locked_nodes),
+        gf_afr_mt_char);
+    if (!local->cont.lk.dom_locked_nodes) {
+        return -ENOMEM;
+    }
+    local->cont.lk.dom_lock_op_ret = GF_CALLOC(
+        priv->child_count, sizeof(*local->cont.lk.dom_lock_op_ret),
+        gf_afr_mt_int32_t);
+    if (!local->cont.lk.dom_lock_op_ret) {
+        return -ENOMEM; /* CALLOC'd members are freed in afr_local_cleanup. */
+    }
+    local->cont.lk.dom_lock_op_errno = GF_CALLOC(
+        priv->child_count, sizeof(*local->cont.lk.dom_lock_op_errno),
+        gf_afr_mt_int32_t);
+    if (!local->cont.lk.dom_lock_op_errno) {
+        return -ENOMEM; /* CALLOC'd members are freed in afr_local_cleanup. */
+    }
+    flock.l_type = F_WRLCK;
+
+    AFR_ONALL(frame, afr_dom_lock_acquire_cbk, finodelk, AFR_LK_HEAL_DOM,
+              local->fd, F_SETLK, &flock, NULL);
+
+    if (!afr_has_quorum(local->cont.lk.dom_locked_nodes, frame->this, NULL))
+        goto blocking_lock;
+
+    /*If any of the bricks returned EAGAIN, we still need blocking locks.*/
+    if (AFR_COUNT(local->cont.lk.dom_locked_nodes, priv->child_count) !=
+        priv->child_count) {
+        for (i = 0; i < priv->child_count; i++) {
+            if (local->cont.lk.dom_lock_op_ret[i] == -1 &&
+                local->cont.lk.dom_lock_op_errno[i] == EAGAIN)
+                goto blocking_lock;
+        }
+    }
+
+    return 0;
+
+blocking_lock:
+    afr_dom_lock_release(frame);
+    AFR_ONALL(frame, afr_dom_lock_acquire_cbk, finodelk, AFR_LK_HEAL_DOM,
+              local->fd, F_SETLKW, &flock, NULL);
+    if (!afr_has_quorum(local->cont.lk.dom_locked_nodes, frame->this, NULL)) {
+        afr_dom_lock_release(frame);
+        return -afr_quorum_errno(priv);
+    }
+
+    return 0;
+}
+
+int
+afr_dom_lock_release_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                         int op_ret, int op_errno, dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    afr_private_t *priv = this->private;
+    int i = (long)cookie;
+
+    if (op_ret < 0) {
+        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_LK_HEAL_DOM,
+               "%s: Failed to release %s on %s", local->loc.path,
+               AFR_LK_HEAL_DOM, priv->children[i]->name);
+    }
+    local->cont.lk.dom_locked_nodes[i] = 0;
+
+    syncbarrier_wake(&local->barrier);
+
+    return 0;
+}
+
+void
+afr_dom_lock_release(call_frame_t *frame)
+{
+    afr_local_t *local = NULL;
+    afr_private_t *priv = NULL;
+    unsigned char *locked_on = NULL;
+    struct gf_flock flock = {
+        0,
+    };
+
+    local = frame->local;
+    priv = frame->this->private;
+    locked_on = local->cont.lk.dom_locked_nodes;
+    if (AFR_COUNT(locked_on, priv->child_count) == 0)
+        return;
+    flock.l_type = F_UNLCK;
+
+    AFR_ONLIST(locked_on, frame, afr_dom_lock_release_cbk, finodelk,
+               AFR_LK_HEAL_DOM, local->fd, F_SETLK, &flock, NULL);
+
+    return;
+}
+
+static void
+afr_lk_heal_info_cleanup(afr_lk_heal_info_t *info)
+{
+    if (!info)
+        return;
+    if (info->xdata_req)
+        dict_unref(info->xdata_req);
+    if (info->fd)
+        fd_unref(info->fd);
+    GF_FREE(info->locked_nodes);
+    GF_FREE(info->child_up_event_gen);
+    GF_FREE(info->child_down_event_gen);
+    GF_FREE(info);
+}
+
+static int
+afr_add_lock_to_saved_locks(call_frame_t *frame, xlator_t *this)
+{
+    afr_private_t *priv = this->private;
+    afr_local_t *local = frame->local;
+    afr_lk_heal_info_t *info = NULL;
+    afr_fd_ctx_t *fd_ctx = NULL;
+    int ret = -ENOMEM;
+
+    info = GF_CALLOC(sizeof(*info), 1, gf_afr_mt_lk_heal_info_t);
+    if (!info) {
+        goto cleanup;
+    }
+    INIT_LIST_HEAD(&info->pos);
+    info->fd = fd_ref(local->fd);
+    info->cmd = local->cont.lk.cmd;
+    info->pid = frame->root->pid;
+    info->flock = local->cont.lk.user_flock;
+    info->xdata_req = dict_copy_with_ref(local->xdata_req, NULL);
+    if (!info->xdata_req) {
+        goto cleanup;
+    }
+    info->lk_owner = frame->root->lk_owner;
+    info->locked_nodes = GF_MALLOC(
+        sizeof(*info->locked_nodes) * priv->child_count, gf_afr_mt_char);
+    if (!info->locked_nodes) {
+        goto cleanup;
+    }
+    memcpy(info->locked_nodes, local->cont.lk.locked_nodes,
+           sizeof(*info->locked_nodes) * priv->child_count);
+    info->child_up_event_gen = GF_CALLOC(sizeof(*info->child_up_event_gen),
+                                         priv->child_count, gf_afr_mt_int32_t);
+    if (!info->child_up_event_gen) {
+        goto cleanup;
+    }
+    info->child_down_event_gen = GF_CALLOC(sizeof(*info->child_down_event_gen),
+                                           priv->child_count,
+                                           gf_afr_mt_int32_t);
+    if (!info->child_down_event_gen) {
+        goto cleanup;
+    }
+
+    LOCK(&local->fd->lock);
+    {
+        fd_ctx = __afr_fd_ctx_get(local->fd, this);
+        if (fd_ctx)
+            fd_ctx->lk_heal_info = info;
+    }
+    UNLOCK(&local->fd->lock);
+    if (!fd_ctx) {
+        goto cleanup;
+    }
+
+    LOCK(&priv->lock);
+    {
+        list_add_tail(&info->pos, &priv->saved_locks);
+    }
+    UNLOCK(&priv->lock);
+
+    return 0;
+cleanup:
+    gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_LK_HEAL_DOM,
+           "%s: Failed to add lock to healq",
+           uuid_utoa(local->fd->inode->gfid));
+    if (info) {
+        afr_lk_heal_info_cleanup(info);
+        if (fd_ctx) {
+            LOCK(&local->fd->lock);
+            {
+                fd_ctx->lk_heal_info = NULL;
+            }
+            UNLOCK(&local->fd->lock);
+        }
+    }
+    return ret;
+}
+
+static int
+afr_remove_lock_from_saved_locks(afr_local_t *local, xlator_t *this)
+{
+    afr_private_t *priv = this->private;
+    struct gf_flock flock = local->cont.lk.user_flock;
+    afr_lk_heal_info_t *info = NULL;
+    afr_fd_ctx_t *fd_ctx = NULL;
+    int ret = -EINVAL;
+
+    fd_ctx = afr_fd_ctx_get(local->fd, this);
+    if (!fd_ctx || !fd_ctx->lk_heal_info) {
+        goto out;
+    }
+
+    info = fd_ctx->lk_heal_info;
+    if ((info->flock.l_start != flock.l_start) ||
+        (info->flock.l_whence != flock.l_whence) ||
+        (info->flock.l_len != flock.l_len)) {
+        /*TODO: Compare lkowners too.*/
+        goto out;
+    }
+
+    LOCK(&priv->lock);
+    {
+        list_del(&fd_ctx->lk_heal_info->pos);
+    }
+    UNLOCK(&priv->lock);
+
+    afr_lk_heal_info_cleanup(info);
+    fd_ctx->lk_heal_info = NULL;
+    ret = 0;
+out:
+    if (ret)
+        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_LK_HEAL_DOM,
+               "%s: Failed to remove lock from healq",
+               uuid_utoa(local->fd->inode->gfid));
+    return ret;
+}
+
+int
+afr_lock_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                  int32_t op_ret, int32_t op_errno, struct gf_flock *lock,
+                  dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    int i = (long)cookie;
+
+    local->replies[i].valid = 1;
+    local->replies[i].op_ret = op_ret;
+    local->replies[i].op_errno = op_errno;
+    if (op_ret != 0) {
+        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_LK_HEAL_DOM,
+               "Failed to heal lock on child %d for %s", i,
+               uuid_utoa(local->fd->inode->gfid));
+    }
+    syncbarrier_wake(&local->barrier);
+    return 0;
+}
+
+int
+afr_getlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
+              int32_t op_errno, struct gf_flock *lock, dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    int i = (long)cookie;
+
+    local->replies[i].valid = 1;
+    local->replies[i].op_ret = op_ret;
+    local->replies[i].op_errno = op_errno;
+    if (op_ret != 0) {
+        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_LK_HEAL_DOM,
+               "Failed getlk for %s", uuid_utoa(local->fd->inode->gfid));
+    } else {
+        local->cont.lk.getlk_rsp[i] = *lock;
+    }
+
+    syncbarrier_wake(&local->barrier);
+    return 0;
+}
+
+static gf_boolean_t
+afr_does_lk_owner_match(call_frame_t *frame, afr_private_t *priv,
+                        afr_lk_heal_info_t *info)
+{
+    int i = 0;
+    afr_local_t *local = frame->local;
+    struct gf_flock flock = {
+        0,
+    };
+    gf_boolean_t ret = _gf_true;
+    char *wind_on = alloca0(priv->child_count);
+    unsigned char *success_replies = alloca0(priv->child_count);
+    local->cont.lk.getlk_rsp = GF_CALLOC(sizeof(*local->cont.lk.getlk_rsp),
+                                         priv->child_count, gf_afr_mt_gf_lock);
+
+    flock = info->flock;
+    for (i = 0; i < priv->child_count; i++) {
+        if (info->locked_nodes[i])
+            wind_on[i] = 1;
+    }
+
+    AFR_ONLIST(wind_on, frame, afr_getlk_cbk, lk, info->fd, F_GETLK, &flock,
+               info->xdata_req);
+
+    afr_fill_success_replies(local, priv, success_replies);
+    if (AFR_COUNT(success_replies, priv->child_count) == 0) {
+        ret = _gf_false;
+        goto out;
+    }
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (!local->replies[i].valid || local->replies[i].op_ret != 0)
+            continue;
+        if (local->cont.lk.getlk_rsp[i].l_type == F_UNLCK)
+            continue;
+        /*TODO: Do we really need to compare lkowner if F_UNLCK is true?*/
+        if (!is_same_lkowner(&local->cont.lk.getlk_rsp[i].l_owner,
+                             &info->lk_owner)) {
+            ret = _gf_false;
+            break;
+        }
+    }
+out:
+    afr_local_replies_wipe(local, priv);
+    GF_FREE(local->cont.lk.getlk_rsp);
+    local->cont.lk.getlk_rsp = NULL;
+    return ret;
+}
+
+static void
+afr_mark_fd_bad(fd_t *fd, xlator_t *this)
+{
+    afr_fd_ctx_t *fd_ctx = NULL;
+
+    if (!fd)
+        return;
+    LOCK(&fd->lock);
+    {
+        fd_ctx = __afr_fd_ctx_get(fd, this);
+        if (fd_ctx) {
+            fd_ctx->is_fd_bad = _gf_true;
+            fd_ctx->lk_heal_info = NULL;
+        }
+    }
+    UNLOCK(&fd->lock);
+}
+
+static void
+afr_add_lock_to_lkhealq(afr_private_t *priv, afr_lk_heal_info_t *info)
+{
+    LOCK(&priv->lock);
+    {
+        list_del(&info->pos);
+        list_add_tail(&info->pos, &priv->lk_healq);
+    }
+    UNLOCK(&priv->lock);
+}
+
+static void
+afr_lock_heal_do(call_frame_t *frame, afr_private_t *priv,
+                 afr_lk_heal_info_t *info)
+{
+    int i = 0;
+    int op_errno = 0;
+    int32_t *current_event_gen = NULL;
+    afr_local_t *local = frame->local;
+    xlator_t *this = frame->this;
+    char *wind_on = alloca0(priv->child_count);
+    gf_boolean_t retry = _gf_true;
+
+    frame->root->pid = info->pid;
+    lk_owner_copy(&frame->root->lk_owner, &info->lk_owner);
+
+    op_errno = -afr_dom_lock_acquire(frame);
+    if ((op_errno != 0)) {
+        goto release;
+    }
+
+    if (!afr_does_lk_owner_match(frame, priv, info)) {
+        gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_LK_HEAL_DOM,
+               "Ignoring lock heal for %s since lk-onwers mismatch. "
+               "Lock possibly pre-empted by another client.",
+               uuid_utoa(info->fd->inode->gfid));
+        goto release;
+    }
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (info->locked_nodes[i])
+            continue;
+        wind_on[i] = 1;
+    }
+
+    current_event_gen = alloca(priv->child_count);
+    memcpy(current_event_gen, info->child_up_event_gen,
+           priv->child_count * sizeof *current_event_gen);
+    AFR_ONLIST(wind_on, frame, afr_lock_heal_cbk, lk, info->fd, info->cmd,
+               &info->flock, info->xdata_req);
+
+    LOCK(&priv->lock);
+    {
+        for (i = 0; i < priv->child_count; i++) {
+            if (!wind_on[i])
+                continue;
+            if ((!local->replies[i].valid) || (local->replies[i].op_ret != 0)) {
+                continue;
+            }
+
+            if ((current_event_gen[i] == info->child_up_event_gen[i]) &&
+                (current_event_gen[i] > info->child_down_event_gen[i])) {
+                info->locked_nodes[i] = 1;
+                retry = _gf_false;
+                list_del_init(&info->pos);
+                list_add_tail(&info->pos, &priv->saved_locks);
+            } else {
+                /*We received subsequent child up/down events while heal was in
+                 * progress; don't mark child as healed. Attempt again on the
+                 * new child up*/
+                gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_LK_HEAL_DOM,
+                       "Event gen mismatch: skipped healing lock on child %d "
+                       "for %s.",
+                       i, uuid_utoa(info->fd->inode->gfid));
+            }
+        }
+    }
+    UNLOCK(&priv->lock);
+
+release:
+    afr_dom_lock_release(frame);
+    if (retry)
+        afr_add_lock_to_lkhealq(priv, info);
+    return;
+}
+
+static int
+afr_lock_heal_done(int ret, call_frame_t *frame, void *opaque)
+{
+    STACK_DESTROY(frame->root);
+    return 0;
+}
+
+static int
+afr_lock_heal(void *opaque)
+{
+    call_frame_t *frame = (call_frame_t *)opaque;
+    call_frame_t *iter_frame = NULL;
+    xlator_t *this = frame->this;
+    afr_private_t *priv = this->private;
+    afr_lk_heal_info_t *info = NULL;
+    afr_lk_heal_info_t *tmp = NULL;
+    struct list_head healq = {
+        0,
+    };
+    int ret = 0;
+
+    iter_frame = afr_copy_frame(frame);
+    if (!iter_frame) {
+        return ENOMEM;
+    }
+
+    INIT_LIST_HEAD(&healq);
+    LOCK(&priv->lock);
+    {
+        list_splice_init(&priv->lk_healq, &healq);
+    }
+    UNLOCK(&priv->lock);
+
+    list_for_each_entry_safe(info, tmp, &healq, pos)
+    {
+        GF_ASSERT((AFR_COUNT(info->locked_nodes, priv->child_count) <
+                   priv->child_count));
+        ((afr_local_t *)(iter_frame->local))->fd = fd_ref(info->fd);
+        afr_lock_heal_do(iter_frame, priv, info);
+        AFR_STACK_RESET(iter_frame);
+        if (iter_frame->local == NULL) {
+            ret = ENOTCONN;
+            gf_msg(frame->this->name, GF_LOG_ERROR, ENOTCONN,
+                   AFR_MSG_LK_HEAL_DOM,
+                   "Aborting processing of lk_healq."
+                   "Healing will be reattempted on next child up for locks "
+                   "that are still in quorum.");
+            LOCK(&priv->lock);
+            {
+                list_add_tail(&healq, &priv->lk_healq);
+            }
+            UNLOCK(&priv->lock);
+            break;
+        }
+    }
+
+    AFR_STACK_DESTROY(iter_frame);
+    return ret;
+}
+
+static int
+__afr_lock_heal_synctask(xlator_t *this, afr_private_t *priv, int child)
+{
+    int ret = 0;
+    call_frame_t *frame = NULL;
+    afr_lk_heal_info_t *info = NULL;
+    afr_lk_heal_info_t *tmp = NULL;
+
+    if (priv->shd.iamshd)
+        return 0;
+
+    list_for_each_entry_safe(info, tmp, &priv->saved_locks, pos)
+    {
+        info->child_up_event_gen[child] = priv->event_generation;
+        list_del_init(&info->pos);
+        list_add_tail(&info->pos, &priv->lk_healq);
+    }
+
+    frame = create_frame(this, this->ctx->pool);
+    if (!frame)
+        return -1;
+
+    ret = synctask_new(this->ctx->env, afr_lock_heal, afr_lock_heal_done, frame,
+                       frame);
+    if (ret)
+        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, AFR_MSG_LK_HEAL_DOM,
+               "Failed to launch lock heal synctask");
+
+    return ret;
+}
+
+static int
+__afr_mark_pending_lk_heal(xlator_t *this, afr_private_t *priv, int child)
+{
+    afr_lk_heal_info_t *info = NULL;
+    afr_lk_heal_info_t *tmp = NULL;
+
+    if (priv->shd.iamshd)
+        return 0;
+    list_for_each_entry_safe(info, tmp, &priv->saved_locks, pos)
+    {
+        info->child_down_event_gen[child] = priv->event_generation;
+        if (info->locked_nodes[child] == 1)
+            info->locked_nodes[child] = 0;
+        if (!afr_has_quorum(info->locked_nodes, this, NULL)) {
+            /* Since the lock was lost on quorum no. of nodes, we should
+             * not attempt to heal it anymore. Some other client could have
+             * acquired the lock, modified data and released it and this
+             * client wouldn't know about it if we heal it.*/
+            afr_mark_fd_bad(info->fd, this);
+            list_del(&info->pos);
+            afr_lk_heal_info_cleanup(info);
+            /* We're not winding an unlock on the node where the lock is still
+             * present because when fencing logic switches over to the new
+             * client (since we marked the fd bad), it should preempt any
+             * existing lock. */
+        }
+    }
+    return 0;
+}
+
 gf_boolean_t
 afr_is_consistent_io_possible(afr_local_t *local, afr_private_t *priv,
                               int32_t *op_errno)
@@ -70,6 +693,19 @@ afr_is_consistent_io_possible(afr_local_t *local, afr_private_t *priv,
     return _gf_true;
 }
 
+gf_boolean_t
+afr_is_lock_mode_mandatory(dict_t *xdata)
+{
+    int ret = 0;
+    uint32_t lk_mode = GF_LK_ADVISORY;
+
+    ret = dict_get_uint32(xdata, GF_LOCK_MODE, &lk_mode);
+    if (!ret && lk_mode == GF_LK_MANDATORY)
+        return _gf_true;
+
+    return _gf_false;
+}
+
 call_frame_t *
 afr_copy_frame(call_frame_t *base)
 {
@@ -98,21 +734,24 @@ afr_is_possibly_under_txn(afr_transaction_type type, afr_local_t *local,
     int tmp = 0;
     afr_private_t *priv = NULL;
     GF_UNUSED char *key = NULL;
+    int keylen = 0;
 
     priv = this->private;
 
-    if (type == AFR_ENTRY_TRANSACTION)
+    if (type == AFR_ENTRY_TRANSACTION) {
         key = GLUSTERFS_PARENT_ENTRYLK;
-    else if (type == AFR_DATA_TRANSACTION)
+        keylen = SLEN(GLUSTERFS_PARENT_ENTRYLK);
+    } else if (type == AFR_DATA_TRANSACTION) {
         /*FIXME: Use GLUSTERFS_INODELK_DOM_COUNT etc. once
          * pl_inodelk_xattr_fill supports separate keys for different
          * domains.*/
         key = GLUSTERFS_INODELK_COUNT;
-
+        keylen = SLEN(GLUSTERFS_INODELK_COUNT);
+    }
     for (i = 0; i < priv->child_count; i++) {
         if (!local->replies[i].xdata)
             continue;
-        if (dict_get_int32(local->replies[i].xdata, key, &tmp) == 0)
+        if (dict_get_int32n(local->replies[i].xdata, key, keylen, &tmp) == 0)
             if (tmp)
                 return _gf_true;
     }
@@ -260,11 +899,7 @@ __afr_set_in_flight_sb_status(xlator_t *this, afr_local_t *local,
 
     count = gf_bits_count(tmp_map);
 
-    if (count == 1)
-        index = gf_bits_index(tmp_map);
-
     for (i = 0; i < priv->child_count; i++) {
-        mask = 0;
         if (!local->transaction.failed_subvols[i])
             continue;
 
@@ -278,25 +913,27 @@ __afr_set_in_flight_sb_status(xlator_t *this, afr_local_t *local,
     switch (txn_type) {
         case AFR_METADATA_TRANSACTION:
             if ((metadatamap_old != 0) && (metadatamap == 0) && (count == 1)) {
+                index = gf_bits_index(tmp_map);
                 local->transaction.in_flight_sb_errno = local->replies[index]
                                                             .op_errno;
                 local->transaction.in_flight_sb = _gf_true;
                 metadatamap |= (1 << index);
             }
             if (metadatamap_old != metadatamap) {
-                event = 0;
+                __afr_inode_need_refresh_set(inode, this);
             }
             break;
 
         case AFR_DATA_TRANSACTION:
             if ((datamap_old != 0) && (datamap == 0) && (count == 1)) {
+                index = gf_bits_index(tmp_map);
                 local->transaction.in_flight_sb_errno = local->replies[index]
                                                             .op_errno;
                 local->transaction.in_flight_sb = _gf_true;
                 datamap |= (1 << index);
             }
             if (datamap_old != datamap)
-                event = 0;
+                __afr_inode_need_refresh_set(inode, this);
             break;
 
         default:
@@ -460,34 +1097,6 @@ out:
 }
 
 int
-__afr_inode_event_gen_reset_small(inode_t *inode, xlator_t *this)
-{
-    int ret = -1;
-    uint16_t datamap = 0;
-    uint16_t metadatamap = 0;
-    uint32_t event = 0;
-    uint64_t val = 0;
-    afr_inode_ctx_t *ctx = NULL;
-
-    ret = __afr_inode_ctx_get(this, inode, &ctx);
-    if (ret)
-        return ret;
-
-    val = ctx->read_subvol;
-
-    metadatamap = (val & 0x000000000000ffff) >> 0;
-    datamap = (val & 0x00000000ffff0000) >> 16;
-    event = 0;
-
-    val = ((uint64_t)metadatamap) | (((uint64_t)datamap) << 16) |
-          (((uint64_t)event) << 32);
-
-    ctx->read_subvol = val;
-
-    return ret;
-}
-
-int
 __afr_inode_read_subvol_get(inode_t *inode, xlator_t *this, unsigned char *data,
                             unsigned char *metadata, int *event_p)
 {
@@ -558,22 +1167,6 @@ out:
 }
 
 int
-__afr_inode_event_gen_reset(inode_t *inode, xlator_t *this)
-{
-    afr_private_t *priv = NULL;
-    int ret = -1;
-
-    priv = this->private;
-
-    if (priv->child_count <= 16)
-        ret = __afr_inode_event_gen_reset_small(inode, this);
-    else
-        ret = -1;
-
-    return ret;
-}
-
-int
 afr_inode_read_subvol_get(inode_t *inode, xlator_t *this, unsigned char *data,
                           unsigned char *metadata, int *event_p)
 {
@@ -639,12 +1232,11 @@ afr_inode_get_readable(call_frame_t *frame, inode_t *inode, xlator_t *this,
     return 0;
 }
 
-int
+static int
 afr_inode_split_brain_choice_get(inode_t *inode, xlator_t *this,
                                  int *spb_choice)
 {
     int ret = -1;
-
     GF_VALIDATE_OR_GOTO(this->name, inode, out);
 
     LOCK(&inode->lock);
@@ -656,6 +1248,40 @@ out:
     return ret;
 }
 
+/*
+ * frame is used to get the favourite policy. Since
+ * afr_inode_split_brain_choice_get was called with afr_open, it is possible to
+ * have a frame with out local->replies. So in that case, frame is passed as
+ * null, hence this function will handle the frame NULL case.
+ */
+int
+afr_split_brain_read_subvol_get(inode_t *inode, xlator_t *this,
+                                call_frame_t *frame, int *spb_subvol)
+{
+    int ret = -1;
+    afr_local_t *local = NULL;
+    afr_private_t *priv = NULL;
+
+    GF_VALIDATE_OR_GOTO("afr", this, out);
+    GF_VALIDATE_OR_GOTO(this->name, this->private, out);
+    GF_VALIDATE_OR_GOTO(this->name, inode, out);
+    GF_VALIDATE_OR_GOTO(this->name, spb_subvol, out);
+
+    priv = this->private;
+
+    ret = afr_inode_split_brain_choice_get(inode, this, spb_subvol);
+    if (*spb_subvol < 0 && priv->fav_child_policy && frame && frame->local) {
+        local = frame->local;
+        *spb_subvol = afr_sh_get_fav_by_policy(this, local->replies, inode,
+                                               NULL);
+        if (*spb_subvol >= 0) {
+            ret = 0;
+        }
+    }
+
+out:
+    return ret;
+}
 int
 afr_inode_read_subvol_set(inode_t *inode, xlator_t *this, unsigned char *data,
                           unsigned char *metadata, int event)
@@ -722,30 +1348,22 @@ out:
     return need_refresh;
 }
 
-static int
-afr_inode_need_refresh_set(inode_t *inode, xlator_t *this)
+int
+__afr_inode_need_refresh_set(inode_t *inode, xlator_t *this)
 {
     int ret = -1;
     afr_inode_ctx_t *ctx = NULL;
 
-    GF_VALIDATE_OR_GOTO(this->name, inode, out);
-
-    LOCK(&inode->lock);
-    {
-        ret = __afr_inode_ctx_get(this, inode, &ctx);
-        if (ret)
-            goto unlock;
-
+    ret = __afr_inode_ctx_get(this, inode, &ctx);
+    if (ret == 0) {
         ctx->need_refresh = _gf_true;
     }
-unlock:
-    UNLOCK(&inode->lock);
-out:
+
     return ret;
 }
 
 int
-afr_inode_event_gen_reset(inode_t *inode, xlator_t *this)
+afr_inode_need_refresh_set(inode_t *inode, xlator_t *this)
 {
     int ret = -1;
 
@@ -753,7 +1371,7 @@ afr_inode_event_gen_reset(inode_t *inode, xlator_t *this)
 
     LOCK(&inode->lock);
     {
-        ret = __afr_inode_event_gen_reset(inode, this);
+        ret = __afr_inode_need_refresh_set(inode, this);
     }
     UNLOCK(&inode->lock);
 out:
@@ -773,6 +1391,7 @@ afr_spb_choice_timeout_cancel(xlator_t *this, inode_t *inode)
     {
         ret = __afr_inode_ctx_get(this, inode, &ctx);
         if (ret < 0 || !ctx) {
+            UNLOCK(&inode->lock);
             gf_msg(this->name, GF_LOG_WARNING, 0,
                    AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR,
                    "Failed to cancel split-brain choice timer.");
@@ -785,8 +1404,8 @@ afr_spb_choice_timeout_cancel(xlator_t *this, inode_t *inode)
         }
         ret = 0;
     }
-out:
     UNLOCK(&inode->lock);
+out:
     return ret;
 }
 
@@ -818,7 +1437,6 @@ afr_set_split_brain_choice(int ret, call_frame_t *frame, void *opaque)
     gf_boolean_t timer_set = _gf_false;
     gf_boolean_t timer_cancelled = _gf_false;
     gf_boolean_t timer_reset = _gf_false;
-    gf_boolean_t need_invalidate = _gf_true;
     int old_spb_choice = -1;
 
     frame = data->frame;
@@ -862,10 +1480,11 @@ afr_set_split_brain_choice(int ret, call_frame_t *frame, void *opaque)
     {
         ret = __afr_inode_ctx_get(this, inode, &ctx);
         if (ret) {
+            UNLOCK(&inode->lock);
             gf_msg(this->name, GF_LOG_ERROR, 0,
                    AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR,
                    "Failed to get inode_ctx for %s", loc->name);
-            goto unlock;
+            goto post_unlock;
         }
 
         old_spb_choice = ctx->spb_choice;
@@ -929,10 +1548,10 @@ afr_set_split_brain_choice(int ret, call_frame_t *frame, void *opaque)
             timer_set = _gf_true;
         if (timer_reset && !ctx->timer)
             timer_cancelled = _gf_true;
-        need_invalidate = _gf_false;
     }
 unlock:
     UNLOCK(&inode->lock);
+post_unlock:
     if (!timer_set)
         inode_unref(inode);
     if (timer_cancelled)
@@ -942,8 +1561,7 @@ unlock:
      * reads from an older cached value despite a change in spb_choice to
      * a new value.
      */
-    if (need_invalidate)
-        inode_invalidate(inode);
+    inode_invalidate(inode);
 out:
     GF_FREE(data);
     AFR_STACK_UNWIND(setxattr, frame, ret, op_errno, NULL);
@@ -988,7 +1606,7 @@ afr_accuse_smallfiles(xlator_t *this, struct afr_reply *replies,
 
     for (i = 0; i < priv->child_count; i++) {
         if (replies[i].valid && replies[i].xdata &&
-            dict_get(replies[i].xdata, GLUSTERFS_BAD_INODE))
+            dict_get_sizen(replies[i].xdata, GLUSTERFS_BAD_INODE))
             continue;
         if (data_accused[i])
             continue;
@@ -1037,7 +1655,7 @@ afr_readables_fill(call_frame_t *frame, xlator_t *this, inode_t *inode,
         if (replies) { /* Lookup */
             if (!replies[i].valid || replies[i].op_ret == -1 ||
                 (replies[i].xdata &&
-                 dict_get(replies[i].xdata, GLUSTERFS_BAD_INODE))) {
+                 dict_get_sizen(replies[i].xdata, GLUSTERFS_BAD_INODE))) {
                 data_readable[i] = 0;
                 metadata_readable[i] = 0;
                 continue;
@@ -1050,6 +1668,8 @@ afr_readables_fill(call_frame_t *frame, xlator_t *this, inode_t *inode,
             ia_type = inode->ia_type;
         }
 
+        if (!xdata)
+            continue; /* mkdir_cbk sends NULL xdata_rsp. */
         afr_accused_fill(this, xdata, data_accused,
                          (ia_type == IA_IFDIR) ? AFR_ENTRY_TRANSACTION
                                                : AFR_DATA_TRANSACTION);
@@ -1153,18 +1773,12 @@ ret:
 }
 
 gf_boolean_t
-afr_selfheal_enabled(xlator_t *this)
+afr_selfheal_enabled(const xlator_t *this)
 {
-    afr_private_t *priv = NULL;
-    gf_boolean_t data = _gf_false;
-    int ret = 0;
+    const afr_private_t *priv = this->private;
 
-    priv = this->private;
-
-    ret = gf_string2boolean(priv->data_self_heal, &data);
-    GF_ASSERT(!ret);
-
-    return data || priv->metadata_self_heal || priv->entry_self_heal;
+    return priv->data_self_heal || priv->metadata_self_heal ||
+           priv->entry_self_heal;
 }
 
 int
@@ -1177,7 +1791,6 @@ afr_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
     inode_t *inode = NULL;
     int event_generation = 0;
     int read_subvol = -1;
-    int op_errno = ENOMEM;
     int ret = 0;
 
     local = frame->local;
@@ -1193,7 +1806,7 @@ afr_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
     ret = afr_inode_get_readable(frame, inode, this, local->readable,
                                  &event_generation, local->transaction.type);
 
-    if (ret == -EIO || (local->is_read_txn && !event_generation)) {
+    if (ret == -EIO) {
         /* No readable subvolume even after refresh ==> splitbrain.*/
         if (!priv->fav_child_policy) {
             err = EIO;
@@ -1206,18 +1819,12 @@ afr_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
             goto refresh_done;
         }
 
-        heal_frame = copy_frame(frame);
+        heal_frame = afr_frame_create(this, NULL);
         if (!heal_frame) {
             err = EIO;
             goto refresh_done;
         }
-        heal_frame->root->pid = GF_CLIENT_PID_SELF_HEALD;
-        heal_local = AFR_FRAME_INIT(heal_frame, op_errno);
-        if (!heal_local) {
-            err = EIO;
-            AFR_STACK_DESTROY(heal_frame);
-            goto refresh_done;
-        }
+        heal_local = heal_frame->local;
         heal_local->xdata_req = dict_new();
         if (!heal_local->xdata_req) {
             err = EIO;
@@ -1238,18 +1845,6 @@ refresh_done:
     return 0;
 }
 
-static void
-afr_fill_success_replies(afr_local_t *local, afr_private_t *priv,
-                         unsigned char *replies)
-{
-    int i = 0;
-
-    for (i = 0; i < priv->child_count; i++) {
-        if (local->replies[i].valid && local->replies[i].op_ret == 0)
-            replies[i] = 1;
-    }
-}
-
 int
 afr_inode_refresh_done(call_frame_t *frame, xlator_t *this, int error)
 {
@@ -1259,7 +1854,6 @@ afr_inode_refresh_done(call_frame_t *frame, xlator_t *this, int error)
     gf_boolean_t start_heal = _gf_false;
     afr_local_t *heal_local = NULL;
     unsigned char *success_replies = NULL;
-    int op_errno = ENOMEM;
     int ret = 0;
 
     if (error != 0) {
@@ -1271,32 +1865,32 @@ afr_inode_refresh_done(call_frame_t *frame, xlator_t *this, int error)
     success_replies = alloca0(priv->child_count);
     afr_fill_success_replies(local, priv, success_replies);
 
-    if (!afr_has_quorum(success_replies, this)) {
-        error = afr_final_errno(frame->local, this->private);
-        if (!error)
-            error = afr_quorum_errno(priv);
-        goto refresh_done;
-    }
-
     if (priv->thin_arbiter_count && local->is_read_txn &&
         AFR_COUNT(success_replies, priv->child_count) != priv->child_count) {
         /* We need to query the good bricks and/or thin-arbiter.*/
+        if (success_replies[0]) {
+            local->read_txn_query_child = AFR_CHILD_ZERO;
+        } else if (success_replies[1]) {
+            local->read_txn_query_child = AFR_CHILD_ONE;
+        }
         error = EINVAL;
         goto refresh_done;
     }
 
+    if (!afr_has_quorum(success_replies, this, frame)) {
+        error = afr_final_errno(frame->local, this->private);
+        if (!error)
+            error = afr_quorum_errno(priv);
+        goto refresh_done;
+    }
+
     ret = afr_replies_interpret(frame, this, local->refreshinode, &start_heal);
 
     if (ret && afr_selfheal_enabled(this) && start_heal) {
-        heal_frame = copy_frame(frame);
+        heal_frame = afr_frame_create(this, NULL);
         if (!heal_frame)
             goto refresh_done;
-        heal_frame->root->pid = GF_CLIENT_PID_SELF_HEALD;
-        heal_local = AFR_FRAME_INIT(heal_frame, op_errno);
-        if (!heal_local) {
-            AFR_STACK_DESTROY(heal_frame);
-            goto refresh_done;
-        }
+        heal_local = heal_frame->local;
         heal_local->refreshinode = inode_ref(local->refreshinode);
         heal_local->heal_frame = heal_frame;
         if (!afr_throttled_selfheal(heal_frame, this)) {
@@ -1333,17 +1927,22 @@ afr_inode_refresh_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (xdata)
             local->replies[call_child].xdata = dict_ref(xdata);
     }
+
     if (xdata) {
         ret = dict_get_int8(xdata, "link-count", &need_heal);
-        local->replies[call_child].need_heal = need_heal;
-    } else {
-        local->replies[call_child].need_heal = need_heal;
+        if (ret) {
+            gf_msg_debug(this->name, -ret, "Unable to get link count");
+        }
     }
 
+    local->replies[call_child].need_heal = need_heal;
     call_count = afr_frame_return(frame);
     if (call_count == 0) {
         afr_set_need_heal(this, local);
         ret = afr_inode_refresh_err(frame, this);
+        if (ret) {
+            gf_msg_debug(this->name, ret, "afr_inode_refresh_err failed");
+        }
         afr_inode_refresh_done(frame, this, ret);
     }
 }
@@ -1452,12 +2051,12 @@ afr_inode_refresh_do(call_frame_t *frame, xlator_t *this)
         return 0;
     }
 
-    ret = dict_set_str(xdata, "link-count", GF_XATTROP_INDEX_COUNT);
+    ret = dict_set_sizen_str_sizen(xdata, "link-count", GF_XATTROP_INDEX_COUNT);
     if (ret) {
         gf_msg_debug(this->name, -ret, "Unable to set link-count in dict ");
     }
 
-    ret = dict_set_str(xdata, GLUSTERFS_INODELK_DOM_COUNT, this->name);
+    ret = dict_set_str_sizen(xdata, GLUSTERFS_INODELK_DOM_COUNT, this->name);
     if (ret) {
         gf_msg_debug(this->name, -ret,
                      "Unable to set inodelk-dom-count in dict ");
@@ -1555,7 +2154,7 @@ afr_xattr_req_prepare(xlator_t *this, dict_t *xattr_req)
                      "query flag");
     }
 
-    ret = dict_set_int32(xattr_req, "list-xattr", 1);
+    ret = dict_set_int32_sizen(xattr_req, "list-xattr", 1);
     if (ret) {
         gf_msg_debug(this->name, -ret, "Unable to set list-xattr in dict ");
     }
@@ -1600,7 +2199,8 @@ afr_lookup_xattr_req_prepare(afr_local_t *local, xlator_t *this,
                GLUSTERFS_PARENT_ENTRYLK);
     }
 
-    ret = dict_set_str(local->xattr_req, "link-count", GF_XATTROP_INDEX_COUNT);
+    ret = dict_set_sizen_str_sizen(local->xattr_req, "link-count",
+                                   GF_XATTROP_INDEX_COUNT);
     if (ret) {
         gf_msg_debug(this->name, -ret, "Unable to set link-count in dict ");
     }
@@ -1611,19 +2211,18 @@ out:
 }
 
 int
-afr_least_pending_reads_child(afr_private_t *priv)
+afr_least_pending_reads_child(afr_private_t *priv, unsigned char *readable)
 {
     int i = 0;
-    int child = 0;
+    int child = -1;
     int64_t read_iter = -1;
     int64_t pending_read = -1;
 
-    pending_read = GF_ATOMIC_GET(priv->pending_reads[0]);
-    for (i = 1; i < priv->child_count; i++) {
-        if (AFR_IS_ARBITER_BRICK(priv, i))
+    for (i = 0; i < priv->child_count; i++) {
+        if (AFR_IS_ARBITER_BRICK(priv, i) || !readable[i])
             continue;
         read_iter = GF_ATOMIC_GET(priv->pending_reads[i]);
-        if (read_iter < pending_read) {
+        if (child == -1 || read_iter < pending_read) {
             pending_read = read_iter;
             child = i;
         }
@@ -1632,8 +2231,54 @@ afr_least_pending_reads_child(afr_private_t *priv)
     return child;
 }
 
+static int32_t
+afr_least_latency_child(afr_private_t *priv, unsigned char *readable)
+{
+    int32_t i = 0;
+    int child = -1;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (AFR_IS_ARBITER_BRICK(priv, i) || !readable[i] ||
+            priv->child_latency[i] < 0)
+            continue;
+
+        if (child == -1 ||
+            priv->child_latency[i] < priv->child_latency[child]) {
+            child = i;
+        }
+    }
+    return child;
+}
+
+static int32_t
+afr_least_latency_times_pending_reads_child(afr_private_t *priv,
+                                            unsigned char *readable)
+{
+    int32_t i = 0;
+    int child = -1;
+    int64_t pending_read = 0;
+    int64_t latency = -1;
+    int64_t least_latency = -1;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (AFR_IS_ARBITER_BRICK(priv, i) || !readable[i] ||
+            priv->child_latency[i] < 0)
+            continue;
+
+        pending_read = GF_ATOMIC_GET(priv->pending_reads[i]);
+        latency = (pending_read + 1) * priv->child_latency[i];
+
+        if (child == -1 || latency < least_latency) {
+            least_latency = latency;
+            child = i;
+        }
+    }
+    return child;
+}
+
 int
-afr_hash_child(afr_read_subvol_args_t *args, afr_private_t *priv)
+afr_hash_child(afr_read_subvol_args_t *args, afr_private_t *priv,
+               unsigned char *readable)
 {
     uuid_t gfid_copy = {
         0,
@@ -1642,14 +2287,14 @@ afr_hash_child(afr_read_subvol_args_t *args, afr_private_t *priv)
     int child = -1;
 
     switch (priv->hash_mode) {
-        case 0:
+        case AFR_READ_POLICY_FIRST_UP:
             break;
-        case 1:
+        case AFR_READ_POLICY_GFID_HASH:
             gf_uuid_copy(gfid_copy, args->gfid);
             child = SuperFastHash((char *)gfid_copy, sizeof(gfid_copy)) %
                     priv->child_count;
             break;
-        case 2:
+        case AFR_READ_POLICY_GFID_PID_HASH:
             if (args->ia_type != IA_IFDIR) {
                 /*
                  * Why getpid?  Because it's one of the cheapest calls
@@ -1661,14 +2306,21 @@ afr_hash_child(afr_read_subvol_args_t *args, afr_private_t *priv)
                  * need is a low probability that multiple clients
                  * won't converge on the same subvolume.
                  */
+                gf_uuid_copy(gfid_copy, args->gfid);
                 pid = getpid();
-                memcpy(gfid_copy, &pid, sizeof(pid));
+                *(pid_t *)gfid_copy ^= pid;
             }
             child = SuperFastHash((char *)gfid_copy, sizeof(gfid_copy)) %
                     priv->child_count;
             break;
-        case 3:
-            child = afr_least_pending_reads_child(priv);
+        case AFR_READ_POLICY_LESS_LOAD:
+            child = afr_least_pending_reads_child(priv, readable);
+            break;
+        case AFR_READ_POLICY_LEAST_LATENCY:
+            child = afr_least_latency_child(priv, readable);
+            break;
+        case AFR_READ_POLICY_LOAD_LATENCY_HYBRID:
+            child = afr_least_latency_times_pending_reads_child(priv, readable);
             break;
     }
 
@@ -1701,7 +2353,7 @@ afr_read_subvol_select_by_policy(inode_t *inode, xlator_t *this,
     }
 
     /* second preference - use hashed mode */
-    read_subvol = afr_hash_child(&local_args, priv);
+    read_subvol = afr_hash_child(&local_args, priv, readable);
     if (read_subvol >= 0 && readable[read_subvol])
         return read_subvol;
 
@@ -1796,11 +2448,9 @@ afr_local_transaction_cleanup(afr_local_t *local, xlator_t *this)
 
     afr_matrix_cleanup(local->pending, priv->child_count);
 
-    GF_FREE(local->internal_lock.locked_nodes);
-
     GF_FREE(local->internal_lock.lower_locked_nodes);
 
-    afr_entry_lockee_cleanup(&local->internal_lock);
+    afr_lockees_cleanup(&local->internal_lock);
 
     GF_FREE(local->transaction.pre_op);
 
@@ -2009,6 +2659,9 @@ afr_local_cleanup(afr_local_t *local, xlator_t *this)
 
     { /* lk */
         GF_FREE(local->cont.lk.locked_nodes);
+        GF_FREE(local->cont.lk.dom_locked_nodes);
+        GF_FREE(local->cont.lk.dom_lock_op_ret);
+        GF_FREE(local->cont.lk.dom_lock_op_errno);
     }
 
     { /* create */
@@ -2239,7 +2892,7 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
 {
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    int spb_choice = -1;
+    int spb_subvol = -1;
     int child_count = -1;
 
     if (*read_subvol != -1)
@@ -2249,13 +2902,15 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
     local = frame->local;
     child_count = priv->child_count;
 
-    afr_inode_split_brain_choice_get(local->inode, this, &spb_choice);
-    if ((spb_choice >= 0) &&
+    afr_split_brain_read_subvol_get(local->inode, this, frame, &spb_subvol);
+    if ((spb_subvol >= 0) &&
         (AFR_COUNT(success_replies, child_count) == child_count)) {
-        *read_subvol = spb_choice;
-    } else if (!priv->quorum_count) {
+        *read_subvol = spb_subvol;
+    } else if (!priv->quorum_count ||
+               frame->root->pid == GF_CLIENT_PID_GLFS_HEAL) {
         *read_subvol = afr_first_up_child(frame, this);
-    } else if (priv->quorum_count && afr_has_quorum(data_readable, this)) {
+    } else if (priv->quorum_count &&
+               afr_has_quorum(data_readable, this, NULL)) {
         /* read_subvol is guaranteed to be valid if we hit this path. */
         *read_subvol = afr_first_up_child(frame, this);
     } else {
@@ -2270,7 +2925,7 @@ afr_attempt_readsubvol_set(call_frame_t *frame, xlator_t *this,
                local->loc.path);
     }
     if (*read_subvol >= 0)
-        dict_del(local->replies[*read_subvol].xdata, GF_CONTENT_KEY);
+        dict_del_sizen(local->replies[*read_subvol].xdata, GF_CONTENT_KEY);
 }
 
 static void
@@ -2291,6 +2946,7 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         0,
     };
     gf_boolean_t locked_entry = _gf_false;
+    gf_boolean_t in_flight_create = _gf_false;
     gf_boolean_t can_interpret = _gf_true;
     inode_t *parent = NULL;
     ia_type_t ia_type = IA_INVAL;
@@ -2334,17 +2990,12 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         if (!replies[i].valid)
             continue;
 
-        if (locked_entry && replies[i].op_ret == -1 &&
-            replies[i].op_errno == ENOENT) {
-            /* Second, check entry is still
-               "underway" in creation */
-            local->op_ret = -1;
-            local->op_errno = ENOENT;
-            goto error;
-        }
-
-        if (replies[i].op_ret == -1)
+        if (replies[i].op_ret == -1) {
+            if (locked_entry && replies[i].op_errno == ENOENT) {
+                in_flight_create = _gf_true;
+            }
             continue;
+        }
 
         if (read_subvol == -1 || !readable[read_subvol]) {
             read_subvol = i;
@@ -2354,6 +3005,12 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         }
     }
 
+    if (in_flight_create && !afr_has_quorum(success_replies, this, NULL)) {
+        local->op_ret = -1;
+        local->op_errno = ENOENT;
+        goto error;
+    }
+
     if (read_subvol == -1)
         goto error;
     /* We now have a read_subvol, which is readable[] (if there
@@ -2382,7 +3039,8 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         /* If we were called from glfsheal and there is still a gfid
          * mismatch, succeed the lookup and let glfsheal print the
          * response via gfid-heal-msg.*/
-        if (!dict_get_str(local->xattr_req, "gfid-heal-msg", &gfid_heal_msg))
+        if (!dict_get_str_sizen(local->xattr_req, "gfid-heal-msg",
+                                &gfid_heal_msg))
             goto cant_interpret;
 
         /* LOG ERROR */
@@ -2397,7 +3055,7 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
     read_subvol = -1;
     memset(readable, 0, sizeof(*readable) * priv->child_count);
     if (can_interpret) {
-        if (!afr_has_quorum(success_replies, this))
+        if (!afr_has_quorum(success_replies, this, NULL))
             goto cant_interpret;
         /* It is safe to call afr_replies_interpret() because we have
            a response from all the UP subvolumes and all of them resolved
@@ -2411,8 +3069,8 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         if (read_subvol == -1)
             goto cant_interpret;
         if (ret) {
-            afr_inode_event_gen_reset(local->inode, this);
-            dict_del(local->replies[read_subvol].xdata, GF_CONTENT_KEY);
+            afr_inode_need_refresh_set(local->inode, this);
+            dict_del_sizen(local->replies[read_subvol].xdata, GF_CONTENT_KEY);
         }
     } else {
     cant_interpret:
@@ -2436,10 +3094,10 @@ afr_lookup_done(call_frame_t *frame, xlator_t *this)
         goto error;
     }
 
-    ret = dict_get_str(local->xattr_req, "gfid-heal-msg", &gfid_heal_msg);
+    ret = dict_get_str_sizen(local->xattr_req, "gfid-heal-msg", &gfid_heal_msg);
     if (!ret) {
-        ret = dict_set_str(local->replies[read_subvol].xdata, "gfid-heal-msg",
-                           gfid_heal_msg);
+        ret = dict_set_str_sizen(local->replies[read_subvol].xdata,
+                                 "gfid-heal-msg", gfid_heal_msg);
         if (ret) {
             gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_DICT_SET_FAILED,
                    "Error setting gfid-heal-msg dict");
@@ -2464,7 +3122,7 @@ error:
  * others in that they must be given higher priority while
  * returning to the user.
  *
- * The hierarchy is ENODATA > ENOENT > ESTALE > others
+ * The hierarchy is ENODATA > ENOENT > ESTALE > ENOSPC others
  */
 
 int
@@ -2476,6 +3134,8 @@ afr_higher_errno(int32_t old_errno, int32_t new_errno)
         return ENOENT;
     if (old_errno == ESTALE || new_errno == ESTALE)
         return ESTALE;
+    if (old_errno == ENOSPC || new_errno == ENOSPC)
+        return ENOSPC;
 
     return new_errno;
 }
@@ -2517,7 +3177,7 @@ afr_local_discovery_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     priv = this->private;
     child_index = (int32_t)(long)cookie;
 
-    ret = dict_get_str(dict, GF_XATTR_PATHINFO_KEY, &pathinfo);
+    ret = dict_get_str_sizen(dict, GF_XATTR_PATHINFO_KEY, &pathinfo);
     if (ret != 0) {
         goto out;
     }
@@ -2605,7 +3265,11 @@ afr_lookup_sh_metadata_wrap(void *opaque)
     dict = dict_new();
     if (!dict)
         goto out;
-    ret = dict_set_str(dict, "link-count", GF_XATTROP_INDEX_COUNT);
+    if (local->xattr_req) {
+        dict_copy(local->xattr_req, dict);
+    }
+
+    ret = dict_set_sizen_str_sizen(dict, "link-count", GF_XATTROP_INDEX_COUNT);
     if (ret) {
         gf_msg_debug(this->name, -ret, "Unable to set link-count in dict ");
     }
@@ -2613,7 +3277,7 @@ afr_lookup_sh_metadata_wrap(void *opaque)
     if (loc_is_nameless(&local->loc)) {
         ret = afr_selfheal_unlocked_discover_on(frame, local->inode,
                                                 local->loc.gfid, local->replies,
-                                                local->child_up);
+                                                local->child_up, dict);
     } else {
         inode = afr_selfheal_unlocked_lookup_on(frame, local->loc.parent,
                                                 local->loc.name, local->replies,
@@ -2787,7 +3451,7 @@ afr_lookup_selfheal_wrap(void *opaque)
 
     inode = afr_selfheal_unlocked_lookup_on(frame, local->loc.parent,
                                             local->loc.name, local->replies,
-                                            local->child_up, NULL);
+                                            local->child_up, local->xattr_req);
     if (inode)
         inode_unref(inode);
 
@@ -2874,7 +3538,7 @@ afr_lookup_entry_heal(call_frame_t *frame, xlator_t *this)
     if (name_state_mismatch) {
         if (!priv->quorum_count)
             goto name_heal;
-        if (!afr_has_quorum(success, this))
+        if (!afr_has_quorum(success, this, NULL))
             goto name_heal;
         if (op_errno)
             goto name_heal;
@@ -2932,7 +3596,7 @@ afr_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * with ESTALE so that a fresh lookup will be sent by the top xlator.
      * So remember it.
      */
-    if (xdata && dict_get(xdata, "gfid-changed"))
+    if (xdata && dict_get_sizen(xdata, "gfid-changed"))
         local->cont.lookup.needs_fresh_lookup = _gf_true;
 
     if (xdata) {
@@ -2962,8 +3626,8 @@ afr_discover_unwind(call_frame_t *frame, xlator_t *this)
 {
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    int op_errno = 0;
     int read_subvol = -1;
+    int ret = 0;
     unsigned char *data_readable = NULL;
     unsigned char *success_replies = NULL;
 
@@ -2976,17 +3640,19 @@ afr_discover_unwind(call_frame_t *frame, xlator_t *this)
     if (AFR_COUNT(success_replies, priv->child_count) > 0)
         local->op_ret = 0;
 
-    op_errno = afr_final_errno(frame->local, this->private);
-
     if (local->op_ret < 0) {
-        AFR_STACK_UNWIND(lookup, frame, -1, op_errno, NULL, NULL, NULL, NULL);
-        return;
+        local->op_ret = -1;
+        local->op_errno = afr_final_errno(frame->local, this->private);
+        goto error;
     }
 
-    if (!afr_has_quorum(success_replies, this))
+    if (!afr_has_quorum(success_replies, this, frame))
         goto unwind;
 
-    afr_replies_interpret(frame, this, local->inode, NULL);
+    ret = afr_replies_interpret(frame, this, local->inode, NULL);
+    if (ret) {
+        afr_inode_need_refresh_set(local->inode, this);
+    }
 
     read_subvol = afr_read_subvol_decide(local->inode, this, NULL,
                                          data_readable);
@@ -2994,11 +3660,8 @@ afr_discover_unwind(call_frame_t *frame, xlator_t *this)
 unwind:
     afr_attempt_readsubvol_set(frame, this, success_replies, data_readable,
                                &read_subvol);
-    if (read_subvol == -1) {
-        AFR_STACK_UNWIND(lookup, frame, local->op_ret, local->op_errno, NULL,
-                         NULL, NULL, NULL);
-        return;
-    }
+    if (read_subvol == -1)
+        goto error;
 
     if (AFR_IS_ARBITER_BRICK(priv, read_subvol) && local->op_ret == 0) {
         local->op_ret = -1;
@@ -3013,6 +3676,11 @@ unwind:
                      local->inode, &local->replies[read_subvol].poststat,
                      local->replies[read_subvol].xdata,
                      &local->replies[read_subvol].postparent);
+    return;
+
+error:
+    AFR_STACK_UNWIND(lookup, frame, local->op_ret, local->op_errno, NULL, NULL,
+                     NULL, NULL);
 }
 
 static int
@@ -3036,7 +3704,7 @@ afr_ta_id_file_check(void *opaque)
     this = opaque;
     priv = this->private;
 
-    ret = afr_fill_ta_loc(this, &loc);
+    ret = afr_fill_ta_loc(this, &loc, _gf_false);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to populate thin-arbiter loc for: %s.", loc.name);
@@ -3209,8 +3877,6 @@ afr_discover(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
     }
 
     if (__is_root_gfid(loc->inode->gfid)) {
-        if (!this->itable)
-            this->itable = loc->inode->table;
         if (!priv->root_inode)
             priv->root_inode = inode_ref(loc->inode);
 
@@ -3229,10 +3895,15 @@ afr_discover(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
 
     local->inode = inode_ref(loc->inode);
 
-    if (xattr_req)
+    if (xattr_req) {
         /* If xattr_req was null, afr_lookup_xattr_req_prepare() will
            allocate one for us */
-        local->xattr_req = dict_ref(xattr_req);
+        local->xattr_req = dict_copy_with_ref(xattr_req, NULL);
+        if (!local->xattr_req) {
+            op_errno = ENOMEM;
+            goto out;
+        }
+    }
 
     if (gf_uuid_is_null(loc->inode->gfid)) {
         afr_discover_do(frame, this, 0);
@@ -3242,11 +3913,7 @@ afr_discover(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
     afr_read_subvol_get(loc->inode, this, NULL, NULL, &event,
                         AFR_DATA_TRANSACTION, NULL);
 
-    if (afr_is_inode_refresh_reqd(loc->inode, this, event,
-                                  local->event_generation))
-        afr_inode_refresh(frame, this, loc->inode, NULL, afr_discover_do);
-    else
-        afr_discover_do(frame, this, 0);
+    afr_discover_do(frame, this, 0);
 
     return 0;
 out:
@@ -3268,7 +3935,6 @@ afr_lookup_do(call_frame_t *frame, xlator_t *this, int err)
 
     if (err < 0) {
         local->op_errno = err;
-        ret = -1;
         goto out;
     }
 
@@ -3279,7 +3945,6 @@ afr_lookup_do(call_frame_t *frame, xlator_t *this, int err)
                                        &local->loc);
     if (ret) {
         local->op_errno = -ret;
-        ret = -1;
         goto out;
     }
 
@@ -3344,16 +4009,15 @@ afr_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
 
     if (loc_is_nameless(loc)) {
         if (xattr_req)
-            dict_del(xattr_req, "gfid-req");
+            dict_del_sizen(xattr_req, "gfid-req");
         afr_discover(frame, this, loc, xattr_req);
         return 0;
     }
 
-    if (__is_root_gfid(loc->parent->gfid)) {
-        if (!strcmp(loc->name, GF_REPLICATE_TRASH_DIR)) {
-            op_errno = EPERM;
-            goto out;
-        }
+    if (afr_is_private_directory(this->private, loc->parent->gfid, loc->name,
+                                 frame->root->pid)) {
+        op_errno = EPERM;
+        goto out;
     }
 
     local = AFR_FRAME_INIT(frame, op_errno);
@@ -3382,18 +4046,14 @@ afr_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
         ret = dict_get_gfuuid(local->xattr_req, "gfid-req",
                               &local->cont.lookup.gfid_req);
         if (ret == 0) {
-            dict_del(local->xattr_req, "gfid-req");
+            dict_del_sizen(local->xattr_req, "gfid-req");
         }
     }
 
     afr_read_subvol_get(loc->parent, this, NULL, NULL, &event,
                         AFR_DATA_TRANSACTION, NULL);
 
-    if (afr_is_inode_refresh_reqd(loc->inode, this, event,
-                                  local->event_generation))
-        afr_inode_refresh(frame, this, loc->parent, NULL, afr_lookup_do);
-    else
-        afr_lookup_do(frame, this, 0);
+    afr_lookup_do(frame, this, 0);
 
     return 0;
 out:
@@ -3403,8 +4063,18 @@ out:
 }
 
 void
-_afr_cleanup_fd_ctx(afr_fd_ctx_t *fd_ctx)
+_afr_cleanup_fd_ctx(xlator_t *this, afr_fd_ctx_t *fd_ctx)
 {
+    afr_private_t *priv = this->private;
+
+    if (fd_ctx->lk_heal_info) {
+        LOCK(&priv->lock);
+        {
+            list_del(&fd_ctx->lk_heal_info->pos);
+        }
+        afr_lk_heal_info_cleanup(fd_ctx->lk_heal_info);
+        fd_ctx->lk_heal_info = NULL;
+    }
     GF_FREE(fd_ctx->opened_on);
     GF_FREE(fd_ctx);
     return;
@@ -3424,7 +4094,7 @@ afr_cleanup_fd_ctx(xlator_t *this, fd_t *fd)
     fd_ctx = (afr_fd_ctx_t *)(long)ctx;
 
     if (fd_ctx) {
-        _afr_cleanup_fd_ctx(fd_ctx);
+        _afr_cleanup_fd_ctx(this, fd_ctx);
     }
 
 out:
@@ -3517,13 +4187,14 @@ __afr_fd_ctx_set(xlator_t *this, fd_t *fd)
     }
 
     fd_ctx->readdir_subvol = -1;
+    fd_ctx->lk_heal_info = NULL;
 
     ret = __fd_ctx_set(fd, this, (uint64_t)(long)fd_ctx);
     if (ret)
         gf_msg_debug(this->name, 0, "failed to set fd ctx (%p)", fd);
 out:
     if (ret && fd_ctx)
-        _afr_cleanup_fd_ctx(fd_ctx);
+        _afr_cleanup_fd_ctx(this, fd_ctx);
     return ret;
 }
 
@@ -3547,11 +4218,10 @@ afr_flush_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
         } else {
             local->op_errno = op_errno;
         }
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
     if (call_count == 0)
         AFR_STACK_UNWIND(flush, frame, local->op_ret, local->op_errno,
                          local->xdata_rsp);
@@ -3647,6 +4317,7 @@ afr_flush(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
     call_stub_t *stub = NULL;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
@@ -3687,11 +4358,10 @@ afr_fsyncdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         } else {
             local->op_errno = op_errno;
         }
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
     if (call_count == 0)
         AFR_STACK_UNWIND(fsyncdir, frame, local->op_ret, local->op_errno,
                          local->xdata_rsp);
@@ -4008,7 +4678,7 @@ afr_fop_lock_done(call_frame_t *frame, xlator_t *this)
 
     if (afr_is_conflicting_lock_present(local->op_ret, local->op_errno)) {
         afr_unlock_locks_and_proceed(frame, this, lock_count);
-    } else if (priv->quorum_count && !afr_has_quorum(success, this)) {
+    } else if (priv->quorum_count && !afr_has_quorum(success, this, NULL)) {
         local->fop_lock_state = AFR_FOP_LOCK_QUORUM_FAILED;
         local->op_ret = -1;
         local->op_errno = afr_final_errno(local, priv);
@@ -4184,9 +4854,9 @@ out:
 }
 
 static int32_t
-afr_handle_inodelk(call_frame_t *frame, glusterfs_fop_t fop, const char *volume,
-                   loc_t *loc, fd_t *fd, int32_t cmd, struct gf_flock *flock,
-                   dict_t *xdata)
+afr_handle_inodelk(call_frame_t *frame, xlator_t *this, glusterfs_fop_t fop,
+                   const char *volume, loc_t *loc, fd_t *fd, int32_t cmd,
+                   struct gf_flock *flock, dict_t *xdata)
 {
     afr_local_t *local = NULL;
     int32_t op_errno = ENOMEM;
@@ -4198,8 +4868,10 @@ afr_handle_inodelk(call_frame_t *frame, glusterfs_fop_t fop, const char *volume,
     local->op = fop;
     if (loc)
         loc_copy(&local->loc, loc);
-    if (fd)
+    if (fd && (flock->l_type != F_UNLCK)) {
+        AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
         local->fd = fd_ref(fd);
+    }
 
     local->cont.inodelk.volume = gf_strdup(volume);
     if (!local->cont.inodelk.volume) {
@@ -4228,8 +4900,8 @@ int32_t
 afr_inodelk(call_frame_t *frame, xlator_t *this, const char *volume, loc_t *loc,
             int32_t cmd, struct gf_flock *flock, dict_t *xdata)
 {
-    afr_handle_inodelk(frame, GF_FOP_INODELK, volume, loc, NULL, cmd, flock,
-                       xdata);
+    afr_handle_inodelk(frame, this, GF_FOP_INODELK, volume, loc, NULL, cmd,
+                       flock, xdata);
     return 0;
 }
 
@@ -4237,15 +4909,16 @@ int32_t
 afr_finodelk(call_frame_t *frame, xlator_t *this, const char *volume, fd_t *fd,
              int32_t cmd, struct gf_flock *flock, dict_t *xdata)
 {
-    afr_handle_inodelk(frame, GF_FOP_FINODELK, volume, NULL, fd, cmd, flock,
-                       xdata);
+    afr_handle_inodelk(frame, this, GF_FOP_FINODELK, volume, NULL, fd, cmd,
+                       flock, xdata);
     return 0;
 }
 
 static int
-afr_handle_entrylk(call_frame_t *frame, glusterfs_fop_t fop, const char *volume,
-                   loc_t *loc, fd_t *fd, const char *basename, entrylk_cmd cmd,
-                   entrylk_type type, dict_t *xdata)
+afr_handle_entrylk(call_frame_t *frame, xlator_t *this, glusterfs_fop_t fop,
+                   const char *volume, loc_t *loc, fd_t *fd,
+                   const char *basename, entrylk_cmd cmd, entrylk_type type,
+                   dict_t *xdata)
 {
     afr_local_t *local = NULL;
     int32_t op_errno = ENOMEM;
@@ -4257,8 +4930,10 @@ afr_handle_entrylk(call_frame_t *frame, glusterfs_fop_t fop, const char *volume,
     local->op = fop;
     if (loc)
         loc_copy(&local->loc, loc);
-    if (fd)
+    if (fd && (cmd != ENTRYLK_UNLOCK)) {
+        AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
         local->fd = fd_ref(fd);
+    }
     local->cont.entrylk.cmd = cmd;
     local->cont.entrylk.in_cmd = cmd;
     local->cont.entrylk.type = type;
@@ -4285,8 +4960,8 @@ afr_entrylk(call_frame_t *frame, xlator_t *this, const char *volume, loc_t *loc,
             const char *basename, entrylk_cmd cmd, entrylk_type type,
             dict_t *xdata)
 {
-    afr_handle_entrylk(frame, GF_FOP_ENTRYLK, volume, loc, NULL, basename, cmd,
-                       type, xdata);
+    afr_handle_entrylk(frame, this, GF_FOP_ENTRYLK, volume, loc, NULL, basename,
+                       cmd, type, xdata);
     return 0;
 }
 
@@ -4295,8 +4970,8 @@ afr_fentrylk(call_frame_t *frame, xlator_t *this, const char *volume, fd_t *fd,
              const char *basename, entrylk_cmd cmd, entrylk_type type,
              dict_t *xdata)
 {
-    afr_handle_entrylk(frame, GF_FOP_FENTRYLK, volume, NULL, fd, basename, cmd,
-                       type, xdata);
+    afr_handle_entrylk(frame, this, GF_FOP_FENTRYLK, volume, NULL, fd, basename,
+                       cmd, type, xdata);
     return 0;
 }
 
@@ -4308,10 +4983,10 @@ afr_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     int call_count = 0;
     struct statvfs *buf = NULL;
 
+    local = frame->local;
+
     LOCK(&frame->lock);
     {
-        local = frame->local;
-
         if (op_ret != 0) {
             local->op_errno = op_errno;
             goto unlock;
@@ -4337,10 +5012,9 @@ afr_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
         }
     }
 unlock:
+    call_count = --local->call_count;
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
     if (call_count == 0)
         AFR_STACK_UNWIND(statfs, frame, local->op_ret, local->op_errno,
                          &local->cont.statfs.buf, local->xdata_rsp);
@@ -4415,9 +5089,10 @@ afr_lk_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     }
 
     call_count = afr_frame_return(frame);
-    if (call_count == 0)
+    if (call_count == 0) {
         AFR_STACK_UNWIND(lk, frame, local->op_ret, local->op_errno, NULL,
                          local->xdata_rsp);
+    }
 
     return 0;
 }
@@ -4499,7 +5174,7 @@ afr_lk_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
                           local->cont.lk.cmd, &local->cont.lk.user_flock,
                           local->xdata_req);
     } else if (priv->quorum_count &&
-               !afr_has_quorum(local->cont.lk.locked_nodes, this)) {
+               !afr_has_quorum(local->cont.lk.locked_nodes, this, NULL)) {
         local->op_ret = -1;
         local->op_errno = afr_final_errno(local, priv);
 
@@ -4516,11 +5191,133 @@ afr_lk_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
 }
 
 int
+afr_lk_transaction_cbk(int ret, call_frame_t *frame, void *opaque)
+{
+    return 0;
+}
+
+int
+afr_lk_txn_wind_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                    int32_t op_ret, int32_t op_errno, struct gf_flock *lock,
+                    dict_t *xdata)
+{
+    afr_local_t *local = NULL;
+    int child_index = -1;
+
+    local = frame->local;
+    child_index = (long)cookie;
+    afr_common_lock_cbk(frame, cookie, this, op_ret, op_errno, xdata);
+    if (op_ret == 0) {
+        local->op_ret = 0;
+        local->op_errno = 0;
+        local->cont.lk.locked_nodes[child_index] = 1;
+        local->cont.lk.ret_flock = *lock;
+    }
+    syncbarrier_wake(&local->barrier);
+    return 0;
+}
+
+int
+afr_lk_txn_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                      int32_t op_ret, int32_t op_errno, struct gf_flock *lock,
+                      dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    afr_private_t *priv = this->private;
+    int child_index = (long)cookie;
+
+    if (op_ret < 0 && op_errno != ENOTCONN && op_errno != EBADFD) {
+        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_UNLOCK_FAIL,
+               "gfid=%s: unlock failed on subvolume %s "
+               "with lock owner %s",
+               uuid_utoa(local->fd->inode->gfid),
+               priv->children[child_index]->name,
+               lkowner_utoa(&frame->root->lk_owner));
+    }
+    return 0;
+}
+int
+afr_lk_transaction(void *opaque)
+{
+    call_frame_t *frame = NULL;
+    xlator_t *this = NULL;
+    afr_private_t *priv = NULL;
+    afr_local_t *local = NULL;
+    char *wind_on = NULL;
+    int op_errno = 0;
+    int i = 0;
+    int ret = 0;
+
+    frame = (call_frame_t *)opaque;
+    local = frame->local;
+    this = frame->this;
+    priv = this->private;
+    wind_on = alloca0(priv->child_count);
+
+    if (priv->arbiter_count || priv->child_count != 3) {
+        op_errno = ENOTSUP;
+        gf_msg(frame->this->name, GF_LOG_ERROR, op_errno, AFR_MSG_LK_HEAL_DOM,
+               "%s: Lock healing supported only for replica 3 volumes.",
+               uuid_utoa(local->fd->inode->gfid));
+        goto err;
+    }
+
+    op_errno = -afr_dom_lock_acquire(frame);  // Released during
+                                              // AFR_STACK_UNWIND
+    if (op_errno != 0) {
+        goto err;
+    }
+    if (priv->quorum_count &&
+        !afr_has_quorum(local->cont.lk.dom_locked_nodes, this, NULL)) {
+        op_errno = afr_final_errno(local, priv);
+        goto err;
+    }
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (priv->child_up[i] && local->cont.lk.dom_locked_nodes[i])
+            wind_on[i] = 1;
+    }
+    AFR_ONLIST(wind_on, frame, afr_lk_txn_wind_cbk, lk, local->fd,
+               local->cont.lk.cmd, &local->cont.lk.user_flock,
+               local->xdata_req);
+
+    if (priv->quorum_count &&
+        !afr_has_quorum(local->cont.lk.locked_nodes, this, NULL)) {
+        local->op_ret = -1;
+        local->op_errno = afr_final_errno(local, priv);
+        goto unlock;
+    } else {
+        if (local->cont.lk.user_flock.l_type == F_UNLCK)
+            ret = afr_remove_lock_from_saved_locks(local, this);
+        else
+            ret = afr_add_lock_to_saved_locks(frame, this);
+        if (ret) {
+            local->op_ret = -1;
+            local->op_errno = -ret;
+            goto unlock;
+        }
+        AFR_STACK_UNWIND(lk, frame, local->op_ret, local->op_errno,
+                         &local->cont.lk.ret_flock, local->xdata_rsp);
+    }
+
+    return 0;
+
+unlock:
+    local->cont.lk.user_flock.l_type = F_UNLCK;
+    AFR_ONLIST(local->cont.lk.locked_nodes, frame, afr_lk_txn_unlock_cbk, lk,
+               local->fd, F_SETLK, &local->cont.lk.user_flock, NULL);
+err:
+    AFR_STACK_UNWIND(lk, frame, -1, op_errno, NULL, NULL);
+    return -1;
+}
+
+int
 afr_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
        struct gf_flock *flock, dict_t *xdata)
 {
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
+    int ret = 0;
     int i = 0;
     int32_t op_errno = ENOMEM;
 
@@ -4531,9 +5328,11 @@ afr_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
         goto out;
 
     local->op = GF_FOP_LK;
-    if (!afr_lk_is_unlock(cmd, flock) &&
-        !afr_is_consistent_io_possible(local, priv, &op_errno))
-        goto out;
+    if (!afr_lk_is_unlock(cmd, flock)) {
+        AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
+        if (!afr_is_consistent_io_possible(local, priv, &op_errno))
+            goto out;
+    }
 
     local->cont.lk.locked_nodes = GF_CALLOC(
         priv->child_count, sizeof(*local->cont.lk.locked_nodes),
@@ -4551,6 +5350,16 @@ afr_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
     if (xdata)
         local->xdata_req = dict_ref(xdata);
 
+    if (afr_is_lock_mode_mandatory(xdata)) {
+        ret = synctask_new(this->ctx->env, afr_lk_transaction,
+                           afr_lk_transaction_cbk, frame, frame);
+        if (ret) {
+            op_errno = ENOMEM;
+            goto out;
+        }
+        return 0;
+    }
+
     STACK_WIND_COOKIE(frame, afr_lk_cbk, (void *)(long)0, priv->children[i],
                       priv->children[i]->fops->lk, fd, cmd, flock,
                       local->xdata_req);
@@ -4654,7 +5463,7 @@ afr_lease_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
                           priv->children[child_index]->fops->lease, &local->loc,
                           &local->cont.lease.user_lease, xdata);
     } else if (priv->quorum_count &&
-               !afr_has_quorum(local->cont.lease.locked_nodes, this)) {
+               !afr_has_quorum(local->cont.lease.locked_nodes, this, NULL)) {
         local->op_ret = -1;
         local->op_errno = afr_final_errno(local, priv);
 
@@ -4872,8 +5681,10 @@ afr_priv_dump(xlator_t *this)
                            GF_ATOMIC_GET(priv->pending_reads[i]));
         sprintf(key, "child_latency[%d]", i);
         gf_proc_dump_write(key, "%" PRId64, priv->child_latency[i]);
+        sprintf(key, "halo_child_up[%d]", i);
+        gf_proc_dump_write(key, "%d", priv->halo_child_up[i]);
     }
-    gf_proc_dump_write("data_self_heal", "%s", priv->data_self_heal);
+    gf_proc_dump_write("data_self_heal", "%d", priv->data_self_heal);
     gf_proc_dump_write("metadata_self_heal", "%d", priv->metadata_self_heal);
     gf_proc_dump_write("entry_self_heal", "%d", priv->entry_self_heal);
     gf_proc_dump_write("read_child", "%d", priv->read_child);
@@ -4884,6 +5695,7 @@ afr_priv_dump(xlator_t *this)
                        priv->background_self_heal_count);
     gf_proc_dump_write("healers", "%d", priv->healers);
     gf_proc_dump_write("read-hash-mode", "%d", priv->hash_mode);
+    gf_proc_dump_write("use-anonymous-inode", "%d", priv->use_anon_inode);
     if (priv->quorum_count == AFR_QUORUM_AUTO) {
         gf_proc_dump_write("quorum-type", "auto");
     } else if (priv->quorum_count == 0) {
@@ -4892,7 +5704,7 @@ afr_priv_dump(xlator_t *this)
         gf_proc_dump_write("quorum-type", "fixed");
         gf_proc_dump_write("quorum-count", "%d", priv->quorum_count);
     }
-    gf_proc_dump_write("up", "%u", afr_has_quorum(priv->child_up, this));
+    gf_proc_dump_write("up", "%u", afr_has_quorum(priv->child_up, this, NULL));
     if (priv->thin_arbiter_count) {
         gf_proc_dump_write("ta_child_up", "%d", priv->ta_child_up);
         gf_proc_dump_write("ta_bad_child_index", "%d",
@@ -4944,13 +5756,31 @@ __afr_get_up_children_count(afr_private_t *priv)
     return up_children;
 }
 
+static int
+__get_heard_from_all_status(xlator_t *this)
+{
+    afr_private_t *priv = this->private;
+    int i;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (!priv->last_event[i]) {
+            return 0;
+        }
+    }
+    if (priv->thin_arbiter_count && !priv->ta_child_up) {
+        return 0;
+    }
+    return 1;
+}
+
 glusterfs_event_t
-__afr_transform_event_from_state(afr_private_t *priv)
+__afr_transform_event_from_state(xlator_t *this)
 {
     int i = 0;
     int up_children = 0;
+    afr_private_t *priv = this->private;
 
-    if (AFR_COUNT(priv->last_event, priv->child_count) == priv->child_count)
+    if (__get_heard_from_all_status(this))
         /* have_heard_from_all. Let afr_notify() do the propagation. */
         return GF_EVENT_MAXVAL;
 
@@ -4992,7 +5822,7 @@ afr_notify_cbk(void *data)
             goto unlock;
         }
         priv->timer = NULL;
-        event = __afr_transform_event_from_state(priv);
+        event = __afr_transform_event_from_state(this);
         if (event != GF_EVENT_MAXVAL)
             propagate = _gf_true;
     }
@@ -5019,22 +5849,6 @@ __afr_launch_notify_timer(xlator_t *this, afr_private_t *priv)
     }
 }
 
-int
-__get_heard_from_all_status(xlator_t *this)
-{
-    afr_private_t *priv = this->private;
-    int heard_from_all = 1;
-    int i = 0;
-
-    for (i = 0; i < priv->child_count; i++) {
-        if (!priv->last_event[i]) {
-            heard_from_all = 0;
-            break;
-        }
-    }
-    return heard_from_all;
-}
-
 static int
 find_best_down_child(xlator_t *this)
 {
@@ -5046,7 +5860,7 @@ find_best_down_child(xlator_t *this)
     priv = this->private;
 
     for (i = 0; i < priv->child_count; i++) {
-        if (priv->child_up[i] && priv->child_latency[i] >= 0 &&
+        if (!priv->child_up[i] && priv->child_latency[i] >= 0 &&
             priv->child_latency[i] < best_latency) {
             best_child = i;
             best_latency = priv->child_latency[i];
@@ -5118,7 +5932,9 @@ __afr_handle_ping_event(xlator_t *this, xlator_t *child_xlator, const int idx,
                    "), "
                    "marking child down.",
                    child_latency_msec, halo_max_latency_msec);
-            *event = GF_EVENT_CHILD_DOWN;
+            if (priv->halo_child_up[idx]) {
+                *event = GF_EVENT_CHILD_DOWN;
+            }
         }
     } else if (child_latency_msec < halo_max_latency_msec &&
                priv->child_up[idx] == 0) {
@@ -5130,7 +5946,9 @@ __afr_handle_ping_event(xlator_t *this, xlator_t *child_xlator, const int idx,
                    "), "
                    "marking child up.",
                    child_latency_msec, halo_max_latency_msec);
-            *event = GF_EVENT_CHILD_UP;
+            if (priv->halo_child_up[idx]) {
+                *event = GF_EVENT_CHILD_UP;
+            }
         } else {
             gf_log(child_xlator->name, GF_LOG_INFO,
                    "Not marking child %d up, "
@@ -5192,9 +6010,15 @@ __afr_handle_child_up_event(xlator_t *this, xlator_t *child_xlator,
      * want to set the child_latency to MAX to indicate
      * the child needs ping data to be available before doing child-up
      */
-    if (child_latency_msec < 0 && priv->halo_enabled) {
+    if (!priv->halo_enabled)
+        goto out;
+
+    if (child_latency_msec < 0) {
         /*set to INT64_MAX-1 so that it is found for best_down_child*/
-        priv->child_latency[idx] = AFR_HALO_MAX_LATENCY;
+        priv->halo_child_up[idx] = 1;
+        if (priv->child_latency[idx] < 0) {
+            priv->child_latency[idx] = AFR_HALO_MAX_LATENCY;
+        }
     }
 
     /*
@@ -5232,13 +6056,14 @@ __afr_handle_child_up_event(xlator_t *this, xlator_t *child_xlator,
                      "up_children (%d) > halo_max_replicas (%d)",
                      worst_up_child, up_children, priv->halo_max_replicas);
     }
-
+out:
     if (up_children == 1) {
         gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_SUBVOL_UP,
                "Subvolume '%s' came back up; "
                "going online.",
                child_xlator->name);
-        gf_event(EVENT_AFR_SUBVOL_UP, "subvol=%s", this->name);
+        gf_event(EVENT_AFR_SUBVOL_UP, "client-pid=%d; subvol=%s",
+                 this->ctx->cmd_args.client_pid, this->name);
     } else {
         *event = GF_EVENT_SOME_DESCENDENT_UP;
     }
@@ -5282,6 +6107,7 @@ __afr_handle_child_down_event(xlator_t *this, xlator_t *child_xlator, int idx,
      */
     if (child_latency_msec < 0) {
         priv->child_latency[idx] = child_latency_msec;
+        priv->halo_child_up[idx] = 0;
     }
     priv->child_up[idx] = 0;
 
@@ -5294,7 +6120,7 @@ __afr_handle_child_down_event(xlator_t *this, xlator_t *child_xlator, int idx,
      * as we want it to be up to date if we are going to
      * begin using it synchronously.
      */
-    if (up_children < priv->halo_min_replicas) {
+    if (priv->halo_enabled && up_children < priv->halo_min_replicas) {
         best_down_child = find_best_down_child(this);
         if (best_down_child >= 0) {
             gf_msg_debug(this->name, 0,
@@ -5306,7 +6132,6 @@ __afr_handle_child_down_event(xlator_t *this, xlator_t *child_xlator, int idx,
             *up_child = best_down_child;
         }
     }
-
     for (i = 0; i < priv->child_count; i++)
         if (priv->child_up[i] == 0)
             down_children++;
@@ -5315,7 +6140,8 @@ __afr_handle_child_down_event(xlator_t *this, xlator_t *child_xlator, int idx,
                "All subvolumes are down. Going "
                "offline until at least one of them "
                "comes back up.");
-        gf_event(EVENT_AFR_SUBVOLS_DOWN, "subvol=%s", this->name);
+        gf_event(EVENT_AFR_SUBVOLS_DOWN, "client-pid=%d; subvol=%s",
+                 this->ctx->cmd_args.client_pid, this->name);
     } else {
         *event = GF_EVENT_SOME_DESCENDENT_DOWN;
     }
@@ -5364,6 +6190,11 @@ afr_handle_inodelk_contention(xlator_t *this, struct gf_upcall *upcall)
     }
     LOCK(&priv->lock);
     {
+        if (priv->release_ta_notify_dom_lock == _gf_true) {
+            /* Ignore multiple release requests from shds.*/
+            UNLOCK(&priv->lock);
+            return;
+        }
         priv->release_ta_notify_dom_lock = _gf_true;
         inmem_count = priv->ta_in_mem_txn_count;
         onwire_count = priv->ta_on_wire_txn_count;
@@ -5371,7 +6202,7 @@ afr_handle_inodelk_contention(xlator_t *this, struct gf_upcall *upcall)
     UNLOCK(&priv->lock);
     if (inmem_count || onwire_count)
         /* lock release will happen in txn code path after
-         * inflight or on-wire txns are over.*/
+         * in-memory or on-wire txns are over.*/
         return;
 
     afr_ta_lock_release_synctask(this);
@@ -5470,14 +6301,15 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
         goto out;
     }
 
-    had_quorum = priv->quorum_count && afr_has_quorum(priv->child_up, this);
-    if (priv->halo_enabled) {
-        halo_max_latency_msec = afr_get_halo_latency(this);
+    had_quorum = priv->quorum_count &&
+                 afr_has_quorum(priv->child_up, this, NULL);
+    if (event == GF_EVENT_CHILD_PING) {
+        child_latency_msec = (int64_t)(uintptr_t)data2;
+        if (priv->halo_enabled) {
+            halo_max_latency_msec = afr_get_halo_latency(this);
 
-        if (event == GF_EVENT_CHILD_PING) {
             /* Calculates the child latency and sets event
              */
-            child_latency_msec = (int64_t)(uintptr_t)data2;
             LOCK(&priv->lock);
             {
                 __afr_handle_ping_event(this, child_xlator, idx,
@@ -5485,6 +6317,12 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
                                         child_latency_msec);
             }
             UNLOCK(&priv->lock);
+        } else {
+            LOCK(&priv->lock);
+            {
+                priv->child_latency[idx] = child_latency_msec;
+            }
+            UNLOCK(&priv->lock);
         }
     }
 
@@ -5528,22 +6366,27 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
                 if (priv->thin_arbiter_count &&
                     (idx == AFR_CHILD_THIN_ARBITER)) {
                     priv->ta_child_up = 1;
+                    priv->ta_event_gen++;
                     break;
                 }
                 __afr_handle_child_up_event(this, child_xlator, idx,
                                             child_latency_msec, &event,
                                             &call_psh, &up_child);
+                __afr_lock_heal_synctask(this, priv, idx);
                 break;
 
             case GF_EVENT_CHILD_DOWN:
                 if (priv->thin_arbiter_count &&
                     (idx == AFR_CHILD_THIN_ARBITER)) {
                     priv->ta_child_up = 0;
+                    priv->ta_event_gen++;
+                    afr_ta_locked_priv_invalidate(priv);
                     break;
                 }
                 __afr_handle_child_down_event(this, child_xlator, idx,
                                               child_latency_msec, &event,
                                               &call_psh, &up_child);
+                __afr_mark_pending_lk_heal(this, priv, idx);
                 break;
 
             case GF_EVENT_CHILD_CONNECTING:
@@ -5585,16 +6428,18 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
     UNLOCK(&priv->lock);
 
     if (priv->quorum_count) {
-        has_quorum = afr_has_quorum(priv->child_up, this);
+        has_quorum = afr_has_quorum(priv->child_up, this, NULL);
         if (!had_quorum && has_quorum) {
             gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_QUORUM_MET,
                    "Client-quorum is met");
-            gf_event(EVENT_AFR_QUORUM_MET, "subvol=%s", this->name);
+            gf_event(EVENT_AFR_QUORUM_MET, "client-pid=%d; subvol=%s",
+                     this->ctx->cmd_args.client_pid, this->name);
         }
         if (had_quorum && !has_quorum) {
             gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_QUORUM_FAIL,
                    "Client-quorum is not met");
-            gf_event(EVENT_AFR_QUORUM_FAIL, "subvol=%s", this->name);
+            gf_event(EVENT_AFR_QUORUM_FAIL, "client-pid=%d; subvol=%s",
+                     this->ctx->cmd_args.client_pid, this->name);
         }
     }
 
@@ -5613,10 +6458,8 @@ afr_notify(xlator_t *this, int32_t event, void *data, void *data2)
          * b) Already heard from everyone, but we now got a child-up
          *    event.
          */
-        if (have_heard_from_all && priv->shd.iamshd) {
-            for (i = 0; i < priv->child_count; i++)
-                if (priv->child_up[i])
-                    afr_selfheal_childup(this, i);
+        if (have_heard_from_all) {
+            afr_selfheal_childup(this, priv);
         }
     }
 out:
@@ -5637,7 +6480,7 @@ afr_local_init(afr_local_t *local, afr_private_t *priv, int32_t *op_errno)
         goto out;
     }
 
-    local->child_up = GF_CALLOC(priv->child_count, sizeof(*local->child_up),
+    local->child_up = GF_MALLOC(priv->child_count * sizeof(*local->child_up),
                                 gf_afr_mt_char);
     if (!local->child_up) {
         if (op_errno)
@@ -5696,7 +6539,11 @@ afr_local_init(afr_local_t *local, afr_private_t *priv, int32_t *op_errno)
     if (priv->thin_arbiter_count) {
         local->ta_child_up = priv->ta_child_up;
         local->ta_failed_subvol = AFR_CHILD_UNKNOWN;
+        local->read_txn_query_child = AFR_CHILD_UNKNOWN;
+        local->ta_event_gen = priv->ta_event_gen;
+        local->fop_state = TA_SUCCESS;
     }
+    local->is_new_entry = _gf_false;
 
     INIT_LIST_HEAD(&local->healer);
     return 0;
@@ -5709,11 +6556,6 @@ afr_internal_lock_init(afr_internal_lock_t *lk, size_t child_count)
 {
     int ret = -ENOMEM;
 
-    lk->locked_nodes = GF_CALLOC(sizeof(*lk->locked_nodes), child_count,
-                                 gf_afr_mt_char);
-    if (NULL == lk->locked_nodes)
-        goto out;
-
     lk->lower_locked_nodes = GF_CALLOC(sizeof(*lk->lower_locked_nodes),
                                        child_count, gf_afr_mt_char);
     if (NULL == lk->lower_locked_nodes)
@@ -5771,6 +6613,10 @@ afr_transaction_local_init(afr_local_t *local, xlator_t *this)
     afr_private_t *priv = NULL;
 
     priv = this->private;
+    INIT_LIST_HEAD(&local->transaction.wait_list);
+    INIT_LIST_HEAD(&local->transaction.owner_list);
+    INIT_LIST_HEAD(&local->ta_waitq);
+    INIT_LIST_HEAD(&local->ta_onwireq);
     ret = afr_internal_lock_init(&local->internal_lock, priv->child_count);
     if (ret < 0)
         goto out;
@@ -5808,10 +6654,6 @@ afr_transaction_local_init(afr_local_t *local, xlator_t *this)
         goto out;
 
     ret = 0;
-    INIT_LIST_HEAD(&local->transaction.wait_list);
-    INIT_LIST_HEAD(&local->transaction.owner_list);
-    INIT_LIST_HEAD(&local->ta_waitq);
-    INIT_LIST_HEAD(&local->ta_onwireq);
 out:
     return ret;
 }
@@ -5830,6 +6672,8 @@ afr_priv_destroy(afr_private_t *priv)
 
     if (!priv)
         goto out;
+
+    GF_FREE(priv->sh_domain);
     GF_FREE(priv->last_event);
 
     child_count = priv->child_count;
@@ -5845,7 +6689,9 @@ afr_priv_destroy(afr_private_t *priv)
     GF_FREE(priv->local);
     GF_FREE(priv->pending_key);
     GF_FREE(priv->children);
+    GF_FREE(priv->anon_inode);
     GF_FREE(priv->child_up);
+    GF_FREE(priv->halo_child_up);
     GF_FREE(priv->child_latency);
     LOCK_DESTROY(&priv->lock);
 
@@ -5896,263 +6742,218 @@ out:
     return changelog;
 }
 
-gf_boolean_t
-afr_decide_heal_info(afr_private_t *priv, unsigned char *sources, int source)
+static dict_t *
+afr_set_heal_info(char *status)
 {
-    int sources_count = 0;
+    dict_t *dict = NULL;
+    int ret = -1;
 
-    if (source < 0)
+    dict = dict_new();
+    if (!dict) {
+        ret = -ENOMEM;
         goto out;
+    }
 
-    sources_count = AFR_COUNT(sources, priv->child_count);
-    if (sources_count == priv->child_count)
-        return _gf_false;
+    ret = dict_set_dynstr_sizen(dict, "heal-info", status);
+    if (ret)
+        gf_msg("", GF_LOG_WARNING, -ret, AFR_MSG_DICT_SET_FAILED,
+               "Failed to set heal-info key to "
+               "%s",
+               status);
 out:
-    return _gf_true;
+    /* Any error other than EINVAL, dict_set_dynstr frees status */
+    if (ret == -ENOMEM || ret == -EINVAL) {
+        GF_FREE(status);
+    }
+
+    if (ret && dict) {
+        dict_unref(dict);
+        dict = NULL;
+    }
+    return dict;
 }
 
-int
-afr_selfheal_locked_metadata_inspect(call_frame_t *frame, xlator_t *this,
-                                     inode_t *inode, gf_boolean_t *msh,
-                                     unsigned char *pending)
+static gf_boolean_t
+afr_is_dirty_count_non_unary_for_txn(xlator_t *this, struct afr_reply *replies,
+                                     afr_transaction_type type)
 {
-    int ret = -1;
-    unsigned char *locked_on = NULL;
-    unsigned char *sources = NULL;
-    unsigned char *sinks = NULL;
-    unsigned char *healed_sinks = NULL;
-    unsigned char *undid_pending = NULL;
-    struct afr_reply *locked_replies = NULL;
-
     afr_private_t *priv = this->private;
+    int *dirty = alloca0(priv->child_count * sizeof(int));
+    int i = 0;
 
-    locked_on = alloca0(priv->child_count);
-    sources = alloca0(priv->child_count);
-    sinks = alloca0(priv->child_count);
-    healed_sinks = alloca0(priv->child_count);
-    undid_pending = alloca0(priv->child_count);
-
-    locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
-
-    ret = afr_selfheal_inodelk(frame, this, inode, this->name, LLONG_MAX - 1, 0,
-                               locked_on);
-    {
-        if (ret == 0) {
-            /* Not a single lock */
-            ret = -afr_final_errno(frame->local, priv);
-            if (ret == 0)
-                ret = -ENOTCONN; /* all invalid responses */
-            goto out;
-        }
-        ret = __afr_selfheal_metadata_prepare(
-            frame, this, inode, locked_on, sources, sinks, healed_sinks,
-            undid_pending, locked_replies, pending);
-        *msh = afr_decide_heal_info(priv, sources, ret);
+    afr_selfheal_extract_xattr(this, replies, type, dirty, NULL);
+    for (i = 0; i < priv->child_count; i++) {
+        if (dirty[i] > 1)
+            return _gf_true;
     }
-    afr_selfheal_uninodelk(frame, this, inode, this->name, LLONG_MAX - 1, 0,
-                           locked_on);
-out:
-    if (locked_replies)
-        afr_replies_wipe(locked_replies, priv->child_count);
-    return ret;
+
+    return _gf_false;
 }
 
-int
-afr_selfheal_locked_data_inspect(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                                 gf_boolean_t *dsh, unsigned char *pflag)
+static gf_boolean_t
+afr_is_dirty_count_non_unary(xlator_t *this, struct afr_reply *replies,
+                             ia_type_t ia_type)
 {
-    int ret = -1;
-    unsigned char *data_lock = NULL;
-    unsigned char *sources = NULL;
-    unsigned char *sinks = NULL;
-    unsigned char *healed_sinks = NULL;
-    unsigned char *undid_pending = NULL;
-    afr_private_t *priv = NULL;
-    struct afr_reply *locked_replies = NULL;
-    inode_t *inode = fd->inode;
+    gf_boolean_t data_chk = _gf_false;
+    gf_boolean_t mdata_chk = _gf_false;
+    gf_boolean_t entry_chk = _gf_false;
 
-    priv = this->private;
-    data_lock = alloca0(priv->child_count);
-    sources = alloca0(priv->child_count);
-    sinks = alloca0(priv->child_count);
-    healed_sinks = alloca0(priv->child_count);
-    undid_pending = alloca0(priv->child_count);
-
-    locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
+    switch (ia_type) {
+        case IA_IFDIR:
+            mdata_chk = _gf_true;
+            entry_chk = _gf_true;
+            break;
+        case IA_IFREG:
+            mdata_chk = _gf_true;
+            data_chk = _gf_true;
+            break;
+        default:
+            /*IA_IFBLK, IA_IFCHR, IA_IFLNK, IA_IFIFO, IA_IFSOCK*/
+            mdata_chk = _gf_true;
+            break;
+    }
 
-    ret = afr_selfheal_inodelk(frame, this, inode, this->name, 0, 0, data_lock);
-    {
-        if (ret == 0) {
-            ret = -afr_final_errno(frame->local, priv);
-            if (ret == 0)
-                ret = -ENOTCONN; /* all invalid responses */
-            goto out;
-        }
-        ret = __afr_selfheal_data_prepare(frame, this, inode, data_lock,
-                                          sources, sinks, healed_sinks,
-                                          undid_pending, locked_replies, pflag);
-        *dsh = afr_decide_heal_info(priv, sources, ret);
+    if (data_chk && afr_is_dirty_count_non_unary_for_txn(
+                        this, replies, AFR_DATA_TRANSACTION)) {
+        return _gf_true;
+    } else if (mdata_chk && afr_is_dirty_count_non_unary_for_txn(
+                                this, replies, AFR_METADATA_TRANSACTION)) {
+        return _gf_true;
+    } else if (entry_chk && afr_is_dirty_count_non_unary_for_txn(
+                                this, replies, AFR_ENTRY_TRANSACTION)) {
+        return _gf_true;
     }
-    afr_selfheal_uninodelk(frame, this, inode, this->name, 0, 0, data_lock);
-out:
-    if (locked_replies)
-        afr_replies_wipe(locked_replies, priv->child_count);
-    return ret;
+
+    return _gf_false;
 }
 
-int
-afr_selfheal_locked_entry_inspect(call_frame_t *frame, xlator_t *this,
-                                  inode_t *inode, gf_boolean_t *esh,
-                                  unsigned char *pflag)
+static int
+afr_update_heal_status(xlator_t *this, struct afr_reply *replies,
+                       ia_type_t ia_type, gf_boolean_t *esh, gf_boolean_t *dsh,
+                       gf_boolean_t *msh, unsigned char pending)
 {
     int ret = -1;
-    int source = -1;
+    GF_UNUSED int ret1 = 0;
+    int i = 0;
+    int io_domain_lk_count = 0;
+    int shd_domain_lk_count = 0;
     afr_private_t *priv = NULL;
-    unsigned char *locked_on = NULL;
-    unsigned char *data_lock = NULL;
-    unsigned char *sources = NULL;
-    unsigned char *sinks = NULL;
-    unsigned char *healed_sinks = NULL;
-    struct afr_reply *locked_replies = NULL;
-    gf_boolean_t granular_locks = _gf_false;
+    char *key1 = NULL;
+    char *key2 = NULL;
 
     priv = this->private;
-    if (strcmp("granular", priv->locking_scheme) == 0)
-        granular_locks = _gf_true;
-    locked_on = alloca0(priv->child_count);
-    data_lock = alloca0(priv->child_count);
-    sources = alloca0(priv->child_count);
-    sinks = alloca0(priv->child_count);
-    healed_sinks = alloca0(priv->child_count);
+    key1 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+                   strlen(this->name));
+    key2 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+                   strlen(priv->sh_domain));
+    sprintf(key1, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, this->name);
+    sprintf(key2, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, priv->sh_domain);
 
-    locked_replies = alloca0(sizeof(*locked_replies) * priv->child_count);
-
-    if (!granular_locks) {
-        ret = afr_selfheal_tryentrylk(frame, this, inode, priv->sh_domain, NULL,
-                                      locked_on);
-    }
-    {
-        if (!granular_locks && ret == 0) {
-            ret = -afr_final_errno(frame->local, priv);
-            if (ret == 0)
-                ret = -ENOTCONN; /* all invalid responses */
-            goto out;
+    for (i = 0; i < priv->child_count; i++) {
+        if ((replies[i].valid != 1) || (replies[i].op_ret != 0))
+            continue;
+        if (!io_domain_lk_count) {
+            ret1 = dict_get_int32(replies[i].xdata, key1, &io_domain_lk_count);
+        }
+        if (!shd_domain_lk_count) {
+            ret1 = dict_get_int32(replies[i].xdata, key2, &shd_domain_lk_count);
         }
+    }
 
-        ret = afr_selfheal_entrylk(frame, this, inode, this->name, NULL,
-                                   data_lock);
-        {
-            if (ret == 0) {
-                ret = -afr_final_errno(frame->local, priv);
-                if (ret == 0)
-                    ret = -ENOTCONN;
-                /* all invalid responses */
-                goto unlock;
-            }
-            ret = __afr_selfheal_entry_prepare(frame, this, inode, data_lock,
-                                               sources, sinks, healed_sinks,
-                                               locked_replies, &source, pflag);
-            if ((ret == 0) && (*pflag & PFLAG_SBRAIN))
-                ret = -EIO;
-            *esh = afr_decide_heal_info(priv, sources, ret);
+    if (!pending) {
+        if ((afr_is_dirty_count_non_unary(this, replies, ia_type)) ||
+            (!io_domain_lk_count)) {
+            /* Needs heal. */
+            ret = 0;
+        } else {
+            /* No heal needed. */
+            *dsh = *esh = *msh = 0;
+        }
+    } else {
+        if (shd_domain_lk_count) {
+            ret = -EAGAIN; /*For 'possibly-healing'. */
+        } else {
+            ret = 0; /*needs heal. Just set a non -ve value so that it is
+                       assumed as the source index.*/
         }
-        afr_selfheal_unentrylk(frame, this, inode, this->name, NULL, data_lock,
-                               NULL);
     }
-unlock:
-    if (!granular_locks)
-        afr_selfheal_unentrylk(frame, this, inode, priv->sh_domain, NULL,
-                               locked_on, NULL);
-out:
-    if (locked_replies)
-        afr_replies_wipe(locked_replies, priv->child_count);
     return ret;
 }
 
+/*return EIO, EAGAIN or pending*/
 int
-afr_selfheal_locked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
-                            inode_t **inode, gf_boolean_t *entry_selfheal,
-                            gf_boolean_t *data_selfheal,
-                            gf_boolean_t *metadata_selfheal,
-                            unsigned char *pending)
-
+afr_lockless_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
+                     inode_t **inode, gf_boolean_t *entry_selfheal,
+                     gf_boolean_t *data_selfheal,
+                     gf_boolean_t *metadata_selfheal, unsigned char *pending)
 {
     int ret = -1;
-    fd_t *fd = NULL;
+    int i = 0;
+    afr_private_t *priv = NULL;
+    struct afr_reply *replies = NULL;
     gf_boolean_t dsh = _gf_false;
     gf_boolean_t msh = _gf_false;
     gf_boolean_t esh = _gf_false;
+    unsigned char *sources = NULL;
+    unsigned char *sinks = NULL;
+    unsigned char *valid_on = NULL;
+    uint64_t *witness = NULL;
+
+    priv = this->private;
+    replies = alloca0(sizeof(*replies) * priv->child_count);
+    sources = alloca0(sizeof(*sources) * priv->child_count);
+    sinks = alloca0(sizeof(*sinks) * priv->child_count);
+    witness = alloca0(sizeof(*witness) * priv->child_count);
+    valid_on = alloca0(sizeof(*valid_on) * priv->child_count);
 
     ret = afr_selfheal_unlocked_inspect(frame, this, gfid, inode, &dsh, &msh,
-                                        &esh);
+                                        &esh, replies);
     if (ret)
         goto out;
-
-    /* For every heal type hold locks and check if it indeed needs heal */
-
-    /* Heal-info does an open() on the file being examined so that the
-     * current eager-lock holding client, if present, at some point sees
-     * open-fd count being > 1 and releases the eager-lock so that heal-info
-     * doesn't remain blocked forever until IO completes.
-     */
-    if ((*inode)->ia_type == IA_IFREG) {
-        ret = afr_selfheal_data_open(this, *inode, &fd);
-        if (ret < 0) {
-            gf_msg_debug(this->name, -ret, "%s: Failed to open",
-                         uuid_utoa((*inode)->gfid));
-            goto out;
+    for (i = 0; i < priv->child_count; i++) {
+        if (replies[i].valid && replies[i].op_ret == 0) {
+            valid_on[i] = 1;
         }
     }
-
     if (msh) {
-        ret = afr_selfheal_locked_metadata_inspect(frame, this, *inode, &msh,
-                                                   pending);
-        if (ret == -EIO)
+        ret = afr_selfheal_find_direction(frame, this, replies,
+                                          AFR_METADATA_TRANSACTION, valid_on,
+                                          sources, sinks, witness, pending);
+        if (*pending & PFLAG_SBRAIN)
+            ret = -EIO;
+        if (ret)
             goto out;
     }
-
     if (dsh) {
-        ret = afr_selfheal_locked_data_inspect(frame, this, fd, &dsh, pending);
-        if (ret == -EIO || (ret == -EAGAIN))
+        ret = afr_selfheal_find_direction(frame, this, replies,
+                                          AFR_DATA_TRANSACTION, valid_on,
+                                          sources, sinks, witness, pending);
+        if (*pending & PFLAG_SBRAIN)
+            ret = -EIO;
+        if (ret)
             goto out;
     }
-
     if (esh) {
-        ret = afr_selfheal_locked_entry_inspect(frame, this, *inode, &esh,
-                                                pending);
+        ret = afr_selfheal_find_direction(frame, this, replies,
+                                          AFR_ENTRY_TRANSACTION, valid_on,
+                                          sources, sinks, witness, pending);
+        if (*pending & PFLAG_SBRAIN)
+            ret = -EIO;
+        if (ret)
+            goto out;
     }
 
+    ret = afr_update_heal_status(this, replies, (*inode)->ia_type, &esh, &dsh,
+                                 &msh, *pending);
 out:
     *data_selfheal = dsh;
     *entry_selfheal = esh;
     *metadata_selfheal = msh;
-    if (fd)
-        fd_unref(fd);
+    if (replies)
+        afr_replies_wipe(replies, priv->child_count);
     return ret;
 }
 
-dict_t *
-afr_set_heal_info(char *status)
-{
-    dict_t *dict = NULL;
-    int ret = -1;
-
-    dict = dict_new();
-    if (!dict) {
-        ret = -ENOMEM;
-        goto out;
-    }
-
-    ret = dict_set_str(dict, "heal-info", status);
-    if (ret)
-        gf_msg("", GF_LOG_WARNING, -ret, AFR_MSG_DICT_SET_FAILED,
-               "Failed to set heal-info key to "
-               "%s",
-               status);
-out:
-    return dict;
-}
-
 int
 afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
 {
@@ -6162,17 +6963,27 @@ afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
     unsigned char pending = 0;
     dict_t *dict = NULL;
     int ret = -1;
-    int op_errno = 0;
+    int op_errno = ENOMEM;
     inode_t *inode = NULL;
     char *substr = NULL;
     char *status = NULL;
+    call_frame_t *heal_frame = NULL;
+    afr_local_t *heal_local = NULL;
+
+    /*Use frame with lk-owner set*/
+    heal_frame = afr_frame_create(frame->this, &op_errno);
+    if (!heal_frame) {
+        ret = -1;
+        goto out;
+    }
+    heal_local = heal_frame->local;
+    heal_frame->local = frame->local;
 
-    ret = afr_selfheal_locked_inspect(frame, this, loc->gfid, &inode,
-                                      &entry_selfheal, &data_selfheal,
-                                      &metadata_selfheal, &pending);
+    ret = afr_lockless_inspect(heal_frame, this, loc->gfid, &inode,
+                               &entry_selfheal, &data_selfheal,
+                               &metadata_selfheal, &pending);
 
     if (ret == -ENOMEM) {
-        op_errno = -ret;
         ret = -1;
         goto out;
     }
@@ -6185,26 +6996,50 @@ afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
 
     if (ret == -EIO) {
         ret = gf_asprintf(&status, "split-brain%s", substr ? substr : "");
-        if (ret < 0)
+        if (ret < 0) {
             goto out;
+        }
         dict = afr_set_heal_info(status);
+        if (!dict) {
+            ret = -1;
+            goto out;
+        }
     } else if (ret == -EAGAIN) {
         ret = gf_asprintf(&status, "possibly-healing%s", substr ? substr : "");
-        if (ret < 0)
+        if (ret < 0) {
             goto out;
+        }
         dict = afr_set_heal_info(status);
+        if (!dict) {
+            ret = -1;
+            goto out;
+        }
     } else if (ret >= 0) {
         /* value of ret = source index
          * so ret >= 0 and at least one of the 3 booleans set to
          * true means a source is identified; heal is required.
          */
         if (!data_selfheal && !entry_selfheal && !metadata_selfheal) {
-            dict = afr_set_heal_info("no-heal");
+            status = gf_strdup("no-heal");
+            if (!status) {
+                ret = -1;
+                goto out;
+            }
+            dict = afr_set_heal_info(status);
+            if (!dict) {
+                ret = -1;
+                goto out;
+            }
         } else {
             ret = gf_asprintf(&status, "heal%s", substr ? substr : "");
-            if (ret < 0)
+            if (ret < 0) {
                 goto out;
+            }
             dict = afr_set_heal_info(status);
+            if (!dict) {
+                ret = -1;
+                goto out;
+            }
         }
     } else if (ret < 0) {
         /* Apart from above checked -ve ret values, there are
@@ -6216,14 +7051,25 @@ afr_get_heal_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
          */
         if (data_selfheal || entry_selfheal || metadata_selfheal) {
             ret = gf_asprintf(&status, "heal%s", substr ? substr : "");
-            if (ret < 0)
+            if (ret < 0) {
                 goto out;
+            }
             dict = afr_set_heal_info(status);
+            if (!dict) {
+                ret = -1;
+                goto out;
+            }
         }
     }
+
     ret = 0;
+    op_errno = 0;
 
 out:
+    if (heal_frame) {
+        heal_frame->local = heal_local;
+        AFR_STACK_DESTROY(heal_frame);
+    }
     AFR_STACK_UNWIND(getxattr, frame, ret, op_errno, dict, NULL);
     if (dict)
         dict_unref(dict);
@@ -6345,10 +7191,10 @@ afr_get_split_brain_status(void *opaque)
     }
 
     /* Calculation for string length :
-     * (child_count X length of child-name) + SLEN ("    Choices :")
+     * (child_count X length of child-name) + SLEN("    Choices :")
      * child-name consists of :
      * a) 251 = max characters for volname according to GD_VOLUME_NAME_MAX
-     * b) strlen ("-client-00,") assuming 16 replicas
+     * b) strlen("-client-00,") assuming 16 replicas
      */
     choices = alloca0(priv->child_count * (256 + SLEN("-client-00,")) +
                       SLEN("    Choices:"));
@@ -6357,8 +7203,8 @@ afr_get_split_brain_status(void *opaque)
     if (ret) {
         op_errno = -ret;
         if (ret == -EAGAIN) {
-            ret = dict_set_str(dict, GF_AFR_SBRAIN_STATUS,
-                               SBRAIN_HEAL_NO_GO_MSG);
+            ret = dict_set_sizen_str_sizen(dict, GF_AFR_SBRAIN_STATUS,
+                                           SBRAIN_HEAL_NO_GO_MSG);
             if (ret) {
                 gf_msg(this->name, GF_LOG_WARNING, -ret,
                        AFR_MSG_DICT_SET_FAILED,
@@ -6387,16 +7233,15 @@ afr_get_split_brain_status(void *opaque)
             op_errno = ENOMEM;
             goto out;
         }
-        ret = dict_set_dynstr(dict, GF_AFR_SBRAIN_STATUS, status);
+        ret = dict_set_dynstr_sizen(dict, GF_AFR_SBRAIN_STATUS, status);
         if (ret) {
             op_errno = -ret;
             ret = -1;
             goto out;
         }
     } else {
-        ret = dict_set_str(dict, GF_AFR_SBRAIN_STATUS,
-                           "The file is not under data or"
-                           " metadata split-brain");
+        ret = dict_set_sizen_str_sizen(dict, GF_AFR_SBRAIN_STATUS,
+                                       SFILE_NOT_UNDER_DATA);
         if (ret) {
             op_errno = -ret;
             ret = -1;
@@ -6421,6 +7266,8 @@ afr_heal_splitbrain_file(call_frame_t *frame, xlator_t *this, loc_t *loc)
     int op_errno = 0;
     dict_t *dict = NULL;
     afr_local_t *local = NULL;
+    afr_local_t *heal_local = NULL;
+    call_frame_t *heal_frame = NULL;
 
     local = frame->local;
     dict = dict_new();
@@ -6430,10 +7277,20 @@ afr_heal_splitbrain_file(call_frame_t *frame, xlator_t *this, loc_t *loc)
         goto out;
     }
 
-    ret = afr_selfheal_do(frame, this, loc->gfid);
+    heal_frame = afr_frame_create(this, &op_errno);
+    if (!heal_frame) {
+        ret = -1;
+        goto out;
+    }
+    heal_local = heal_frame->local;
+    heal_frame->local = frame->local;
+    /*Initiate heal with heal_frame with lk-owner set so that inodelk/entrylk
+     * work correctly*/
+    ret = afr_selfheal_do(heal_frame, this, loc->gfid);
 
     if (ret == 1 || ret == 2) {
-        ret = dict_set_str(dict, "sh-fail-msg", "File not in split-brain");
+        ret = dict_set_sizen_str_sizen(dict, "sh-fail-msg",
+                                       SFILE_NOT_IN_SPLIT_BRAIN);
         if (ret)
             gf_msg(this->name, GF_LOG_WARNING, -ret, AFR_MSG_DICT_SET_FAILED,
                    "Failed to set sh-fail-msg in dict");
@@ -6451,6 +7308,10 @@ afr_heal_splitbrain_file(call_frame_t *frame, xlator_t *this, loc_t *loc)
     }
 
 out:
+    if (heal_frame) {
+        heal_frame->local = heal_local;
+        AFR_STACK_DESTROY(heal_frame);
+    }
     if (local->op == GF_FOP_GETXATTR)
         AFR_STACK_UNWIND(getxattr, frame, ret, op_errno, dict, NULL);
     else if (local->op == GF_FOP_SETXATTR)
@@ -6593,7 +7454,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
         ret = afr_selfheal_inodelk(heal_frame, this, inode, this->name, 0, 0,
                                    locked_on);
         {
-            if (ret < AFR_SH_MIN_PARTICIPANTS)
+            if (ret < priv->child_count)
                 goto data_unlock;
             ret = __afr_selfheal_data_prepare(
                 heal_frame, this, inode, locked_on, sources, sinks,
@@ -6610,7 +7471,7 @@ afr_fav_child_reset_sink_xattrs(void *opaque)
         ret = afr_selfheal_inodelk(heal_frame, this, inode, this->name,
                                    LLONG_MAX - 1, 0, locked_on);
         {
-            if (ret < AFR_SH_MIN_PARTICIPANTS)
+            if (ret < priv->child_count)
                 goto mdata_unlock;
             ret = __afr_selfheal_metadata_prepare(
                 heal_frame, this, inode, locked_on, sources, sinks,
@@ -6637,12 +7498,14 @@ afr_serialize_xattrs_with_delimiter(call_frame_t *frame, xlator_t *this,
     char *xattr = NULL;
     int i = 0;
     int len = 0;
+    int keylen = 0;
     size_t str_len = 0;
     int ret = -1;
 
     priv = this->private;
     local = frame->local;
 
+    keylen = strlen(local->cont.getxattr.name);
     for (i = 0; i < priv->child_count; i++) {
         if (!local->replies[i].valid || local->replies[i].op_ret) {
             str_len = strlen(default_str);
@@ -6651,8 +7514,8 @@ afr_serialize_xattrs_with_delimiter(call_frame_t *frame, xlator_t *this,
             buf[len++] = delimiter;
             buf[len] = '\0';
         } else {
-            ret = dict_get_str(local->replies[i].xattr,
-                               local->cont.getxattr.name, &xattr);
+            ret = dict_get_strn(local->replies[i].xattr,
+                                local->cont.getxattr.name, keylen, &xattr);
             if (ret) {
                 gf_msg("TEST", GF_LOG_ERROR, -ret, AFR_MSG_DICT_GET_FAILED,
                        "Failed to get the node_uuid of brick "
@@ -6939,3 +7802,77 @@ afr_ta_has_quorum(afr_private_t *priv, afr_local_t *local)
 
     return _gf_false;
 }
+
+static gf_boolean_t
+afr_is_add_replica_mount_lookup_on_root(call_frame_t *frame)
+{
+    afr_local_t *local = NULL;
+
+    if (frame->root->pid != GF_CLIENT_PID_ADD_REPLICA_MOUNT)
+        return _gf_false;
+
+    local = frame->local;
+
+    if (local->op != GF_FOP_LOOKUP)
+        /* TODO:If the replica count is being increased on a plain distribute
+         * volume that was never mounted, we need to allow setxattr on '/' with
+         * GF_CLIENT_PID_NO_ROOT_SQUASH to accomodate for DHT layout setting */
+        return _gf_false;
+
+    if (local->inode == NULL)
+        return _gf_false;
+
+    if (!__is_root_gfid(local->inode->gfid))
+        return _gf_false;
+
+    return _gf_true;
+}
+
+gf_boolean_t
+afr_lookup_has_quorum(call_frame_t *frame, const unsigned int up_children_count)
+{
+    if (frame && (up_children_count > 0) &&
+        afr_is_add_replica_mount_lookup_on_root(frame))
+        return _gf_true;
+
+    return _gf_false;
+}
+
+void
+afr_handle_replies_quorum(call_frame_t *frame, xlator_t *this)
+{
+    afr_local_t *local = frame->local;
+    afr_private_t *priv = this->private;
+    unsigned char *success_replies = NULL;
+
+    success_replies = alloca0(priv->child_count);
+    afr_fill_success_replies(local, priv, success_replies);
+
+    if (priv->quorum_count && !afr_has_quorum(success_replies, this, NULL)) {
+        local->op_errno = afr_final_errno(local, priv);
+        if (!local->op_errno)
+            local->op_errno = afr_quorum_errno(priv);
+        local->op_ret = -1;
+    }
+}
+
+gf_boolean_t
+afr_ta_dict_contains_pending_xattr(dict_t *dict, afr_private_t *priv, int child)
+{
+    int *pending = NULL;
+    int ret = 0;
+    int i = 0;
+
+    ret = dict_get_ptr(dict, priv->pending_key[child], (void *)&pending);
+    if (ret == 0) {
+        for (i = 0; i < AFR_NUM_CHANGE_LOGS; i++) {
+            /* Not doing a ntoh32(pending) as we just want to check
+             * if it is non-zero or not. */
+            if (pending[i]) {
+                return _gf_true;
+            }
+        }
+    }
+
+    return _gf_false;
+}
diff --git a/xlators/cluster/afr/src/afr-dir-read.c b/xlators/cluster/afr/src/afr-dir-read.c
index 4c40e85f393..f8bf8340dab 100644
--- a/xlators/cluster/afr/src/afr-dir-read.c
+++ b/xlators/cluster/afr/src/afr-dir-read.c
@@ -10,24 +10,17 @@
 
 #include <libgen.h>
 #include <unistd.h>
-#include <fnmatch.h>
 #include <sys/time.h>
 #include <stdlib.h>
 #include <signal.h>
 #include <string.h>
 
-#include "glusterfs.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
-#include "checksum.h"
+#include <glusterfs/glusterfs.h>
+#include <glusterfs/dict.h>
+#include <glusterfs/list.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
 
 #include "afr.h"
 #include "afr-transaction.h"
@@ -45,6 +38,10 @@ afr_opendir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     fd_ctx = local->fd_ctx;
     child_index = (long)cookie;
 
+    local->replies[child_index].valid = 1;
+    local->replies[child_index].op_ret = op_ret;
+    local->replies[child_index].op_errno = op_errno;
+
     LOCK(&frame->lock);
     {
         if (op_ret == -1) {
@@ -56,19 +53,22 @@ afr_opendir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
             if (!local->xdata_rsp && xdata)
                 local->xdata_rsp = dict_ref(xdata);
         }
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
-    if (call_count == 0)
+    if (call_count == 0) {
+        afr_handle_replies_quorum(frame, this);
         AFR_STACK_UNWIND(opendir, frame, local->op_ret, local->op_errno,
                          local->fd, NULL);
+    }
+
     return 0;
 }
 
 int
-afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd)
+afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd,
+            dict_t *xdata)
 {
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
@@ -84,6 +84,12 @@ afr_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd)
         goto out;
 
     local->op = GF_FOP_OPENDIR;
+
+    if (priv->quorum_count && !afr_has_quorum(local->child_up, this, NULL)) {
+        op_errno = afr_quorum_errno(priv);
+        goto out;
+    }
+
     if (!afr_is_consistent_io_possible(local, priv, &op_errno))
         goto out;
 
@@ -158,8 +164,8 @@ afr_validate_read_subvol(inode_t *inode, xlator_t *this, int par_read_subvol)
 }
 
 static void
-afr_readdir_transform_entries(gf_dirent_t *subvol_entries, int subvol,
-                              gf_dirent_t *entries, fd_t *fd)
+afr_readdir_transform_entries(call_frame_t *frame, gf_dirent_t *subvol_entries,
+                              int subvol, gf_dirent_t *entries, fd_t *fd)
 {
     int ret = -1;
     gf_dirent_t *entry = NULL;
@@ -177,8 +183,8 @@ afr_readdir_transform_entries(gf_dirent_t *subvol_entries, int subvol,
 
     list_for_each_entry_safe(entry, tmp, &subvol_entries->list, list)
     {
-        if (__is_root_gfid(fd->inode->gfid) &&
-            !strcmp(entry->d_name, GF_REPLICATE_TRASH_DIR)) {
+        if (afr_is_private_directory(priv, fd->inode->gfid, entry->d_name,
+                                     frame->root->pid)) {
             continue;
         }
 
@@ -222,8 +228,8 @@ afr_readdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     }
 
     if (op_ret >= 0)
-        afr_readdir_transform_entries(subvol_entries, (long)cookie, &entries,
-                                      local->fd);
+        afr_readdir_transform_entries(frame, subvol_entries, (long)cookie,
+                                      &entries, local->fd);
 
     AFR_STACK_UNWIND(readdir, frame, op_ret, op_errno, &entries, xdata);
 
diff --git a/xlators/cluster/afr/src/afr-dir-write.c b/xlators/cluster/afr/src/afr-dir-write.c
index 5725b1c5cb3..b7cceb79158 100644
--- a/xlators/cluster/afr/src/afr-dir-write.c
+++ b/xlators/cluster/afr/src/afr-dir-write.c
@@ -10,24 +10,20 @@
 
 #include <libgen.h>
 #include <unistd.h>
-#include <fnmatch.h>
 #include <sys/time.h>
 #include <stdlib.h>
 #include <signal.h>
 
-#include "glusterfs.h"
+#include <glusterfs/glusterfs.h>
 #include "afr.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
-#include "byte-order.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/list.h>
+#include <glusterfs/defaults.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
+#include <glusterfs/byte-order.h>
 
 #include "afr.h"
 #include "afr-transaction.h"
@@ -98,7 +94,9 @@ __afr_dir_write_finalize(call_frame_t *frame, xlator_t *this)
     }
 
     if (local->inode) {
-        afr_replies_interpret(frame, this, local->inode, NULL);
+        if (local->op != GF_FOP_RENAME && local->op != GF_FOP_LINK)
+            afr_replies_interpret(frame, this, local->inode, NULL);
+
         inode_read_subvol = afr_data_subvol_get(local->inode, this, NULL, NULL,
                                                 NULL, &args);
     }
@@ -121,11 +119,11 @@ __afr_dir_write_finalize(call_frame_t *frame, xlator_t *this)
             continue;
         if (local->replies[i].op_ret < 0) {
             if (local->inode)
-                afr_inode_event_gen_reset(local->inode, this);
+                afr_inode_need_refresh_set(local->inode, this);
             if (local->parent)
-                afr_inode_event_gen_reset(local->parent, this);
+                afr_inode_need_refresh_set(local->parent, this);
             if (local->parent2)
-                afr_inode_event_gen_reset(local->parent2, this);
+                afr_inode_need_refresh_set(local->parent2, this);
             continue;
         }
 
@@ -231,9 +229,9 @@ __afr_dir_write_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         __afr_dir_write_fill(frame, this, child_index, op_ret, op_errno, buf,
                              preparent, postparent, preparent2, postparent2,
                              xdata);
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
-    call_count = afr_frame_return(frame);
 
     if (call_count == 0) {
         __afr_dir_write_finalize(frame, this);
@@ -347,6 +345,7 @@ afr_mark_entry_pending_changelog(call_frame_t *frame, xlator_t *this)
     afr_private_t *priv = NULL;
     int pre_op_count = 0;
     int failed_count = 0;
+    unsigned char *success_replies = NULL;
 
     local = frame->local;
     priv = this->private;
@@ -362,9 +361,22 @@ afr_mark_entry_pending_changelog(call_frame_t *frame, xlator_t *this)
     failed_count = AFR_COUNT(local->transaction.failed_subvols,
                              priv->child_count);
 
+    /* FOP succeeded on all bricks. */
     if (pre_op_count == priv->child_count && !failed_count)
         return;
 
+    /* FOP did not suceed on quorum no. of bricks. */
+    success_replies = alloca0(priv->child_count);
+    afr_fill_success_replies(local, priv, success_replies);
+    if (!afr_has_quorum(success_replies, this, NULL))
+        return;
+
+    if (priv->thin_arbiter_count) {
+        /*Mark new entry using ta file*/
+        local->is_new_entry = _gf_true;
+        return;
+    }
+
     afr_mark_new_entry_changelog(frame, this);
 
     return;
@@ -423,15 +435,11 @@ int
 afr_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
            mode_t mode, mode_t umask, fd_t *fd, dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -473,16 +481,6 @@ afr_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -553,15 +551,11 @@ int
 afr_mknod(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
           dev_t dev, mode_t umask, dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -596,16 +590,6 @@ afr_mknod(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -674,15 +658,11 @@ int
 afr_mkdir(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
           mode_t umask, dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -698,7 +678,7 @@ afr_mkdir(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
     local->cont.mkdir.mode = mode;
     local->umask = umask;
 
-    if (!xdata || !dict_get(xdata, "gfid-req")) {
+    if (!xdata || !dict_get_sizen(xdata, "gfid-req")) {
         op_errno = EPERM;
         gf_msg_callingfn(this->name, GF_LOG_WARNING, op_errno,
                          AFR_MSG_GFID_NULL,
@@ -724,16 +704,6 @@ afr_mkdir(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -802,15 +772,11 @@ int
 afr_link(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
          dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -845,16 +811,6 @@ afr_link(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(newloc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -924,15 +880,11 @@ int
 afr_symlink(call_frame_t *frame, xlator_t *this, const char *linkpath,
             loc_t *loc, mode_t umask, dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -966,16 +918,6 @@ afr_symlink(call_frame_t *frame, xlator_t *this, const char *linkpath,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -1048,15 +990,10 @@ int
 afr_rename(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
            dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
-    int nlockee = 0;
-
-    priv = this->private;
 
     transaction_frame = copy_frame(frame);
     if (!transaction_frame) {
@@ -1099,35 +1036,6 @@ afr_rename(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(oldloc->path);
     local->transaction.new_basename = AFR_BASENAME(newloc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = nlockee = 0;
-    ret = afr_init_entry_lockee(
-        &int_lock->lockee[nlockee], local, &local->transaction.new_parent_loc,
-        local->transaction.new_basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    nlockee++;
-    ret = afr_init_entry_lockee(&int_lock->lockee[nlockee], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    nlockee++;
-    if (local->newloc.inode && IA_ISDIR(local->newloc.inode->ia_type)) {
-        ret = afr_init_entry_lockee(&int_lock->lockee[nlockee], local,
-                                    &local->newloc, NULL, priv->child_count);
-        if (ret)
-            goto out;
-
-        nlockee++;
-    }
-    qsort(int_lock->lockee, nlockee, sizeof(*int_lock->lockee),
-          afr_entry_lockee_cmp);
-    int_lock->lockee_count = nlockee;
-
     ret = afr_transaction(transaction_frame, this,
                           AFR_ENTRY_RENAME_TRANSACTION);
     if (ret < 0) {
@@ -1196,15 +1104,11 @@ int
 afr_unlink(call_frame_t *frame, xlator_t *this, loc_t *loc, int xflag,
            dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
 
-    priv = this->private;
-
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -1237,16 +1141,6 @@ afr_unlink(call_frame_t *frame, xlator_t *this, loc_t *loc, int xflag,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[0], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    int_lock->lockee_count++;
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
@@ -1313,15 +1207,10 @@ int
 afr_rmdir(call_frame_t *frame, xlator_t *this, loc_t *loc, int flags,
           dict_t *xdata)
 {
-    afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int op_errno = ENOMEM;
-    int nlockee = 0;
-
-    priv = this->private;
 
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
@@ -1355,26 +1244,6 @@ afr_rmdir(call_frame_t *frame, xlator_t *this, loc_t *loc, int flags,
 
     local->transaction.main_frame = frame;
     local->transaction.basename = AFR_BASENAME(loc->path);
-    int_lock = &local->internal_lock;
-
-    int_lock->lockee_count = nlockee = 0;
-    ret = afr_init_entry_lockee(&int_lock->lockee[nlockee], local,
-                                &local->transaction.parent_loc,
-                                local->transaction.basename, priv->child_count);
-    if (ret)
-        goto out;
-
-    nlockee++;
-    ret = afr_init_entry_lockee(&int_lock->lockee[nlockee], local, &local->loc,
-                                NULL, priv->child_count);
-    if (ret)
-        goto out;
-
-    nlockee++;
-    qsort(int_lock->lockee, nlockee, sizeof(*int_lock->lockee),
-          afr_entry_lockee_cmp);
-    int_lock->lockee_count = nlockee;
-
     ret = afr_transaction(transaction_frame, this, AFR_ENTRY_TRANSACTION);
     if (ret < 0) {
         op_errno = -ret;
diff --git a/xlators/cluster/afr/src/afr-inode-read.c b/xlators/cluster/afr/src/afr-inode-read.c
index 03338592618..c5521704de2 100644
--- a/xlators/cluster/afr/src/afr-inode-read.c
+++ b/xlators/cluster/afr/src/afr-inode-read.c
@@ -15,20 +15,17 @@
 #include <stdlib.h>
 #include <signal.h>
 
-#include "glusterfs.h"
+#include <glusterfs/glusterfs.h>
 #include "afr.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "byte-order.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
-#include "quota-common-utils.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/list.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/defaults.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
+#include <glusterfs/quota-common-utils.h>
 
 #include "afr-transaction.h"
 #include "afr-messages.h"
@@ -305,6 +302,7 @@ afr_fstat(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
     afr_local_t *local = NULL;
     int op_errno = 0;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
@@ -531,12 +529,16 @@ afr_fgetxattr_clrlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int32_t callcnt = 0;
     long int cky = 0;
     int ret = 0;
+    int keylen = 0;
+    int children_keylen = 0;
 
     priv = this->private;
     children = priv->children;
 
     local = frame->local;
     cky = (long)cookie;
+    keylen = strlen(local->cont.getxattr.name);
+    children_keylen = strlen(children[cky]->name);
 
     LOCK(&frame->lock);
     {
@@ -547,11 +549,12 @@ afr_fgetxattr_clrlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (!local->dict)
             local->dict = dict_new();
         if (local->dict) {
-            ret = dict_get_str(dict, local->cont.getxattr.name, &tmp_report);
+            ret = dict_get_strn(dict, local->cont.getxattr.name, keylen,
+                                &tmp_report);
             if (ret)
                 goto unlock;
-            ret = dict_set_dynstr(local->dict, children[cky]->name,
-                                  gf_strdup(tmp_report));
+            ret = dict_set_dynstrn(local->dict, children[cky]->name,
+                                   children_keylen, gf_strdup(tmp_report));
             if (ret)
                 goto unlock;
         }
@@ -575,8 +578,8 @@ unlock:
         }
         if (serz_len == -1)
             snprintf(lk_summary, sizeof(lk_summary), "No locks cleared.");
-        ret = dict_set_dynstr(xattr, local->cont.getxattr.name,
-                              gf_strdup(lk_summary));
+        ret = dict_set_dynstrn(xattr, local->cont.getxattr.name, keylen,
+                               gf_strdup(lk_summary));
         if (ret) {
             op_ret = -1;
             op_errno = ENOMEM;
@@ -613,6 +616,8 @@ afr_getxattr_clrlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int32_t callcnt = 0;
     long int cky = 0;
     int ret = 0;
+    int keylen = 0;
+    int children_keylen = 0;
 
     priv = this->private;
     children = priv->children;
@@ -620,6 +625,9 @@ afr_getxattr_clrlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     local = frame->local;
     cky = (long)cookie;
 
+    keylen = strlen(local->cont.getxattr.name);
+    children_keylen = strlen(children[cky]->name);
+
     LOCK(&frame->lock);
     {
         callcnt = --local->call_count;
@@ -629,11 +637,12 @@ afr_getxattr_clrlk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (!local->dict)
             local->dict = dict_new();
         if (local->dict) {
-            ret = dict_get_str(dict, local->cont.getxattr.name, &tmp_report);
+            ret = dict_get_strn(dict, local->cont.getxattr.name, keylen,
+                                &tmp_report);
             if (ret)
                 goto unlock;
-            ret = dict_set_dynstr(local->dict, children[cky]->name,
-                                  gf_strdup(tmp_report));
+            ret = dict_set_dynstrn(local->dict, children[cky]->name,
+                                   children_keylen, gf_strdup(tmp_report));
             if (ret)
                 goto unlock;
         }
@@ -657,8 +666,8 @@ unlock:
         }
         if (serz_len == -1)
             snprintf(lk_summary, sizeof(lk_summary), "No locks cleared.");
-        ret = dict_set_dynstr(xattr, local->cont.getxattr.name,
-                              gf_strdup(lk_summary));
+        ret = dict_set_dynstrn(xattr, local->cont.getxattr.name, keylen,
+                               gf_strdup(lk_summary));
         if (ret) {
             op_ret = -1;
             op_errno = ENOMEM;
@@ -802,15 +811,18 @@ unlock:
         if (ret) {
             local->op_ret = -1;
             local->op_errno = ENOMEM;
+            GF_FREE(xattr_serz);
             goto unwind;
         }
-        ret = dict_set_dynstr(local->dict, GF_XATTR_LIST_NODE_UUIDS_KEY,
-                              xattr_serz);
+        ret = dict_set_dynstr_sizen(local->dict, GF_XATTR_LIST_NODE_UUIDS_KEY,
+                                    xattr_serz);
         if (ret) {
             gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                    "Cannot set node_uuid key in dict");
             local->op_ret = -1;
             local->op_errno = ENOMEM;
+            if (ret == -EINVAL)
+                GF_FREE(xattr_serz);
         } else {
             local->op_ret = local->cont.getxattr.xattr_len - 1;
             local->op_errno = 0;
@@ -934,24 +946,13 @@ unlock:
             goto unwind;
         }
 
-        len = dict_serialized_length(local->dict);
-        if (len <= 0) {
-            goto unwind;
-        }
-
-        lockinfo_buf = GF_CALLOC(1, len, gf_common_mt_char);
-        if (!lockinfo_buf) {
+        op_ret = dict_allocate_and_serialize(
+            local->dict, (char **)&lockinfo_buf, (unsigned int *)&len);
+        if (op_ret != 0) {
             local->op_ret = -1;
-            local->op_errno = ENOMEM;
             goto unwind;
         }
 
-        op_ret = dict_serialize(local->dict, lockinfo_buf);
-        if (op_ret < 0) {
-            local->op_ret = -1;
-            local->op_errno = -op_ret;
-        }
-
         op_ret = dict_set_dynptr(newdict, GF_XATTR_LOCKINFO_KEY,
                                  (void *)lockinfo_buf, len);
         if (op_ret < 0) {
@@ -1050,24 +1051,13 @@ unlock:
             goto unwind;
         }
 
-        len = dict_serialized_length(local->dict);
-        if (len <= 0) {
-            goto unwind;
-        }
-
-        lockinfo_buf = GF_CALLOC(1, len, gf_common_mt_char);
-        if (!lockinfo_buf) {
+        op_ret = dict_allocate_and_serialize(
+            local->dict, (char **)&lockinfo_buf, (unsigned int *)&len);
+        if (op_ret != 0) {
             local->op_ret = -1;
-            local->op_errno = ENOMEM;
             goto unwind;
         }
 
-        op_ret = dict_serialize(local->dict, lockinfo_buf);
-        if (op_ret < 0) {
-            local->op_ret = -1;
-            local->op_errno = -op_ret;
-        }
-
         op_ret = dict_set_dynptr(newdict, GF_XATTR_LOCKINFO_KEY,
                                  (void *)lockinfo_buf, len);
         if (op_ret < 0) {
@@ -1096,9 +1086,11 @@ afr_fgetxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int ret = 0;
     char *xattr = NULL;
     char *xattr_serz = NULL;
+    int keylen = 0;
     char xattr_cky[1024] = {
         0,
     };
+    int xattr_cky_len = 0;
     dict_t *nxattr = NULL;
     long cky = 0;
     int32_t padding = 0;
@@ -1111,7 +1103,9 @@ afr_fgetxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     local = frame->local;
     cky = (long)cookie;
-
+    keylen = strlen(local->cont.getxattr.name);
+    xattr_cky_len = snprintf(xattr_cky, sizeof(xattr_cky), "%s-%ld",
+                             local->cont.getxattr.name, cky);
     LOCK(&frame->lock);
     {
         callcnt = --local->call_count;
@@ -1127,31 +1121,30 @@ afr_fgetxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (!dict || (op_ret < 0))
             goto unlock;
 
-        if (!local->dict)
+        if (!local->dict) {
             local->dict = dict_new();
-
-        if (local->dict) {
-            ret = dict_get_str(dict, local->cont.getxattr.name, &xattr);
-            if (ret)
+            if (!local->dict)
                 goto unlock;
+        }
+        ret = dict_get_strn(dict, local->cont.getxattr.name, keylen, &xattr);
+        if (ret)
+            goto unlock;
 
-            xattr = gf_strdup(xattr);
-
-            (void)snprintf(xattr_cky, sizeof(xattr_cky), "%s-%ld",
-                           local->cont.getxattr.name, cky);
-            ret = dict_set_dynstr(local->dict, xattr_cky, xattr);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
-                       "Cannot set xattr cookie key");
-                goto unlock;
-            }
+        xattr = gf_strdup(xattr);
 
-            local->cont.getxattr.xattr_len += strlen(xattr) + 1;
+        ret = dict_set_dynstrn(local->dict, xattr_cky, xattr_cky_len, xattr);
+        if (ret) {
+            UNLOCK(&frame->lock);
+            gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
+                   "Cannot set xattr cookie key");
+            goto post_unlock;
         }
+
+        local->cont.getxattr.xattr_len += strlen(xattr) + 1;
     }
 unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     if (!callcnt) {
         if (!local->cont.getxattr.xattr_len)
             goto unwind;
@@ -1178,6 +1171,7 @@ unlock:
         ret = dict_serialize_value_with_delim(
             local->dict, xattr_serz + xattr_serz_len, &tlen, ' ');
         if (ret) {
+            GF_FREE(xattr_serz);
             goto unwind;
         }
 
@@ -1185,10 +1179,14 @@ unlock:
         *(xattr_serz + padding + tlen) = ')';
         *(xattr_serz + padding + tlen + 1) = '\0';
 
-        ret = dict_set_dynstr(nxattr, local->cont.getxattr.name, xattr_serz);
-        if (ret)
+        ret = dict_set_dynstrn(nxattr, local->cont.getxattr.name, keylen,
+                               xattr_serz);
+        if (ret) {
             gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                    "Cannot set pathinfo key in dict");
+            if (ret == -EINVAL)
+                GF_FREE(xattr_serz);
+        }
 
     unwind:
         AFR_STACK_UNWIND(fgetxattr, frame, local->op_ret, local->op_errno,
@@ -1215,6 +1213,8 @@ afr_getxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     char xattr_cky[1024] = {
         0,
     };
+    int keylen = 0;
+    int xattr_cky_len = 0;
     dict_t *nxattr = NULL;
     long cky = 0;
     int32_t padding = 0;
@@ -1227,7 +1227,9 @@ afr_getxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     local = frame->local;
     cky = (long)cookie;
-
+    keylen = strlen(local->cont.getxattr.name);
+    xattr_cky_len = snprintf(xattr_cky, sizeof(xattr_cky), "%s-%ld",
+                             local->cont.getxattr.name, cky);
     LOCK(&frame->lock);
     {
         callcnt = --local->call_count;
@@ -1243,32 +1245,30 @@ afr_getxattr_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (!dict || (op_ret < 0))
             goto unlock;
 
-        if (!local->dict)
+        if (!local->dict) {
             local->dict = dict_new();
-
-        if (local->dict) {
-            ret = dict_get_str(dict, local->cont.getxattr.name, &xattr);
-            if (ret)
+            if (!local->dict)
                 goto unlock;
+        }
+        ret = dict_get_strn(dict, local->cont.getxattr.name, keylen, &xattr);
+        if (ret)
+            goto unlock;
 
-            xattr = gf_strdup(xattr);
-
-            (void)snprintf(xattr_cky, 1024, "%s-%ld", local->cont.getxattr.name,
-                           cky);
-            ret = dict_set_dynstr(local->dict, xattr_cky, xattr);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
-                       "Cannot set xattr "
-                       "cookie key");
-                goto unlock;
-            }
+        xattr = gf_strdup(xattr);
 
-            local->cont.getxattr.xattr_len += strlen(xattr) + 1;
+        ret = dict_set_dynstrn(local->dict, xattr_cky, xattr_cky_len, xattr);
+        if (ret) {
+            UNLOCK(&frame->lock);
+            gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
+                   "Cannot set xattr cookie key");
+            goto post_unlock;
         }
+
+        local->cont.getxattr.xattr_len += strlen(xattr) + 1;
     }
 unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     if (!callcnt) {
         if (!local->cont.getxattr.xattr_len)
             goto unwind;
@@ -1295,6 +1295,7 @@ unlock:
         ret = dict_serialize_value_with_delim(
             local->dict, xattr_serz + xattr_serz_len, &tlen, ' ');
         if (ret) {
+            GF_FREE(xattr_serz);
             goto unwind;
         }
 
@@ -1302,10 +1303,14 @@ unlock:
         *(xattr_serz + padding + tlen) = ')';
         *(xattr_serz + padding + tlen + 1) = '\0';
 
-        ret = dict_set_dynstr(nxattr, local->cont.getxattr.name, xattr_serz);
-        if (ret)
+        ret = dict_set_dynstrn(nxattr, local->cont.getxattr.name, keylen,
+                               xattr_serz);
+        if (ret) {
             gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                    "Cannot set pathinfo key in dict");
+            if (ret == -EINVAL)
+                GF_FREE(xattr_serz);
+        }
 
     unwind:
         AFR_STACK_UNWIND(getxattr, frame, local->op_ret, local->op_errno,
@@ -1694,6 +1699,7 @@ afr_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *name,
     int32_t op_errno = 0;
     fop_fgetxattr_cbk_t cbk = NULL;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
@@ -1787,6 +1793,7 @@ afr_readv(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
     afr_local_t *local = NULL;
     int32_t op_errno = 0;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
@@ -1862,6 +1869,7 @@ afr_seek(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
     afr_local_t *local = NULL;
     int32_t op_errno = 0;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
diff --git a/xlators/cluster/afr/src/afr-inode-read.h b/xlators/cluster/afr/src/afr-inode-read.h
index 1627ee2c426..8c982bc7e6f 100644
--- a/xlators/cluster/afr/src/afr-inode-read.h
+++ b/xlators/cluster/afr/src/afr-inode-read.h
@@ -38,5 +38,8 @@ afr_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *name,
               dict_t *xdata);
 
 int
+afr_seek(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
+         gf_seek_what_t what, dict_t *xdata);
+int
 afr_handle_quota_size(call_frame_t *frame, xlator_t *this);
 #endif /* __INODE_READ_H__ */
diff --git a/xlators/cluster/afr/src/afr-inode-write.c b/xlators/cluster/afr/src/afr-inode-write.c
index c6fe0939841..1d6e4f3570a 100644
--- a/xlators/cluster/afr/src/afr-inode-write.c
+++ b/xlators/cluster/afr/src/afr-inode-write.c
@@ -8,27 +8,21 @@
   cases as published by the Free Software Foundation.
 */
 
-#include <libgen.h>
 #include <unistd.h>
-#include <fnmatch.h>
 #include <sys/time.h>
 #include <stdlib.h>
 #include <signal.h>
 
-#include "glusterfs.h"
+#include <glusterfs/glusterfs.h>
 #include "afr.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/defaults.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
 #include "protocol-common.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 #include "afr-transaction.h"
 #include "afr-self-heal.h"
 #include "afr-messages.h"
@@ -180,11 +174,10 @@ __afr_inode_write_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     {
         __afr_inode_write_fill(frame, this, child_index, op_ret, op_errno,
                                prebuf, postbuf, xattr, xdata);
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
     if (call_count == 0) {
         __afr_inode_write_finalize(frame, this);
 
@@ -498,6 +491,7 @@ afr_writev(call_frame_t *frame, xlator_t *this, fd_t *fd, struct iovec *vector,
     int op_errno = ENOMEM;
     int ret = -1;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local)
         goto out;
@@ -737,6 +731,7 @@ afr_ftruncate(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -947,6 +942,7 @@ afr_fsetattr(call_frame_t *frame, xlator_t *this, fd_t *fd, struct iatt *buf,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -1059,15 +1055,14 @@ afr_emptyb_set_pending_changelog_cbk(call_frame_t *frame, void *cookie,
     local->replies[i].op_ret = op_ret;
     local->replies[i].op_errno = op_errno;
 
-    ret = dict_get_str(local->xdata_req, "replicate-brick-op", &op_type);
+    ret = dict_get_str_sizen(local->xdata_req, "replicate-brick-op", &op_type);
     if (ret)
         goto out;
 
-    gf_msg(this->name, op_ret ? GF_LOG_ERROR : GF_LOG_INFO,
-           op_ret ? op_errno : 0, afr_get_msg_id(op_type),
-           "Set of pending xattr %s on"
-           " %s.",
-           op_ret ? "failed" : "succeeded", priv->children[i]->name);
+    gf_smsg(this->name, op_ret ? GF_LOG_ERROR : GF_LOG_INFO,
+            op_ret ? op_errno : 0, AFR_MSG_SET_PEND_XATTR, "name=%s",
+            priv->children[i]->name, "op_ret=%s",
+            op_ret ? "failed" : "succeeded", NULL);
 
 out:
     syncbarrier_wake(&local->barrier);
@@ -1105,10 +1100,10 @@ out:
     return -ret;
 }
 
-int
+static int
 _afr_handle_empty_brick_type(xlator_t *this, call_frame_t *frame, loc_t *loc,
                              int empty_index, afr_transaction_type type,
-                             char *op_type)
+                             char *op_type, const int op_type_len)
 {
     int count = 0;
     int ret = -ENOMEM;
@@ -1139,7 +1134,8 @@ _afr_handle_empty_brick_type(xlator_t *this, call_frame_t *frame, loc_t *loc,
     if (!local->xdata_req)
         goto out;
 
-    ret = dict_set_str(local->xdata_req, "replicate-brick-op", op_type);
+    ret = dict_set_nstrn(local->xdata_req, "replicate-brick-op",
+                         SLEN("replicate-brick-op"), op_type, op_type_len);
     if (ret)
         goto out;
 
@@ -1160,9 +1156,8 @@ _afr_handle_empty_brick_type(xlator_t *this, call_frame_t *frame, loc_t *loc,
     }
 
     if (!count) {
-        gf_msg(this->name, GF_LOG_ERROR, EAGAIN, AFR_MSG_REPLACE_BRICK_STATUS,
-               "Couldn't acquire lock on"
-               " any child.");
+        gf_smsg(this->name, GF_LOG_ERROR, EAGAIN, AFR_MSG_REPLACE_BRICK_STATUS,
+                NULL);
         ret = -EAGAIN;
         goto unlock;
     }
@@ -1211,26 +1206,41 @@ _afr_handle_empty_brick(void *opaque)
     call_frame_t *frame = NULL;
     xlator_t *this = NULL;
     char *op_type = NULL;
+    int op_type_len = 0;
     afr_empty_brick_args_t *data = NULL;
+    call_frame_t *op_frame = NULL;
 
     data = opaque;
     frame = data->frame;
     empty_index = data->empty_index;
+    if (!data->op_type)
+        goto out;
+
+    op_frame = copy_frame(frame);
+    if (!op_frame) {
+        ret = -1;
+        op_errno = ENOMEM;
+        goto out;
+    }
+
     op_type = data->op_type;
-    this = frame->this;
+    op_type_len = strlen(op_type);
+    this = op_frame->this;
     priv = this->private;
 
-    local = AFR_FRAME_INIT(frame, op_errno);
+    afr_set_lk_owner(op_frame, this, op_frame->root);
+    local = AFR_FRAME_INIT(op_frame, op_errno);
     if (!local)
         goto out;
 
     loc_copy(&local->loc, &data->loc);
 
-    gf_msg(this->name, GF_LOG_INFO, 0, 0, "New brick is : %s",
-           priv->children[empty_index]->name);
+    gf_smsg(this->name, GF_LOG_INFO, 0, AFR_MSG_NEW_BRICK, "name=%s",
+            priv->children[empty_index]->name, NULL);
 
-    ret = _afr_handle_empty_brick_type(this, frame, &local->loc, empty_index,
-                                       AFR_METADATA_TRANSACTION, op_type);
+    ret = _afr_handle_empty_brick_type(this, op_frame, &local->loc, empty_index,
+                                       AFR_METADATA_TRANSACTION, op_type,
+                                       op_type_len);
     if (ret) {
         op_errno = -ret;
         ret = -1;
@@ -1244,8 +1254,9 @@ _afr_handle_empty_brick(void *opaque)
     local->xattr_req = NULL;
     local->xdata_req = NULL;
 
-    ret = _afr_handle_empty_brick_type(this, frame, &local->loc, empty_index,
-                                       AFR_ENTRY_TRANSACTION, op_type);
+    ret = _afr_handle_empty_brick_type(this, op_frame, &local->loc, empty_index,
+                                       AFR_ENTRY_TRANSACTION, op_type,
+                                       op_type_len);
     if (ret) {
         op_errno = -ret;
         ret = -1;
@@ -1253,6 +1264,9 @@ _afr_handle_empty_brick(void *opaque)
     }
     ret = 0;
 out:
+    if (op_frame) {
+        AFR_STACK_DESTROY(op_frame);
+    }
     AFR_STACK_UNWIND(setxattr, frame, ret, op_errno, NULL);
     return 0;
 }
@@ -1273,14 +1287,14 @@ afr_split_brain_resolve_do(call_frame_t *frame, xlator_t *this, loc_t *loc,
         goto out;
     }
 
-    ret = dict_set_int32(local->xdata_req, "heal-op",
-                         GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK);
+    ret = dict_set_int32_sizen(local->xdata_req, "heal-op",
+                               GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK);
     if (ret) {
         op_errno = -ret;
         ret = -1;
         goto out;
     }
-    ret = dict_set_str(local->xdata_req, "child-name", data);
+    ret = dict_set_str_sizen(local->xdata_req, "child-name", data);
     if (ret) {
         op_errno = -ret;
         ret = -1;
@@ -1297,9 +1311,8 @@ afr_split_brain_resolve_do(call_frame_t *frame, xlator_t *this, loc_t *loc,
      */
     ret = afr_inode_split_brain_choice_set(loc->inode, this, -1);
     if (ret)
-        gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR,
-               "Failed to set"
-               "split-brain choice to -1");
+        gf_smsg(this->name, GF_LOG_WARNING, 0, AFR_MSG_SPLIT_BRAIN_SET_FAILED,
+                NULL);
     afr_heal_splitbrain_file(frame, this, loc);
     ret = 0;
 out:
@@ -1322,8 +1335,8 @@ afr_get_split_brain_child_index(xlator_t *this, void *value, size_t len)
 
     spb_child_index = afr_get_child_index_from_name(this, spb_child_str);
     if (spb_child_index < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_SUBVOL,
-               "Invalid subvol: %s", spb_child_str);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_SUBVOL,
+                "subvol=%s", spb_child_str, NULL);
     }
     return spb_child_index;
 }
@@ -1345,11 +1358,9 @@ afr_can_set_split_brain_choice(void *opaque)
                              &data->m_spb);
 
     if (ret)
-        gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR,
-               "Failed to determine if %s"
-               " is in split-brain. "
-               "Aborting split-brain-choice set.",
-               uuid_utoa(loc->gfid));
+        gf_smsg(this->name, GF_LOG_ERROR, 0,
+                AFR_MSG_SPLIT_BRAIN_DETERMINE_FAILED, "gfid=%s",
+                uuid_utoa(loc->gfid), NULL);
     return ret;
 }
 
@@ -1357,7 +1368,8 @@ int
 afr_handle_split_brain_commands(xlator_t *this, call_frame_t *frame, loc_t *loc,
                                 dict_t *dict)
 {
-    void *value = NULL;
+    void *choice_value = NULL;
+    void *resolve_value = NULL;
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
     afr_spbc_timeout_t *data = NULL;
@@ -1368,6 +1380,14 @@ afr_handle_split_brain_commands(xlator_t *this, call_frame_t *frame, loc_t *loc,
 
     priv = this->private;
 
+    ret = dict_get_ptr_and_len(dict, GF_AFR_SBRAIN_CHOICE, &choice_value, &len);
+    ret = dict_get_ptr_and_len(dict, GF_AFR_SBRAIN_RESOLVE, &resolve_value,
+                               &len);
+    if (!choice_value && !resolve_value) {
+        ret = -1;
+        goto out;
+    }
+
     local = AFR_FRAME_INIT(frame, op_errno);
     if (!local) {
         ret = 1;
@@ -1376,9 +1396,9 @@ afr_handle_split_brain_commands(xlator_t *this, call_frame_t *frame, loc_t *loc,
 
     local->op = GF_FOP_SETXATTR;
 
-    ret = dict_get_ptr_and_len(dict, GF_AFR_SBRAIN_CHOICE, &value, &len);
-    if (value) {
-        spb_child_index = afr_get_split_brain_child_index(this, value, len);
+    if (choice_value) {
+        spb_child_index = afr_get_split_brain_child_index(this, choice_value,
+                                                          len);
         if (spb_child_index < 0) {
             /* Case where value was "none" */
             if (spb_child_index == -2)
@@ -1402,12 +1422,8 @@ afr_handle_split_brain_commands(xlator_t *this, call_frame_t *frame, loc_t *loc,
         ret = synctask_new(this->ctx->env, afr_can_set_split_brain_choice,
                            afr_set_split_brain_choice, NULL, data);
         if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, 0,
-                   AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR,
-                   "Failed to create"
-                   " synctask. Aborting split-brain choice set"
-                   " for %s",
-                   loc->name);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN_STATUS,
+                    "name=%s", loc->name, NULL);
             ret = 1;
             op_errno = ENOMEM;
             goto out;
@@ -1416,9 +1432,9 @@ afr_handle_split_brain_commands(xlator_t *this, call_frame_t *frame, loc_t *loc,
         goto out;
     }
 
-    ret = dict_get_ptr_and_len(dict, GF_AFR_SBRAIN_RESOLVE, &value, &len);
-    if (value) {
-        spb_child_index = afr_get_split_brain_child_index(this, value, len);
+    if (resolve_value) {
+        spb_child_index = afr_get_split_brain_child_index(this, resolve_value,
+                                                          len);
         if (spb_child_index < 0) {
             ret = 1;
             goto out;
@@ -1470,20 +1486,20 @@ afr_handle_empty_brick(xlator_t *this, call_frame_t *frame, loc_t *loc,
     char *op_type = NULL;
     afr_empty_brick_args_t *data = NULL;
 
-    ret = dict_get_str(dict, GF_AFR_REPLACE_BRICK, &empty_brick);
+    ret = dict_get_str_sizen(dict, GF_AFR_REPLACE_BRICK, &empty_brick);
     if (!ret)
         op_type = GF_AFR_REPLACE_BRICK;
 
-    ab_ret = dict_get_str(dict, GF_AFR_ADD_BRICK, &empty_brick);
+    ab_ret = dict_get_str_sizen(dict, GF_AFR_ADD_BRICK, &empty_brick);
     if (!ab_ret)
         op_type = GF_AFR_ADD_BRICK;
 
     if (ret && ab_ret)
         goto out;
 
-    if (frame->root->pid != GF_CLIENT_PID_SELF_HEALD) {
-        gf_msg(this->name, GF_LOG_ERROR, EPERM, afr_get_msg_id(op_type),
-               "'%s' is an internal extended attribute.", op_type);
+    if (frame->root->pid != GF_CLIENT_PID_ADD_REPLICA_MOUNT) {
+        gf_smsg(this->name, GF_LOG_ERROR, EPERM, AFR_MSG_INTERNAL_ATTR,
+                "op_type=%s", op_type, NULL);
         ret = 1;
         goto out;
     }
@@ -1509,8 +1525,8 @@ afr_handle_empty_brick(xlator_t *this, call_frame_t *frame, loc_t *loc,
         ret = synctask_new(this->ctx->env, _afr_handle_empty_brick,
                            _afr_handle_empty_brick_cbk, NULL, data);
         if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, afr_get_msg_id(op_type),
-                   "Failed to create synctask.");
+            gf_smsg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN_STATUS,
+                    NULL);
             ret = 1;
             op_errno = ENOMEM;
             afr_brick_args_cleanup(data);
@@ -1668,6 +1684,7 @@ afr_fsetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *dict,
 
     GF_IF_INTERNAL_XATTR_GOTO("trusted.glusterfs.afr.*", dict, op_errno, out);
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -1876,6 +1893,7 @@ afr_fremovexattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
 
     GF_IF_NATIVE_XATTR_GOTO("trusted.glusterfs.afr.*", name, op_errno, out);
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -1976,6 +1994,7 @@ afr_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t mode,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -2085,6 +2104,7 @@ afr_discard(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -2191,6 +2211,7 @@ afr_zerofill(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -2222,7 +2243,7 @@ afr_zerofill(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
 
     local->transaction.main_frame = frame;
 
-    local->transaction.start = local->cont.discard.offset;
+    local->transaction.start = local->cont.zerofill.offset;
     local->transaction.len = len;
 
     afr_fix_open(fd, this);
@@ -2390,6 +2411,7 @@ afr_fxattrop(call_frame_t *frame, xlator_t *this, fd_t *fd,
     int ret = -1;
     int op_errno = ENOMEM;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -2484,7 +2506,9 @@ afr_fsync(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t datasync,
     call_frame_t *transaction_frame = NULL;
     int ret = -1;
     int32_t op_errno = ENOMEM;
+    int8_t last_fsync = 0;
 
+    AFR_ERROR_OUT_IF_FDCTX_INVALID(fd, this, op_errno, out);
     transaction_frame = copy_frame(frame);
     if (!transaction_frame)
         goto out;
@@ -2493,10 +2517,16 @@ afr_fsync(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t datasync,
     if (!local)
         goto out;
 
-    if (xdata)
+    if (xdata) {
         local->xdata_req = dict_copy_with_ref(xdata, NULL);
-    else
+        if (dict_get_int8(xdata, "last-fsync", &last_fsync) == 0) {
+            if (last_fsync) {
+                local->transaction.disable_delayed_post_op = _gf_true;
+            }
+        }
+    } else {
         local->xdata_req = dict_new();
+    }
 
     if (!local->xdata_req)
         goto out;
diff --git a/xlators/cluster/afr/src/afr-lk-common.c b/xlators/cluster/afr/src/afr-lk-common.c
index 95e52ff4a09..bc8eabe0f43 100644
--- a/xlators/cluster/afr/src/afr-lk-common.c
+++ b/xlators/cluster/afr/src/afr-lk-common.c
@@ -8,9 +8,9 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "dict.h"
-#include "byte-order.h"
-#include "common-utils.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/common-utils.h>
 
 #include "afr.h"
 #include "afr-transaction.h"
@@ -22,11 +22,40 @@
 #define LOCKED_YES 0x1   /* for DATA, METADATA, ENTRY and higher_path */
 #define LOCKED_LOWER 0x2 /* for lower path */
 
+void
+afr_lockee_cleanup(afr_lockee_t *lockee)
+{
+    if (lockee->fd) {
+        fd_unref(lockee->fd);
+        lockee->fd = NULL;
+    } else {
+        loc_wipe(&lockee->loc);
+    }
+
+    GF_FREE(lockee->basename);
+    lockee->basename = NULL;
+    GF_FREE(lockee->locked_nodes);
+    lockee->locked_nodes = NULL;
+
+    return;
+}
+
+void
+afr_lockees_cleanup(afr_internal_lock_t *int_lock)
+{
+    int i = 0;
+
+    for (i = 0; i < int_lock->lockee_count; i++) {
+        afr_lockee_cleanup(&int_lock->lockee[i]);
+    }
+
+    return;
+}
 int
 afr_entry_lockee_cmp(const void *l1, const void *l2)
 {
-    const afr_entry_lockee_t *r1 = l1;
-    const afr_entry_lockee_t *r2 = l2;
+    const afr_lockee_t *r1 = l1;
+    const afr_lockee_t *r2 = l2;
     int ret = 0;
     uuid_t gfid1 = {0};
     uuid_t gfid2 = {0};
@@ -81,31 +110,14 @@ internal_lock_count(call_frame_t *frame, xlator_t *this)
 }
 
 int
-afr_is_inodelk_transaction(afr_transaction_type type)
-{
-    int ret = 0;
-
-    switch (type) {
-        case AFR_DATA_TRANSACTION:
-        case AFR_METADATA_TRANSACTION:
-            ret = 1;
-            break;
-
-        case AFR_ENTRY_RENAME_TRANSACTION:
-        case AFR_ENTRY_TRANSACTION:
-            ret = 0;
-            break;
-    }
-
-    return ret;
-}
-
-int
-afr_init_entry_lockee(afr_entry_lockee_t *lockee, afr_local_t *local,
-                      loc_t *loc, char *basename, int child_count)
+afr_add_entry_lockee(afr_local_t *local, loc_t *loc, char *basename,
+                     int child_count)
 {
-    int ret = -1;
+    int ret = -ENOMEM;
+    afr_internal_lock_t *int_lock = &local->internal_lock;
+    afr_lockee_t *lockee = &int_lock->lockee[int_lock->lockee_count];
 
+    GF_ASSERT(int_lock->lockee_count < AFR_LOCKEE_COUNT_MAX);
     loc_copy(&lockee->loc, loc);
     lockee->basename = (basename) ? gf_strdup(basename) : NULL;
     if (basename && !lockee->basename)
@@ -119,28 +131,45 @@ afr_init_entry_lockee(afr_entry_lockee_t *lockee, afr_local_t *local,
         goto out;
 
     ret = 0;
+    int_lock->lockee_count++;
 out:
+    if (ret) {
+        afr_lockee_cleanup(lockee);
+    }
     return ret;
 }
 
-void
-afr_entry_lockee_cleanup(afr_internal_lock_t *int_lock)
+int
+afr_add_inode_lockee(afr_local_t *local, int child_count)
 {
-    int i = 0;
+    int ret = -ENOMEM;
+    afr_internal_lock_t *int_lock = &local->internal_lock;
+    afr_lockee_t *lockee = &int_lock->lockee[int_lock->lockee_count];
 
-    for (i = 0; i < int_lock->lockee_count; i++) {
-        loc_wipe(&int_lock->lockee[i].loc);
-        if (int_lock->lockee[i].basename)
-            GF_FREE(int_lock->lockee[i].basename);
-        if (int_lock->lockee[i].locked_nodes)
-            GF_FREE(int_lock->lockee[i].locked_nodes);
+    if (local->fd) {
+        lockee->fd = fd_ref(local->fd);
+    } else {
+        loc_copy(&lockee->loc, &local->loc);
     }
 
-    return;
+    lockee->locked_count = 0;
+    lockee->locked_nodes = GF_CALLOC(child_count, sizeof(*lockee->locked_nodes),
+                                     gf_afr_mt_afr_node_character);
+
+    if (!lockee->locked_nodes)
+        goto out;
+
+    ret = 0;
+    int_lock->lockee_count++;
+out:
+    if (ret) {
+        afr_lockee_cleanup(lockee);
+    }
+    return ret;
 }
 
 static int
-initialize_entrylk_variables(call_frame_t *frame, xlator_t *this)
+initialize_internal_lock_variables(call_frame_t *frame, xlator_t *this)
 {
     afr_local_t *local = NULL;
     afr_internal_lock_t *int_lock = NULL;
@@ -152,9 +181,10 @@ initialize_entrylk_variables(call_frame_t *frame, xlator_t *this)
     local = frame->local;
     int_lock = &local->internal_lock;
 
-    int_lock->entrylk_lock_count = 0;
+    int_lock->lock_count = 0;
     int_lock->lock_op_ret = -1;
     int_lock->lock_op_errno = 0;
+    int_lock->lk_attempted_count = 0;
 
     for (i = 0; i < AFR_LOCKEE_COUNT_MAX; i++) {
         if (!int_lock->lockee[i].locked_nodes)
@@ -167,28 +197,6 @@ initialize_entrylk_variables(call_frame_t *frame, xlator_t *this)
     return 0;
 }
 
-static int
-initialize_inodelk_variables(call_frame_t *frame, xlator_t *this)
-{
-    afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
-    afr_private_t *priv = NULL;
-
-    priv = this->private;
-    local = frame->local;
-    int_lock = &local->internal_lock;
-
-    int_lock->lock_count = 0;
-    int_lock->lk_attempted_count = 0;
-    int_lock->lock_op_ret = -1;
-    int_lock->lock_op_errno = 0;
-
-    memset(int_lock->locked_nodes, 0,
-           sizeof(*int_lock->locked_nodes) * priv->child_count);
-
-    return 0;
-}
-
 int
 afr_lockee_locked_nodes_count(afr_internal_lock_t *int_lock)
 {
@@ -216,19 +224,74 @@ afr_locked_nodes_count(unsigned char *locked_nodes, int child_count)
     return call_count;
 }
 
-/* FIXME: What if UNLOCK fails */
+static void
+afr_log_locks_failure(call_frame_t *frame, char *where, char *what,
+                      int op_errno)
+{
+    xlator_t *this = frame->this;
+    gf_lkowner_t *lk_owner = &frame->root->lk_owner;
+    afr_local_t *local = frame->local;
+    const char *fop = NULL;
+    char *gfid = NULL;
+    const char *name = NULL;
+
+    fop = gf_fop_list[local->op];
+
+    switch (local->transaction.type) {
+        case AFR_ENTRY_RENAME_TRANSACTION:
+        case AFR_ENTRY_TRANSACTION:
+            switch (local->op) {
+                case GF_FOP_LINK:
+                    gfid = uuid_utoa(local->newloc.pargfid);
+                    name = local->newloc.name;
+                    break;
+                default:
+                    gfid = uuid_utoa(local->loc.pargfid);
+                    name = local->loc.name;
+                    break;
+            }
+            gf_msg(this->name, GF_LOG_WARNING, op_errno,
+                   AFR_MSG_INTERNAL_LKS_FAILED,
+                   "Unable to do entry %s with lk-owner:%s on %s "
+                   "while attempting %s on {pgfid:%s, name:%s}.",
+                   what, lkowner_utoa(lk_owner), where, fop, gfid, name);
+            break;
+        case AFR_DATA_TRANSACTION:
+        case AFR_METADATA_TRANSACTION:
+            gfid = uuid_utoa(local->inode->gfid);
+            gf_msg(this->name, GF_LOG_WARNING, op_errno,
+                   AFR_MSG_INTERNAL_LKS_FAILED,
+                   "Unable to do inode %s with lk-owner:%s on %s "
+                   "while attempting %s on gfid:%s.",
+                   what, lkowner_utoa(lk_owner), where, fop, gfid);
+            break;
+    }
+}
+
 static int32_t
 afr_unlock_common_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
     afr_local_t *local = NULL;
+    afr_private_t *priv = NULL;
     afr_internal_lock_t *int_lock = NULL;
+    int lockee_num = 0;
     int call_count = 0;
+    int child_index = 0;
     int ret = 0;
 
     local = frame->local;
     int_lock = &local->internal_lock;
+    priv = this->private;
+    lockee_num = (int)((long)cookie) / priv->child_count;
+    child_index = (int)((long)cookie) % priv->child_count;
+
+    if (op_ret < 0 && op_errno != ENOTCONN && op_errno != EBADFD) {
+        afr_log_locks_failure(frame, priv->children[child_index]->name,
+                              "unlock", op_errno);
+    }
 
+    int_lock->lockee[lockee_num].locked_nodes[child_index] &= LOCKED_NO;
     if (local->transaction.type == AFR_DATA_TRANSACTION && op_ret != 1)
         ret = afr_write_subvol_reset(frame, this);
 
@@ -239,7 +302,6 @@ afr_unlock_common_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     UNLOCK(&frame->lock);
 
     if (call_count == 0) {
-        gf_msg_trace(this->name, 0, "All internal locks unlocked");
         int_lock->lock_cbk(frame, this);
     }
 
@@ -247,143 +309,88 @@ afr_unlock_common_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 }
 
 void
-afr_update_uninodelk(afr_local_t *local, afr_internal_lock_t *int_lock,
-                     int32_t child_index)
-{
-    int_lock->locked_nodes[child_index] &= LOCKED_NO;
-}
-
-static int32_t
-afr_unlock_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    afr_local_t *local = NULL;
-    afr_internal_lock_t *int_lock = NULL;
-    int32_t child_index = (long)cookie;
-    afr_private_t *priv = NULL;
-
-    local = frame->local;
-    int_lock = &local->internal_lock;
-
-    priv = this->private;
-
-    if (op_ret < 0 && op_errno != ENOTCONN && op_errno != EBADFD) {
-        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_UNLOCK_FAIL,
-               "path=%s gfid=%s: unlock failed on subvolume %s "
-               "with lock owner %s",
-               local->loc.path, loc_gfid_utoa(&(local->loc)),
-               priv->children[child_index]->name,
-               lkowner_utoa(&frame->root->lk_owner));
-    }
-
-    afr_update_uninodelk(local, int_lock, child_index);
-
-    afr_unlock_common_cbk(frame, cookie, this, op_ret, op_errno, xdata);
-
-    return 0;
-}
-
-static int
-afr_unlock_inodelk(call_frame_t *frame, xlator_t *this)
+afr_internal_lock_wind(call_frame_t *frame,
+                       int32_t (*cbk)(call_frame_t *, void *, xlator_t *,
+                                      int32_t, int32_t, dict_t *),
+                       void *cookie, int child, int lockee_num,
+                       gf_boolean_t blocking, gf_boolean_t unlock)
 {
-    afr_internal_lock_t *int_lock = NULL;
-    afr_local_t *local = NULL;
-    afr_private_t *priv = NULL;
+    afr_local_t *local = frame->local;
+    xlator_t *this = frame->this;
+    afr_private_t *priv = this->private;
+    afr_internal_lock_t *int_lock = &local->internal_lock;
+    entrylk_cmd cmd = ENTRYLK_LOCK_NB;
+    int32_t cmd1 = F_SETLK;
     struct gf_flock flock = {
         0,
     };
-    int call_count = 0;
-    int i = 0;
-
-    local = frame->local;
-    int_lock = &local->internal_lock;
-    priv = this->private;
-
-    flock.l_start = int_lock->flock.l_start;
-    flock.l_len = int_lock->flock.l_len;
-    flock.l_type = F_UNLCK;
-
-    call_count = afr_locked_nodes_count(int_lock->locked_nodes,
-                                        priv->child_count);
-
-    int_lock->lk_call_count = call_count;
-
-    if (!call_count) {
-        GF_ASSERT(!local->transaction.do_eager_unlock);
-        gf_msg_trace(this->name, 0, "No internal locks unlocked");
-
-        int_lock->lock_cbk(frame, this);
-        goto out;
-    }
 
-    for (i = 0; i < priv->child_count; i++) {
-        if ((int_lock->locked_nodes[i] & LOCKED_YES) != LOCKED_YES)
-            continue;
-
-        if (local->fd) {
-            STACK_WIND_COOKIE(
-                frame, afr_unlock_inodelk_cbk, (void *)(long)i,
-                priv->children[i], priv->children[i]->fops->finodelk,
-                int_lock->domain, local->fd, F_SETLK, &flock, NULL);
-        } else {
-            STACK_WIND_COOKIE(
-                frame, afr_unlock_inodelk_cbk, (void *)(long)i,
-                priv->children[i], priv->children[i]->fops->inodelk,
-                int_lock->domain, &local->loc, F_SETLK, &flock, NULL);
-        }
+    switch (local->transaction.type) {
+        case AFR_ENTRY_TRANSACTION:
+        case AFR_ENTRY_RENAME_TRANSACTION:
+            if (unlock) {
+                cmd = ENTRYLK_UNLOCK;
+            } else if (blocking) { /*Doesn't make sense to have blocking
+                                      unlock*/
+                cmd = ENTRYLK_LOCK;
+            }
 
-        if (!--call_count)
+            if (local->fd) {
+                STACK_WIND_COOKIE(frame, cbk, cookie, priv->children[child],
+                                  priv->children[child]->fops->fentrylk,
+                                  int_lock->domain,
+                                  int_lock->lockee[lockee_num].fd,
+                                  int_lock->lockee[lockee_num].basename, cmd,
+                                  ENTRYLK_WRLCK, NULL);
+            } else {
+                STACK_WIND_COOKIE(frame, cbk, cookie, priv->children[child],
+                                  priv->children[child]->fops->entrylk,
+                                  int_lock->domain,
+                                  &int_lock->lockee[lockee_num].loc,
+                                  int_lock->lockee[lockee_num].basename, cmd,
+                                  ENTRYLK_WRLCK, NULL);
+            }
             break;
-    }
-out:
-    return 0;
-}
-
-static int32_t
-afr_unlock_entrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    afr_local_t *local = NULL;
-    afr_private_t *priv = NULL;
-    afr_internal_lock_t *int_lock = NULL;
-    int32_t child_index = 0;
-    int lockee_no = 0;
-
-    priv = this->private;
-    lockee_no = (int)((long)cookie) / priv->child_count;
-    child_index = (int)((long)cookie) % priv->child_count;
 
-    local = frame->local;
-    int_lock = &local->internal_lock;
+        case AFR_DATA_TRANSACTION:
+        case AFR_METADATA_TRANSACTION:
+            flock = int_lock->lockee[lockee_num].flock;
+            if (unlock) {
+                flock.l_type = F_UNLCK;
+            } else if (blocking) { /*Doesn't make sense to have blocking
+                                      unlock*/
+                cmd1 = F_SETLKW;
+            }
 
-    if (op_ret < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, op_errno, AFR_MSG_ENTRY_UNLOCK_FAIL,
-               "%s: unlock failed on %s", local->loc.path,
-               priv->children[child_index]->name);
+            if (local->fd) {
+                STACK_WIND_COOKIE(
+                    frame, cbk, cookie, priv->children[child],
+                    priv->children[child]->fops->finodelk, int_lock->domain,
+                    int_lock->lockee[lockee_num].fd, cmd1, &flock, NULL);
+            } else {
+                STACK_WIND_COOKIE(
+                    frame, cbk, cookie, priv->children[child],
+                    priv->children[child]->fops->inodelk, int_lock->domain,
+                    &int_lock->lockee[lockee_num].loc, cmd1, &flock, NULL);
+            }
+            break;
     }
-
-    int_lock->lockee[lockee_no].locked_nodes[child_index] &= LOCKED_NO;
-    afr_unlock_common_cbk(frame, cookie, this, op_ret, op_errno, NULL);
-
-    return 0;
 }
 
 static int
-afr_unlock_entrylk(call_frame_t *frame, xlator_t *this)
+afr_unlock_now(call_frame_t *frame, xlator_t *this)
 {
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
     int call_count = 0;
-    int index = 0;
-    int lockee_no = 0;
-    int copies = 0;
+    int child_index = 0;
+    int lockee_num = 0;
     int i = -1;
 
     local = frame->local;
     int_lock = &local->internal_lock;
     priv = this->private;
-    copies = priv->child_count;
 
     call_count = afr_lockee_locked_nodes_count(int_lock);
 
@@ -396,16 +403,13 @@ afr_unlock_entrylk(call_frame_t *frame, xlator_t *this)
     }
 
     for (i = 0; i < int_lock->lockee_count * priv->child_count; i++) {
-        lockee_no = i / copies;
-        index = i % copies;
-        if (int_lock->lockee[lockee_no].locked_nodes[index] & LOCKED_YES) {
-            STACK_WIND_COOKIE(
-                frame, afr_unlock_entrylk_cbk, (void *)(long)i,
-                priv->children[index], priv->children[index]->fops->entrylk,
-                int_lock->domain, &int_lock->lockee[lockee_no].loc,
-                int_lock->lockee[lockee_no].basename, ENTRYLK_UNLOCK,
-                ENTRYLK_WRLCK, NULL);
-
+        lockee_num = i / priv->child_count;
+        child_index = i % priv->child_count;
+        if (int_lock->lockee[lockee_num].locked_nodes[child_index] &
+            LOCKED_YES) {
+            afr_internal_lock_wind(frame, afr_unlock_common_cbk,
+                                   (void *)(long)i, child_index, lockee_num,
+                                   _gf_false, _gf_true);
             if (!--call_count)
                 break;
         }
@@ -415,18 +419,6 @@ out:
     return 0;
 }
 
-int32_t
-afr_unlock_now(call_frame_t *frame, xlator_t *this)
-{
-    afr_local_t *local = frame->local;
-
-    if (afr_is_inodelk_transaction(local->transaction.type))
-        afr_unlock_inodelk(frame, this);
-    else
-        afr_unlock_entrylk(frame, this);
-    return 0;
-}
-
 static int32_t
 afr_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
              int32_t op_errno, dict_t *xdata)
@@ -436,14 +428,14 @@ afr_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
     afr_private_t *priv = NULL;
     int cky = (long)cookie;
     int child_index = 0;
-    int lockee_no = 0;
+    int lockee_num = 0;
 
     priv = this->private;
     local = frame->local;
     int_lock = &local->internal_lock;
 
     child_index = ((int)cky) % priv->child_count;
-    lockee_no = ((int)cky) / priv->child_count;
+    lockee_num = ((int)cky) / priv->child_count;
 
     LOCK(&frame->lock);
     {
@@ -470,23 +462,16 @@ afr_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
         afr_unlock_now(frame, this);
     } else {
         if (op_ret == 0) {
-            if (local->transaction.type == AFR_ENTRY_TRANSACTION ||
-                local->transaction.type == AFR_ENTRY_RENAME_TRANSACTION) {
-                int_lock->lockee[lockee_no]
-                    .locked_nodes[child_index] |= LOCKED_YES;
-                int_lock->lockee[lockee_no].locked_count++;
-                int_lock->entrylk_lock_count++;
-            } else {
-                int_lock->locked_nodes[child_index] |= LOCKED_YES;
-                int_lock->lock_count++;
-
-                if (local->transaction.type == AFR_DATA_TRANSACTION) {
-                    LOCK(&local->inode->lock);
-                    {
-                        local->inode_ctx->lock_count++;
-                    }
-                    UNLOCK(&local->inode->lock);
+            int_lock->lockee[lockee_num]
+                .locked_nodes[child_index] |= LOCKED_YES;
+            int_lock->lockee[lockee_num].locked_count++;
+            int_lock->lock_count++;
+            if (local->transaction.type == AFR_DATA_TRANSACTION) {
+                LOCK(&local->inode->lock);
+                {
+                    local->inode_ctx->lock_count++;
                 }
+                UNLOCK(&local->inode->lock);
             }
         }
         afr_lock_blocking(frame, this, cky + 1);
@@ -495,30 +480,6 @@ afr_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
     return 0;
 }
 
-static int32_t
-afr_blocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                         int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    afr_lock_cbk(frame, cookie, this, op_ret, op_errno, xdata);
-    return 0;
-}
-
-static int32_t
-afr_blocking_entrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                         int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    afr_lock_cbk(frame, cookie, this, op_ret, op_errno, xdata);
-    return 0;
-}
-
-static gf_boolean_t
-afr_is_entrylk(afr_transaction_type trans_type)
-{
-    if (afr_is_inodelk_transaction(trans_type))
-        return _gf_false;
-    return _gf_true;
-}
-
 static gf_boolean_t
 _is_lock_wind_needed(afr_local_t *local, int child_index)
 {
@@ -528,40 +489,12 @@ _is_lock_wind_needed(afr_local_t *local, int child_index)
     return _gf_true;
 }
 
-static void
-afr_log_entry_locks_failure(xlator_t *this, afr_local_t *local,
-                            afr_internal_lock_t *int_lock)
-{
-    const char *fop = NULL;
-    char *pargfid = NULL;
-    const char *name = NULL;
-
-    fop = gf_fop_list[local->op];
-
-    switch (local->op) {
-        case GF_FOP_LINK:
-            pargfid = uuid_utoa(local->newloc.pargfid);
-            name = local->newloc.name;
-            break;
-        default:
-            pargfid = uuid_utoa(local->loc.pargfid);
-            name = local->loc.name;
-            break;
-    }
-
-    gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_BLOCKING_LKS_FAILED,
-           "Unable to obtain sufficient blocking entry locks on at least "
-           "one child while attempting %s on {pgfid:%s, name:%s}.",
-           fop, pargfid, name);
-}
-
 static gf_boolean_t
 is_blocking_locks_count_sufficient(call_frame_t *frame, xlator_t *this)
 {
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
     afr_internal_lock_t *int_lock = NULL;
-    gf_boolean_t is_entrylk = _gf_false;
     int child = 0;
     int nlockee = 0;
     int lockee_count = 0;
@@ -571,42 +504,26 @@ is_blocking_locks_count_sufficient(call_frame_t *frame, xlator_t *this)
     priv = this->private;
     int_lock = &local->internal_lock;
     lockee_count = int_lock->lockee_count;
-    is_entrylk = afr_is_entrylk(local->transaction.type);
-
-    if (!is_entrylk) {
-        if (int_lock->lock_count == 0) {
-            gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_BLOCKING_LKS_FAILED,
-                   "Unable to obtain "
-                   "blocking inode lock on even one child for "
-                   "gfid:%s.",
-                   uuid_utoa(local->inode->gfid));
-            return _gf_false;
-        } else {
-            /*inodelk succeeded on at least one child. */
-            return _gf_true;
-        }
 
-    } else {
-        if (int_lock->entrylk_lock_count == 0) {
-            afr_log_entry_locks_failure(this, local, int_lock);
-            return _gf_false;
-        }
-        /* For FOPS that take multiple sets of locks (mkdir, rename),
-         * there must be at least one brick on which the locks from
-         * all lock sets were successful. */
-        for (child = 0; child < priv->child_count; child++) {
-            ret = _gf_true;
-            for (nlockee = 0; nlockee < lockee_count; nlockee++) {
-                if (!(int_lock->lockee[nlockee].locked_nodes[child] &
-                      LOCKED_YES))
-                    ret = _gf_false;
-            }
-            if (ret)
-                return ret;
+    if (int_lock->lock_count == 0) {
+        afr_log_locks_failure(frame, "any subvolume", "lock",
+                              int_lock->lock_op_errno);
+        return _gf_false;
+    }
+    /* For FOPS that take multiple sets of locks (mkdir, rename),
+     * there must be at least one brick on which the locks from
+     * all lock sets were successful. */
+    for (child = 0; child < priv->child_count; child++) {
+        ret = _gf_true;
+        for (nlockee = 0; nlockee < lockee_count; nlockee++) {
+            if (!(int_lock->lockee[nlockee].locked_nodes[child] & LOCKED_YES))
+                ret = _gf_false;
         }
-        if (!ret)
-            afr_log_entry_locks_failure(this, local, int_lock);
+        if (ret)
+            return ret;
     }
+    if (!ret)
+        afr_log_locks_failure(frame, "all", "lock", int_lock->lock_op_errno);
 
     return ret;
 }
@@ -617,27 +534,16 @@ afr_lock_blocking(call_frame_t *frame, xlator_t *this, int cookie)
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
-    struct gf_flock flock = {
-        0,
-    };
     uint64_t ctx = 0;
     int ret = 0;
     int child_index = 0;
-    int lockee_no = 0;
-    gf_boolean_t is_entrylk = _gf_false;
+    int lockee_num = 0;
 
     local = frame->local;
     int_lock = &local->internal_lock;
     priv = this->private;
     child_index = cookie % priv->child_count;
-    lockee_no = cookie / priv->child_count;
-    is_entrylk = afr_is_entrylk(local->transaction.type);
-
-    if (!is_entrylk) {
-        flock.l_start = int_lock->flock.l_start;
-        flock.l_len = int_lock->flock.l_len;
-        flock.l_type = int_lock->flock.l_type;
-    }
+    lockee_num = cookie / priv->child_count;
 
     if (local->fd) {
         ret = fd_ctx_get(local->fd, this, &ctx);
@@ -681,52 +587,8 @@ afr_lock_blocking(call_frame_t *frame, xlator_t *this, int cookie)
         return 0;
     }
 
-    switch (local->transaction.type) {
-        case AFR_DATA_TRANSACTION:
-        case AFR_METADATA_TRANSACTION:
-
-            if (local->fd) {
-                STACK_WIND_COOKIE(
-                    frame, afr_blocking_inodelk_cbk, (void *)(long)child_index,
-                    priv->children[child_index],
-                    priv->children[child_index]->fops->finodelk,
-                    int_lock->domain, local->fd, F_SETLKW, &flock, NULL);
-
-            } else {
-                STACK_WIND_COOKIE(
-                    frame, afr_blocking_inodelk_cbk, (void *)(long)child_index,
-                    priv->children[child_index],
-                    priv->children[child_index]->fops->inodelk,
-                    int_lock->domain, &local->loc, F_SETLKW, &flock, NULL);
-            }
-
-            break;
-
-        case AFR_ENTRY_RENAME_TRANSACTION:
-        case AFR_ENTRY_TRANSACTION:
-            /*Accounting for child_index increments on 'down'
-             *and 'fd-less' children */
-
-            if (local->fd) {
-                STACK_WIND_COOKIE(frame, afr_blocking_entrylk_cbk,
-                                  (void *)(long)cookie,
-                                  priv->children[child_index],
-                                  priv->children[child_index]->fops->fentrylk,
-                                  int_lock->domain, local->fd,
-                                  int_lock->lockee[lockee_no].basename,
-                                  ENTRYLK_LOCK, ENTRYLK_WRLCK, NULL);
-            } else {
-                STACK_WIND_COOKIE(
-                    frame, afr_blocking_entrylk_cbk, (void *)(long)cookie,
-                    priv->children[child_index],
-                    priv->children[child_index]->fops->entrylk,
-                    int_lock->domain, &int_lock->lockee[lockee_no].loc,
-                    int_lock->lockee[lockee_no].basename, ENTRYLK_LOCK,
-                    ENTRYLK_WRLCK, NULL);
-            }
-
-            break;
-    }
+    afr_internal_lock_wind(frame, afr_lock_cbk, (void *)(long)cookie,
+                           child_index, lockee_num, _gf_true, _gf_false);
 
     return 0;
 }
@@ -743,20 +605,10 @@ afr_blocking_lock(call_frame_t *frame, xlator_t *this)
     local = frame->local;
     int_lock = &local->internal_lock;
 
-    switch (local->transaction.type) {
-        case AFR_DATA_TRANSACTION:
-        case AFR_METADATA_TRANSACTION:
-            initialize_inodelk_variables(frame, this);
-            break;
-
-        case AFR_ENTRY_RENAME_TRANSACTION:
-        case AFR_ENTRY_TRANSACTION:
-            up_count = AFR_COUNT(local->child_up, priv->child_count);
-            int_lock->lk_call_count = int_lock->lk_expected_count =
-                (int_lock->lockee_count * up_count);
-            initialize_entrylk_variables(frame, this);
-            break;
-    }
+    up_count = AFR_COUNT(local->child_up, priv->child_count);
+    int_lock->lk_call_count = int_lock->lk_expected_count =
+        (int_lock->lockee_count * up_count);
+    initialize_internal_lock_variables(frame, this);
 
     afr_lock_blocking(frame, this, 0);
 
@@ -764,171 +616,20 @@ afr_blocking_lock(call_frame_t *frame, xlator_t *this)
 }
 
 static int32_t
-afr_nonblocking_entrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                            int32_t op_ret, int32_t op_errno, dict_t *xdata)
+afr_nb_internal_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                         int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
     int call_count = 0;
-    int child_index = (long)cookie;
-    int copies = 0;
-    int index = 0;
-    int lockee_no = 0;
-    afr_private_t *priv = NULL;
-
-    priv = this->private;
-
-    copies = priv->child_count;
-    index = child_index % copies;
-    lockee_no = child_index / copies;
-
-    local = frame->local;
-    int_lock = &local->internal_lock;
-
-    LOCK(&frame->lock);
-    {
-        if (op_ret < 0) {
-            if (op_errno == ENOSYS) {
-                /* return ENOTSUP */
-                gf_msg(this->name, GF_LOG_ERROR, ENOSYS,
-                       AFR_MSG_LOCK_XLATOR_NOT_LOADED,
-                       "subvolume does not support "
-                       "locking. please load features/locks"
-                       " xlator on server");
-                local->op_ret = op_ret;
-                int_lock->lock_op_ret = op_ret;
-
-                int_lock->lock_op_errno = op_errno;
-                local->op_errno = op_errno;
-            }
-        } else if (op_ret == 0) {
-            int_lock->lockee[lockee_no].locked_nodes[index] |= LOCKED_YES;
-            int_lock->lockee[lockee_no].locked_count++;
-            int_lock->entrylk_lock_count++;
-        }
-
-        call_count = --int_lock->lk_call_count;
-    }
-    UNLOCK(&frame->lock);
-
-    if (call_count == 0) {
-        gf_msg_trace(this->name, 0, "Last locking reply received");
-        /* all locks successful. Proceed to call FOP */
-        if (int_lock->entrylk_lock_count == int_lock->lk_expected_count) {
-            gf_msg_trace(this->name, 0, "All servers locked. Calling the cbk");
-            int_lock->lock_op_ret = 0;
-            int_lock->lock_cbk(frame, this);
-        }
-        /* Not all locks were successful. Unlock and try locking
-           again, this time with serially blocking locks */
-        else {
-            gf_msg_trace(this->name, 0,
-                         "%d servers locked. Trying again "
-                         "with blocking calls",
-                         int_lock->lock_count);
-
-            afr_unlock_now(frame, this);
-        }
-    }
-
-    return 0;
-}
-
-int
-afr_nonblocking_entrylk(call_frame_t *frame, xlator_t *this)
-{
-    afr_internal_lock_t *int_lock = NULL;
-    afr_local_t *local = NULL;
+    int child_index = 0;
+    int lockee_num = 0;
     afr_private_t *priv = NULL;
-    afr_fd_ctx_t *fd_ctx = NULL;
-    int copies = 0;
-    int index = 0;
-    int lockee_no = 0;
-    int32_t call_count = 0;
-    int i = 0;
 
-    local = frame->local;
-    int_lock = &local->internal_lock;
     priv = this->private;
 
-    copies = priv->child_count;
-    initialize_entrylk_variables(frame, this);
-
-    if (local->fd) {
-        fd_ctx = afr_fd_ctx_get(local->fd, this);
-        if (!fd_ctx) {
-            gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_FD_CTX_GET_FAILED,
-                   "unable to get fd ctx for fd=%p", local->fd);
-
-            local->op_ret = -1;
-            int_lock->lock_op_ret = -1;
-            local->op_errno = EINVAL;
-            int_lock->lock_op_errno = EINVAL;
-
-            afr_unlock_now(frame, this);
-            return -1;
-        }
-
-        call_count = int_lock->lockee_count * internal_lock_count(frame, this);
-        int_lock->lk_call_count = call_count;
-        int_lock->lk_expected_count = call_count;
-
-        if (!call_count) {
-            gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_INFO_COMMON,
-                   "fd not open on any subvolumes. aborting.");
-            afr_unlock_now(frame, this);
-            goto out;
-        }
-
-        /* Send non-blocking entrylk calls only on up children
-           and where the fd has been opened */
-        for (i = 0; i < int_lock->lockee_count * priv->child_count; i++) {
-            index = i % copies;
-            lockee_no = i / copies;
-            if (local->child_up[index]) {
-                STACK_WIND_COOKIE(frame, afr_nonblocking_entrylk_cbk,
-                                  (void *)(long)i, priv->children[index],
-                                  priv->children[index]->fops->fentrylk,
-                                  this->name, local->fd,
-                                  int_lock->lockee[lockee_no].basename,
-                                  ENTRYLK_LOCK_NB, ENTRYLK_WRLCK, NULL);
-                if (!--call_count)
-                    break;
-            }
-        }
-    } else {
-        call_count = int_lock->lockee_count * internal_lock_count(frame, this);
-        int_lock->lk_call_count = call_count;
-        int_lock->lk_expected_count = call_count;
-
-        for (i = 0; i < int_lock->lockee_count * priv->child_count; i++) {
-            index = i % copies;
-            lockee_no = i / copies;
-            if (local->child_up[index]) {
-                STACK_WIND_COOKIE(frame, afr_nonblocking_entrylk_cbk,
-                                  (void *)(long)i, priv->children[index],
-                                  priv->children[index]->fops->entrylk,
-                                  this->name, &int_lock->lockee[lockee_no].loc,
-                                  int_lock->lockee[lockee_no].basename,
-                                  ENTRYLK_LOCK_NB, ENTRYLK_WRLCK, NULL);
-
-                if (!--call_count)
-                    break;
-            }
-        }
-    }
-out:
-    return 0;
-}
-
-int32_t
-afr_nonblocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                            int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    afr_internal_lock_t *int_lock = NULL;
-    afr_local_t *local = NULL;
-    int call_count = 0;
-    int child_index = (long)cookie;
+    child_index = ((long)cookie) % priv->child_count;
+    lockee_num = ((long)cookie) / priv->child_count;
 
     local = frame->local;
     int_lock = &local->internal_lock;
@@ -953,11 +654,14 @@ afr_nonblocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                        " xlator on server");
                 local->op_ret = op_ret;
                 int_lock->lock_op_ret = op_ret;
+
                 int_lock->lock_op_errno = op_errno;
                 local->op_errno = op_errno;
             }
-        } else {
-            int_lock->locked_nodes[child_index] |= LOCKED_YES;
+        } else if (op_ret == 0) {
+            int_lock->lockee[lockee_num]
+                .locked_nodes[child_index] |= LOCKED_YES;
+            int_lock->lockee[lockee_num].locked_count++;
             int_lock->lock_count++;
         }
 
@@ -966,7 +670,7 @@ afr_nonblocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     UNLOCK(&frame->lock);
 
     if (call_count == 0) {
-        gf_msg_trace(this->name, 0, "Last inode locking reply received");
+        gf_msg_trace(this->name, 0, "Last locking reply received");
         /* all locks successful. Proceed to call FOP */
         if (int_lock->lock_count == int_lock->lk_expected_count) {
             gf_msg_trace(this->name, 0, "All servers locked. Calling the cbk");
@@ -977,8 +681,8 @@ afr_nonblocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
            again, this time with serially blocking locks */
         else {
             gf_msg_trace(this->name, 0,
-                         "%d servers locked. "
-                         "Trying again with blocking calls",
+                         "%d servers locked. Trying again "
+                         "with blocking calls",
                          int_lock->lock_count);
 
             afr_unlock_now(frame, this);
@@ -989,12 +693,14 @@ afr_nonblocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 }
 
 int
-afr_nonblocking_inodelk(call_frame_t *frame, xlator_t *this)
+afr_lock_nonblocking(call_frame_t *frame, xlator_t *this)
 {
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
     afr_fd_ctx_t *fd_ctx = NULL;
+    int child = 0;
+    int lockee_num = 0;
     int32_t call_count = 0;
     int i = 0;
     int ret = 0;
@@ -1003,7 +709,7 @@ afr_nonblocking_inodelk(call_frame_t *frame, xlator_t *this)
     int_lock = &local->internal_lock;
     priv = this->private;
 
-    initialize_inodelk_variables(frame, this);
+    initialize_internal_lock_variables(frame, this);
 
     if (local->fd) {
         fd_ctx = afr_fd_ctx_get(local->fd, this);
@@ -1022,36 +728,29 @@ afr_nonblocking_inodelk(call_frame_t *frame, xlator_t *this)
         }
     }
 
-    call_count = internal_lock_count(frame, this);
+    call_count = int_lock->lockee_count * internal_lock_count(frame, this);
     int_lock->lk_call_count = call_count;
     int_lock->lk_expected_count = call_count;
 
     if (!call_count) {
-        gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_SUBVOLS_DOWN,
-               "All bricks are down, aborting.");
+        gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_INFO_COMMON,
+               "fd not open on any subvolumes. aborting.");
         afr_unlock_now(frame, this);
         goto out;
     }
 
-    /* Send non-blocking inodelk calls only on up children
+    /* Send non-blocking lock calls only on up children
        and where the fd has been opened */
-    for (i = 0; i < priv->child_count; i++) {
-        if (!local->child_up[i])
-            continue;
-
-        if (local->fd) {
-            STACK_WIND_COOKIE(
-                frame, afr_nonblocking_inodelk_cbk, (void *)(long)i,
-                priv->children[i], priv->children[i]->fops->finodelk,
-                int_lock->domain, local->fd, F_SETLK, &int_lock->flock, NULL);
-        } else {
-            STACK_WIND_COOKIE(
-                frame, afr_nonblocking_inodelk_cbk, (void *)(long)i,
-                priv->children[i], priv->children[i]->fops->inodelk,
-                int_lock->domain, &local->loc, F_SETLK, &int_lock->flock, NULL);
+    for (i = 0; i < int_lock->lockee_count * priv->child_count; i++) {
+        child = i % priv->child_count;
+        lockee_num = i / priv->child_count;
+        if (local->child_up[child]) {
+            afr_internal_lock_wind(frame, afr_nb_internal_lock_cbk,
+                                   (void *)(long)i, child, lockee_num,
+                                   _gf_false, _gf_false);
+            if (!--call_count)
+                break;
         }
-        if (!--call_count)
-            break;
     }
 out:
     return ret;
diff --git a/xlators/cluster/afr/src/afr-mem-types.h b/xlators/cluster/afr/src/afr-mem-types.h
index 531184b4bd7..816065fb57a 100644
--- a/xlators/cluster/afr/src/afr-mem-types.h
+++ b/xlators/cluster/afr/src/afr-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __AFR_MEM_TYPES_H__
 #define __AFR_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_afr_mem_types_ {
     gf_afr_mt_afr_fd_ctx_t = gf_common_mt_end + 1,
@@ -31,6 +31,8 @@ enum gf_afr_mem_types_ {
     gf_afr_mt_empty_brick_t,
     gf_afr_mt_child_latency_t,
     gf_afr_mt_atomic_t,
+    gf_afr_mt_lk_heal_info_t,
+    gf_afr_mt_gf_lock,
     gf_afr_mt_end
 };
 #endif
diff --git a/xlators/cluster/afr/src/afr-messages.h b/xlators/cluster/afr/src/afr-messages.h
index 696336889d3..e73fd997765 100644
--- a/xlators/cluster/afr/src/afr-messages.h
+++ b/xlators/cluster/afr/src/afr-messages.h
@@ -11,7 +11,7 @@
 #ifndef _AFR_MESSAGES_H_
 #define _AFR_MESSAGES_H_
 
-#include "glfs-message-id.h"
+#include <glusterfs/glfs-message-id.h>
 
 /* To add new message IDs, append new identifiers at the end of the list.
  *
@@ -23,25 +23,145 @@
  * glfs-message-id.h.
  */
 
-GLFS_MSGID(AFR, AFR_MSG_QUORUM_FAIL, AFR_MSG_QUORUM_MET,
-           AFR_MSG_QUORUM_OVERRIDE, AFR_MSG_INVALID_CHILD_UP, AFR_MSG_SUBVOL_UP,
-           AFR_MSG_SUBVOLS_DOWN, AFR_MSG_ENTRY_UNLOCK_FAIL, AFR_MSG_SPLIT_BRAIN,
-           AFR_MSG_OPEN_FAIL, AFR_MSG_UNLOCK_FAIL, AFR_MSG_REPLACE_BRICK_STATUS,
-           AFR_MSG_GFID_NULL, AFR_MSG_FD_CREATE_FAILED, AFR_MSG_DICT_SET_FAILED,
-           AFR_MSG_EXPUNGING_FILE_OR_DIR, AFR_MSG_MIGRATION_IN_PROGRESS,
-           AFR_MSG_CHILD_MISCONFIGURED, AFR_MSG_VOL_MISCONFIGURED,
-           AFR_MSG_BLOCKING_LKS_FAILED, AFR_MSG_INVALID_FD, AFR_MSG_LOCK_INFO,
-           AFR_MSG_LOCK_XLATOR_NOT_LOADED, AFR_MSG_FD_CTX_GET_FAILED,
-           AFR_MSG_INVALID_SUBVOL, AFR_MSG_PUMP_XLATOR_ERROR,
-           AFR_MSG_SELF_HEAL_INFO, AFR_MSG_READ_SUBVOL_ERROR,
-           AFR_MSG_DICT_GET_FAILED, AFR_MSG_INFO_COMMON,
-           AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR, AFR_MSG_LOCAL_CHILD,
-           AFR_MSG_INVALID_DATA, AFR_MSG_INVALID_ARG,
-           AFR_MSG_INDEX_DIR_GET_FAILED, AFR_MSG_FSYNC_FAILED,
-           AFR_MSG_FAVORITE_CHILD, AFR_MSG_SELF_HEAL_FAILED,
-           AFR_MSG_SPLIT_BRAIN_STATUS, AFR_MSG_ADD_BRICK_STATUS,
-           AFR_MSG_NO_CHANGELOG, AFR_MSG_TIMER_CREATE_FAIL,
-           AFR_MSG_SBRAIN_FAV_CHILD_POLICY, AFR_MSG_INODE_CTX_GET_FAILED,
-           AFR_MSG_THIN_ARB);
+GLFS_MSGID(
+    AFR, AFR_MSG_QUORUM_FAIL, AFR_MSG_QUORUM_MET, AFR_MSG_QUORUM_OVERRIDE,
+    AFR_MSG_INVALID_CHILD_UP, AFR_MSG_SUBVOL_UP, AFR_MSG_SUBVOLS_DOWN,
+    AFR_MSG_ENTRY_UNLOCK_FAIL, AFR_MSG_SPLIT_BRAIN, AFR_MSG_OPEN_FAIL,
+    AFR_MSG_UNLOCK_FAIL, AFR_MSG_REPLACE_BRICK_STATUS, AFR_MSG_GFID_NULL,
+    AFR_MSG_FD_CREATE_FAILED, AFR_MSG_DICT_SET_FAILED,
+    AFR_MSG_EXPUNGING_FILE_OR_DIR, AFR_MSG_MIGRATION_IN_PROGRESS,
+    AFR_MSG_CHILD_MISCONFIGURED, AFR_MSG_VOL_MISCONFIGURED,
+    AFR_MSG_INTERNAL_LKS_FAILED, AFR_MSG_INVALID_FD, AFR_MSG_LOCK_INFO,
+    AFR_MSG_LOCK_XLATOR_NOT_LOADED, AFR_MSG_FD_CTX_GET_FAILED,
+    AFR_MSG_INVALID_SUBVOL, AFR_MSG_PUMP_XLATOR_ERROR, AFR_MSG_SELF_HEAL_INFO,
+    AFR_MSG_READ_SUBVOL_ERROR, AFR_MSG_DICT_GET_FAILED, AFR_MSG_INFO_COMMON,
+    AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR, AFR_MSG_LOCAL_CHILD, AFR_MSG_INVALID_DATA,
+    AFR_MSG_INVALID_ARG, AFR_MSG_INDEX_DIR_GET_FAILED, AFR_MSG_FSYNC_FAILED,
+    AFR_MSG_FAVORITE_CHILD, AFR_MSG_SELF_HEAL_FAILED,
+    AFR_MSG_SPLIT_BRAIN_STATUS, AFR_MSG_ADD_BRICK_STATUS, AFR_MSG_NO_CHANGELOG,
+    AFR_MSG_TIMER_CREATE_FAIL, AFR_MSG_SBRAIN_FAV_CHILD_POLICY,
+    AFR_MSG_INODE_CTX_GET_FAILED, AFR_MSG_THIN_ARB,
+    AFR_MSG_THIN_ARB_XATTROP_FAILED, AFR_MSG_THIN_ARB_LOC_POP_FAILED,
+    AFR_MSG_GET_PEND_VAL, AFR_MSG_THIN_ARB_SKIP_SHD, AFR_MSG_UNKNOWN_SET,
+    AFR_MSG_NO_XL_ID, AFR_MSG_SELF_HEAL_INFO_START,
+    AFR_MSG_SELF_HEAL_INFO_FINISH, AFR_MSG_INCRE_COUNT,
+    AFR_MSG_ADD_TO_OUTPUT_FAILED, AFR_MSG_SET_TIME_FAILED,
+    AFR_MSG_GFID_MISMATCH_DETECTED, AFR_MSG_GFID_HEAL_MSG,
+    AFR_MSG_THIN_ARB_LOOKUP_FAILED, AFR_MSG_DICT_CREATE_FAILED,
+    AFR_MSG_NO_MAJORITY_TO_RESOLVE, AFR_MSG_TYPE_MISMATCH,
+    AFR_MSG_SIZE_POLICY_NOT_APPLICABLE, AFR_MSG_NO_CHILD_SELECTED,
+    AFR_MSG_INVALID_CHILD, AFR_MSG_RESOLVE_CONFLICTING_DATA,
+    SERROR_GETTING_SRC_BRICK, SNO_DIFF_IN_MTIME, SNO_BIGGER_FILE,
+    SALL_BRICKS_UP_TO_RESOLVE, AFR_MSG_UNLOCK_FAILED, AFR_MSG_POST_OP_FAILED,
+    AFR_MSG_TA_FRAME_CREATE_FAILED, AFR_MSG_SET_KEY_XATTROP_FAILED,
+    AFR_MSG_BLOCKING_ENTRYLKS_FAILED, AFR_MSG_FOP_FAILED,
+    AFR_MSG_CLEAN_UP_FAILED, AFR_MSG_UNABLE_TO_FETCH, AFR_MSG_XATTR_SET_FAILED,
+    AFR_MSG_SPLIT_BRAIN_REPLICA, AFR_MSG_INODE_CTX_FAILED,
+    AFR_MSG_LOOKUP_FAILED, AFR_MSG_ALL_SUBVOLS_DOWN,
+    AFR_MSG_RELEASE_LOCK_FAILED, AFR_MSG_CLEAR_TIME_SPLIT_BRAIN,
+    AFR_MSG_READ_FAILED, AFR_MSG_LAUNCH_FAILED, AFR_MSG_READ_SUBVOL_NOT_UP,
+    AFR_MSG_LK_HEAL_DOM, AFR_MSG_NEW_BRICK, AFR_MSG_SPLIT_BRAIN_SET_FAILED,
+    AFR_MSG_SPLIT_BRAIN_DETERMINE_FAILED, AFR_MSG_HEALER_SPAWN_FAILED,
+    AFR_MSG_ADD_CRAWL_EVENT_FAILED, AFR_MSG_NULL_DEREF, AFR_MSG_SET_PEND_XATTR,
+    AFR_MSG_INTERNAL_ATTR);
 
+#define AFR_MSG_DICT_GET_FAILED_STR "Dict get failed"
+#define AFR_MSG_DICT_SET_FAILED_STR "Dict set failed"
+#define AFR_MSG_HEALER_SPAWN_FAILED_STR "Healer spawn failed"
+#define AFR_MSG_ADD_CRAWL_EVENT_FAILED_STR "Adding crawl event failed"
+#define AFR_MSG_INVALID_ARG_STR "Invalid argument"
+#define AFR_MSG_INDEX_DIR_GET_FAILED_STR "unable to get index-dir on "
+#define AFR_MSG_THIN_ARB_LOOKUP_FAILED_STR "Failed lookup on file"
+#define AFR_MSG_DICT_CREATE_FAILED_STR "Failed to create dict."
+#define AFR_MSG_THIN_ARB_XATTROP_FAILED_STR "Xattrop failed."
+#define AFR_MSG_THIN_ARB_LOC_POP_FAILED_STR                                    \
+    "Failed to populate loc for thin-arbiter"
+#define AFR_MSG_GET_PEND_VAL_STR "Error getting value of pending"
+#define AFR_MSG_THIN_ARB_SKIP_SHD_STR "I am not the god shd. skipping."
+#define AFR_MSG_UNKNOWN_SET_STR "Unknown set"
+#define AFR_MSG_NO_XL_ID_STR "xl does not have id"
+#define AFR_MSG_SELF_HEAL_INFO_START_STR "starting full sweep on"
+#define AFR_MSG_SELF_HEAL_INFO_FINISH_STR "finished full sweep on"
+#define AFR_MSG_INCRE_COUNT_STR "Could not increment the counter."
+#define AFR_MSG_ADD_TO_OUTPUT_FAILED_STR "Could not add to output"
+#define AFR_MSG_SET_TIME_FAILED_STR "Could not set time"
+#define AFR_MSG_GFID_HEAL_MSG_STR "Error setting gfid-heal-msg dict"
+#define AFR_MSG_NO_MAJORITY_TO_RESOLVE_STR                                     \
+    "No majority to resolve gfid split brain"
+#define AFR_MSG_GFID_MISMATCH_DETECTED_STR "Gfid mismatch dectected"
+#define AFR_MSG_SELF_HEAL_INFO_STR "performing selfheal"
+#define AFR_MSG_TYPE_MISMATCH_STR "TYPE mismatch"
+#define AFR_MSG_SIZE_POLICY_NOT_APPLICABLE_STR                                 \
+    "Size policy is not applicable to directories."
+#define AFR_MSG_NO_CHILD_SELECTED_STR                                          \
+    "No child selected by favorite-child policy"
+#define AFR_MSG_INVALID_CHILD_STR "Invalid child"
+#define AFR_MSG_RESOLVE_CONFLICTING_DATA_STR                                   \
+    "selected as authentic to resolve conflicting data"
+#define SERROR_GETTING_SRC_BRICK_STR "Error getting the source brick"
+#define SNO_DIFF_IN_MTIME_STR "No difference in mtime"
+#define SNO_BIGGER_FILE_STR "No bigger file"
+#define SALL_BRICKS_UP_TO_RESOLVE_STR                                          \
+    "All the bricks should be up to resolve the gfid split brain"
+#define AFR_MSG_UNLOCK_FAILED_STR "Failed to unlock"
+#define AFR_MSG_POST_OP_FAILED_STR "Post-op on thin-arbiter failed"
+#define AFR_MSG_TA_FRAME_CREATE_FAILED_STR "Failed to create ta_frame"
+#define AFR_MSG_SET_KEY_XATTROP_FAILED_STR "Could not set key during xattrop"
+#define AFR_MSG_BLOCKING_ENTRYLKS_FAILED_STR "Blocking entrylks failed"
+#define AFR_MSG_FSYNC_FAILED_STR "fsync failed"
+#define AFR_MSG_QUORUM_FAIL_STR "quorum is not met"
+#define AFR_MSG_FOP_FAILED_STR "Failing Fop"
+#define AFR_MSG_INVALID_SUBVOL_STR "not a subvolume"
+#define AFR_MSG_VOL_MISCONFIGURED_STR "Volume is dangling"
+#define AFR_MSG_CHILD_MISCONFIGURED_STR                                        \
+    "replicate translator needs more than one subvolume defined"
+#define AFR_MSG_CLEAN_UP_FAILED_STR "Failed to clean up healer threads"
+#define AFR_MSG_QUORUM_OVERRIDE_STR "overriding quorum-count"
+#define AFR_MSG_UNABLE_TO_FETCH_STR                                            \
+    "Unable to fetch afr-pending-xattr option from volfile. Falling back to "  \
+    "using client translator names"
+#define AFR_MSG_NULL_DEREF_STR "possible NULL deref"
+#define AFR_MSG_XATTR_SET_FAILED_STR "Cannot set xattr cookie key"
+#define AFR_MSG_SPLIT_BRAIN_STATUS_STR "Failed to create synctask"
+#define AFR_MSG_SUBVOLS_DOWN_STR "All subvolumes are not up"
+#define AFR_MSG_SPLIT_BRAIN_CHOICE_ERROR_STR                                   \
+    "Failed to cancel split-brain choice"
+#define AFR_MSG_SPLIT_BRAIN_REPLICA_STR                                        \
+    "Cannot set replica. File is not in data/metadata split-brain"
+#define AFR_MSG_INODE_CTX_FAILED_STR "Failed to get inode_ctx"
+#define AFR_MSG_READ_SUBVOL_ERROR_STR "no read subvols"
+#define AFR_MSG_LOCAL_CHILD_STR "selecting local read-child"
+#define AFR_MSG_LOOKUP_FAILED_STR "Failed to lookup/create thin-arbiter id file"
+#define AFR_MSG_TIMER_CREATE_FAIL_STR                                          \
+    "Cannot create timer for delayed initialization"
+#define AFR_MSG_SUBVOL_UP_STR "Subvolume came back up; going online"
+#define AFR_MSG_ALL_SUBVOLS_DOWN_STR                                           \
+    "All subvolumes are down. Going offline until atleast one of them is up"
+#define AFR_MSG_RELEASE_LOCK_FAILED_STR "Failed to release lock"
+#define AFR_MSG_INVALID_CHILD_UP_STR "Received child_up from invalid subvolume"
+#define AFR_MSG_QUORUM_MET_STR "Client-quorum is met"
+#define AFR_MSG_EXPUNGING_FILE_OR_DIR_STR "expunging file or dir"
+#define AFR_MSG_SELF_HEAL_FAILED_STR "Invalid"
+#define AFR_MSG_SPLIT_BRAIN_STR "Skipping conservative mergeon the file"
+#define AFR_MSG_CLEAR_TIME_SPLIT_BRAIN_STR "clear time split brain"
+#define AFR_MSG_READ_FAILED_STR "Failing read since good brick is down"
+#define AFR_MSG_LAUNCH_FAILED_STR "Failed to launch synctask"
+#define AFR_MSG_READ_SUBVOL_NOT_UP_STR                                         \
+    "read subvolume in this generation is not up"
+#define AFR_MSG_INTERNAL_LKS_FAILED_STR                                        \
+    "Unable to work with lk-owner while attempting fop"
+#define AFR_MSG_LOCK_XLATOR_NOT_LOADED_STR                                     \
+    "subvolume does not support locking. please load features/locks xlator "   \
+    "on server."
+#define AFR_MSG_FD_CTX_GET_FAILED_STR "unable to get fd ctx"
+#define AFR_MSG_INFO_COMMON_STR "fd not open on any subvolumes, aborting."
+#define AFR_MSG_REPLACE_BRICK_STATUS_STR "Couldn't acquire lock on any child."
+#define AFR_MSG_NEW_BRICK_STR "New brick"
+#define AFR_MSG_SPLIT_BRAIN_SET_FAILED_STR                                     \
+    "Failed to set split-brain choice to -1"
+#define AFR_MSG_SPLIT_BRAIN_DETERMINE_FAILED_STR                               \
+    "Failed to determine split-brain. Aborting split-brain-choice set"
+#define AFR_MSG_OPEN_FAIL_STR "Failed to open subvolume"
+#define AFR_MSG_SET_PEND_XATTR_STR "Set of pending xattr"
+#define AFR_MSG_INTERNAL_ATTR_STR "is an internal extended attribute"
 #endif /* !_AFR_MESSAGES_H_ */
diff --git a/xlators/cluster/afr/src/afr-open.c b/xlators/cluster/afr/src/afr-open.c
index 1763dc0a4cc..64856042b65 100644
--- a/xlators/cluster/afr/src/afr-open.c
+++ b/xlators/cluster/afr/src/afr-open.c
@@ -8,32 +8,22 @@
   cases as published by the Free Software Foundation.
 */
 
-#include <libgen.h>
 #include <unistd.h>
-#include <fnmatch.h>
 #include <sys/time.h>
 #include <stdlib.h>
 #include <signal.h>
 
-#include "glusterfs.h"
+#include <glusterfs/glusterfs.h>
 #include "afr.h"
-#include "dict.h"
-#include "xlator.h"
-#include "hashfn.h"
-#include "logging.h"
-#include "list.h"
-#include "call-stub.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat-errno.h"
-#include "compat.h"
-#include "byte-order.h"
-#include "statedump.h"
-
-#include "afr-inode-read.h"
-#include "afr-inode-write.h"
-#include "afr-dir-read.h"
-#include "afr-dir-write.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/defaults.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/compat.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/statedump.h>
+
 #include "afr-transaction.h"
 
 gf_boolean_t
@@ -73,6 +63,10 @@ afr_open_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
     local = frame->local;
     fd_ctx = local->fd_ctx;
 
+    local->replies[child_index].valid = 1;
+    local->replies[child_index].op_ret = op_ret;
+    local->replies[child_index].op_errno = op_errno;
+
     LOCK(&frame->lock);
     {
         if (op_ret == -1) {
@@ -84,13 +78,16 @@ afr_open_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
             if (!local->xdata_rsp && xdata)
                 local->xdata_rsp = dict_ref(xdata);
         }
+        call_count = --local->call_count;
     }
     UNLOCK(&frame->lock);
 
-    call_count = afr_frame_return(frame);
-
     if (call_count == 0) {
-        if ((fd_ctx->flags & O_TRUNC) && (local->op_ret >= 0)) {
+        afr_handle_replies_quorum(frame, this);
+        if (local->op_ret == -1) {
+            AFR_STACK_UNWIND(open, frame, local->op_ret, local->op_errno, NULL,
+                             NULL);
+        } else if (fd_ctx->flags & O_TRUNC) {
             STACK_WIND(frame, afr_open_ftruncate_cbk, this,
                        this->fops->ftruncate, fd, 0, NULL);
         } else {
@@ -140,7 +137,7 @@ afr_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
 {
     afr_private_t *priv = NULL;
     afr_local_t *local = NULL;
-    int spb_choice = 0;
+    int spb_subvol = 0;
     int event_generation = 0;
     int ret = 0;
     int32_t op_errno = 0;
@@ -161,6 +158,11 @@ afr_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
         goto out;
     }
 
+    if (priv->quorum_count && !afr_has_quorum(local->child_up, this, NULL)) {
+        op_errno = afr_quorum_errno(priv);
+        goto out;
+    }
+
     if (!afr_is_consistent_io_possible(local, priv, &op_errno))
         goto out;
 
@@ -177,9 +179,9 @@ afr_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
     ret = afr_inode_get_readable(frame, local->inode, this, NULL,
                                  &event_generation, AFR_DATA_TRANSACTION);
     if ((ret < 0) &&
-        (afr_inode_split_brain_choice_get(local->inode, this, &spb_choice) ==
-         0) &&
-        spb_choice < 0) {
+        (afr_split_brain_read_subvol_get(local->inode, this, NULL,
+                                         &spb_subvol) == 0) &&
+        spb_subvol < 0) {
         afr_inode_refresh(frame, this, local->inode, local->inode->gfid,
                           afr_open_continue);
     } else {
@@ -213,11 +215,9 @@ afr_openfd_fix_open_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      "successfully on subvolume %s",
                      local->loc.path, priv->children[child_index]->name);
     } else {
-        gf_msg(this->name, fop_log_level(GF_FOP_OPEN, op_errno), op_errno,
-               AFR_MSG_OPEN_FAIL,
-               "Failed to open %s on "
-               "subvolume %s",
-               local->loc.path, priv->children[child_index]->name);
+        gf_smsg(this->name, fop_log_level(GF_FOP_OPEN, op_errno), op_errno,
+                AFR_MSG_OPEN_FAIL, "path=%s", local->loc.path, "subvolume=%s",
+                priv->children[child_index]->name, NULL);
     }
 
     fd_ctx = local->fd_ctx;
diff --git a/xlators/cluster/afr/src/afr-read-txn.c b/xlators/cluster/afr/src/afr-read-txn.c
index 1cd5c2eee3b..6fc2c75145c 100644
--- a/xlators/cluster/afr/src/afr-read-txn.c
+++ b/xlators/cluster/afr/src/afr-read-txn.c
@@ -30,27 +30,6 @@ afr_pending_read_decrement(afr_private_t *priv, int child_index)
     GF_ATOMIC_DEC(priv->pending_reads[child_index]);
 }
 
-static gf_boolean_t
-afr_ta_dict_contains_pending_xattr(dict_t *dict, afr_private_t *priv, int child)
-{
-    int *pending = NULL;
-    int ret = 0;
-    int i = 0;
-
-    ret = dict_get_ptr(dict, priv->pending_key[child], (void *)&pending);
-    if (ret == 0) {
-        for (i = 0; i < AFR_NUM_CHANGE_LOGS; i++) {
-            /* Not doing a ntoh32(pending) as we just want to check
-             * if it is non-zero or not. */
-            if (pending[i]) {
-                return _gf_true;
-            }
-        }
-    }
-
-    return _gf_false;
-}
-
 void
 afr_read_txn_wind(call_frame_t *frame, xlator_t *this, int subvol)
 {
@@ -114,7 +93,7 @@ afr_ta_read_txn(void *opaque)
     call_frame_t *frame = NULL;
     xlator_t *this = NULL;
     int read_subvol = -1;
-    int up_child = AFR_CHILD_UNKNOWN;
+    int query_child = AFR_CHILD_UNKNOWN;
     int possible_bad_child = AFR_CHILD_UNKNOWN;
     int ret = 0;
     int op_errno = ENOMEM;
@@ -134,18 +113,18 @@ afr_ta_read_txn(void *opaque)
     this = frame->this;
     local = frame->local;
     priv = this->private;
+    query_child = local->read_txn_query_child;
 
-    if (local->child_up[AFR_CHILD_ZERO]) {
-        up_child = AFR_CHILD_ZERO;
+    if (query_child == AFR_CHILD_ZERO) {
         possible_bad_child = AFR_CHILD_ONE;
-    } else if (local->child_up[AFR_CHILD_ONE]) {
-        up_child = AFR_CHILD_ONE;
+    } else if (query_child == AFR_CHILD_ONE) {
         possible_bad_child = AFR_CHILD_ZERO;
+    } else {
+        /*read_txn_query_child is AFR_CHILD_UNKNOWN*/
+        goto out;
     }
 
-    GF_ASSERT(up_child != AFR_CHILD_UNKNOWN);
-
-    /* Query the up_child to see if it blames the down one. */
+    /* Ask the query_child to see if it blames the possibly bad one. */
     xdata_req = dict_new();
     if (!xdata_req)
         goto out;
@@ -159,30 +138,33 @@ afr_ta_read_txn(void *opaque)
         goto out;
 
     if (local->fd) {
-        ret = syncop_fxattrop(priv->children[up_child], local->fd,
+        ret = syncop_fxattrop(priv->children[query_child], local->fd,
                               GF_XATTROP_ADD_ARRAY, xdata_req, NULL, &xdata_rsp,
                               NULL);
     } else {
-        ret = syncop_xattrop(priv->children[up_child], &local->loc,
+        ret = syncop_xattrop(priv->children[query_child], &local->loc,
                              GF_XATTROP_ADD_ARRAY, xdata_req, NULL, &xdata_rsp,
                              NULL);
     }
     if (ret || !xdata_rsp) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed xattrop for gfid %s on %s",
-               uuid_utoa(local->inode->gfid), priv->children[up_child]->name);
+               uuid_utoa(local->inode->gfid),
+               priv->children[query_child]->name);
         op_errno = -ret;
         goto out;
     }
 
     if (afr_ta_dict_contains_pending_xattr(xdata_rsp, priv,
                                            possible_bad_child)) {
-        read_subvol = up_child;
+        read_subvol = query_child;
         goto out;
     }
     dict_unref(xdata_rsp);
-    /* Query thin-arbiter to see if it blames any data brick. */
-    ret = afr_fill_ta_loc(this, &loc);
+    xdata_rsp = NULL;
+
+    /* It doesn't. So query thin-arbiter to see if it blames any data brick. */
+    ret = afr_fill_ta_loc(this, &loc, _gf_true);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to populate thin-arbiter loc for: %s.", loc.name);
@@ -211,8 +193,8 @@ afr_ta_read_txn(void *opaque)
         goto unlock;
     }
 
-    if (!afr_ta_dict_contains_pending_xattr(xdata_rsp, priv, up_child)) {
-        read_subvol = up_child;
+    if (!afr_ta_dict_contains_pending_xattr(xdata_rsp, priv, query_child)) {
+        read_subvol = query_child;
     } else {
         gf_msg(this->name, GF_LOG_ERROR, EIO, AFR_MSG_THIN_ARB,
                "Failing read for gfid %s since good brick %s is down",
@@ -290,7 +272,7 @@ afr_read_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
     int read_subvol = -1;
     inode_t *inode = NULL;
     int ret = -1;
-    int spb_choice = -1;
+    int spb_subvol = -1;
 
     local = frame->local;
     inode = local->inode;
@@ -321,9 +303,9 @@ afr_read_txn_refresh_done(call_frame_t *frame, xlator_t *this, int err)
     local->read_attempted[read_subvol] = 1;
 readfn:
     if (read_subvol == -1) {
-        ret = afr_inode_split_brain_choice_get(inode, this, &spb_choice);
-        if ((ret == 0) && spb_choice >= 0)
-            read_subvol = spb_choice;
+        ret = afr_split_brain_read_subvol_get(inode, this, frame, &spb_subvol);
+        if ((ret == 0) && spb_subvol >= 0)
+            read_subvol = spb_subvol;
     }
 
     if (read_subvol == -1) {
@@ -431,7 +413,7 @@ afr_read_txn(call_frame_t *frame, xlator_t *this, inode_t *inode,
     local->is_read_txn = _gf_true;
     local->transaction.type = type;
 
-    if (priv->quorum_count && !afr_has_quorum(local->child_up, this)) {
+    if (priv->quorum_count && !afr_has_quorum(local->child_up, this, NULL)) {
         local->op_ret = -1;
         local->op_errno = afr_quorum_errno(priv);
         goto read;
@@ -450,6 +432,11 @@ afr_read_txn(call_frame_t *frame, xlator_t *this, inode_t *inode,
 
     if (priv->thin_arbiter_count &&
         AFR_COUNT(local->child_up, priv->child_count) != priv->child_count) {
+        if (local->child_up[0]) {
+            local->read_txn_query_child = AFR_CHILD_ZERO;
+        } else if (local->child_up[1]) {
+            local->read_txn_query_child = AFR_CHILD_ONE;
+        }
         afr_ta_read_txn_synctask(frame, this);
         return 0;
     }
diff --git a/xlators/cluster/afr/src/afr-self-heal-common.c b/xlators/cluster/afr/src/afr-self-heal-common.c
index 7a96b894cd7..a580a1584cc 100644
--- a/xlators/cluster/afr/src/afr-self-heal-common.c
+++ b/xlators/cluster/afr/src/afr-self-heal-common.c
@@ -10,10 +10,10 @@
 
 #include "afr.h"
 #include "afr-self-heal.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 #include "protocol-common.h"
 #include "afr-messages.h"
-#include "events.h"
+#include <glusterfs/events.h>
 
 void
 afr_heal_synctask(xlator_t *this, afr_local_t *local);
@@ -55,7 +55,8 @@ afr_lookup_and_heal_gfid(xlator_t *this, inode_t *parent, const char *name,
     for (i = 0; i < priv->child_count; i++) {
         if (source == -1) {
             /* case (a) above. */
-            if (replies[i].valid && replies[i].op_ret == 0) {
+            if (replies[i].valid && replies[i].op_ret == 0 &&
+                replies[i].poststat.ia_type != IA_INVAL) {
                 ia_type = replies[i].poststat.ia_type;
                 break;
             }
@@ -63,7 +64,8 @@ afr_lookup_and_heal_gfid(xlator_t *this, inode_t *parent, const char *name,
             /* case (b) above. */
             if (i == source)
                 continue;
-            if (sources[i] && replies[i].valid && replies[i].op_ret == 0) {
+            if (sources[i] && replies[i].valid && replies[i].op_ret == 0 &&
+                replies[i].poststat.ia_type != IA_INVAL) {
                 ia_type = replies[i].poststat.ia_type;
                 break;
             }
@@ -77,6 +79,12 @@ heal:
     for (i = 0; i < priv->child_count; i++) {
         if (!replies[i].valid || replies[i].op_ret != 0)
             continue;
+
+        if (gf_uuid_is_null(gfid) &&
+            !gf_uuid_is_null(replies[i].poststat.ia_gfid) &&
+            replies[i].poststat.ia_type == ia_type)
+            gfid = replies[i].poststat.ia_gfid;
+
         if (!gf_uuid_is_null(replies[i].poststat.ia_gfid) ||
             replies[i].poststat.ia_type != ia_type)
             continue;
@@ -132,7 +140,7 @@ heal:
         }
     }
 out:
-    if (gfid_idx && (*gfid_idx == -1) && (ret == 0)) {
+    if (gfid_idx && (*gfid_idx == -1) && (ret == 0) && local) {
         ret = -afr_final_errno(local, priv);
     }
     loc_wipe(&loc);
@@ -167,27 +175,23 @@ afr_selfheal_gfid_mismatch_by_majority(struct afr_reply *replies,
 {
     int j = 0;
     int i = 0;
-    int src = -1;
-    int votes[child_count];
+    int votes;
 
     for (i = 0; i < child_count; i++) {
         if (!replies[i].valid || replies[i].op_ret == -1)
             continue;
 
-        votes[i] = 1;
+        votes = 1;
         for (j = i + 1; j < child_count; j++) {
             if ((!gf_uuid_compare(replies[i].poststat.ia_gfid,
                                   replies[j].poststat.ia_gfid)))
-                votes[i]++;
-            if (votes[i] > child_count / 2) {
-                src = i;
-                goto out;
-            }
+                votes++;
+            if (votes > child_count / 2)
+                return i;
         }
     }
 
-out:
-    return src;
+    return -1;
 }
 
 int
@@ -263,10 +267,8 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
                "All the bricks should be up to resolve the gfid split "
                "barin");
         if (xdata) {
-            ret = dict_set_str(xdata, "gfid-heal-msg",
-                               "All the "
-                               "bricks should be up to resolve the"
-                               " gfid split barin");
+            ret = dict_set_sizen_str_sizen(xdata, "gfid-heal-msg",
+                                           SALL_BRICKS_UP_TO_RESOLVE);
             if (ret)
                 gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_DICT_SET_FAILED,
                        "Error setting"
@@ -276,7 +278,7 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
     }
 
     if (xdata) {
-        ret = dict_get_int32(xdata, "heal-op", &heal_op);
+        ret = dict_get_int32_sizen(xdata, "heal-op", &heal_op);
         if (ret)
             goto fav_child;
     } else {
@@ -289,10 +291,10 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
                                                            priv->child_count);
             if (*src == -1) {
                 gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN,
-                       "No bigger file");
+                       SNO_BIGGER_FILE);
                 if (xdata) {
-                    ret = dict_set_str(xdata, "gfid-heal-msg",
-                                       "No bigger file");
+                    ret = dict_set_sizen_str_sizen(xdata, "gfid-heal-msg",
+                                                   SNO_BIGGER_FILE);
                     if (ret)
                         gf_msg(this->name, GF_LOG_ERROR, 0,
                                AFR_MSG_DICT_SET_FAILED,
@@ -307,10 +309,10 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
                                                             priv->child_count);
             if (*src == -1) {
                 gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN,
-                       "No difference in mtime");
+                       SNO_DIFF_IN_MTIME);
                 if (xdata) {
-                    ret = dict_set_str(xdata, "gfid-heal-msg",
-                                       "No difference in mtime");
+                    ret = dict_set_sizen_str_sizen(xdata, "gfid-heal-msg",
+                                                   SNO_DIFF_IN_MTIME);
                     if (ret)
                         gf_msg(this->name, GF_LOG_ERROR, 0,
                                AFR_MSG_DICT_SET_FAILED,
@@ -321,7 +323,7 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
             break;
 
         case GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK:
-            ret = dict_get_str(xdata, "child-name", &src_brick);
+            ret = dict_get_str_sizen(xdata, "child-name", &src_brick);
             if (ret) {
                 gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN,
                        "Error getting the source "
@@ -332,12 +334,10 @@ afr_gfid_split_brain_source(xlator_t *this, struct afr_reply *replies,
                                                          src_brick);
             if (*src == -1) {
                 gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN,
-                       "Error getting the source "
-                       "brick");
+                       SERROR_GETTING_SRC_BRICK);
                 if (xdata) {
-                    ret = dict_set_str(xdata, "gfid-heal-msg",
-                                       "Error getting the source "
-                                       "brick");
+                    ret = dict_set_sizen_str_sizen(xdata, "gfid-heal-msg",
+                                                   SERROR_GETTING_SRC_BRICK);
                     if (ret)
                         gf_msg(this->name, GF_LOG_ERROR, 0,
                                AFR_MSG_DICT_SET_FAILED,
@@ -391,11 +391,12 @@ out:
                uuid_utoa_r(replies[src_idx].poststat.ia_gfid, g2),
                priv->children[src_idx]->name);
         gf_event(EVENT_AFR_SPLIT_BRAIN,
+                 "client-pid=%d;"
                  "subvol=%s;type=gfid;file="
                  "<gfid:%s>/%s>;count=2;child-%d=%s;gfid-%d=%s;"
                  "child-%d=%s;gfid-%d=%s",
-                 this->name, uuid_utoa(pargfid), bname, child_idx,
-                 priv->children[child_idx]->name, child_idx,
+                 this->ctx->cmd_args.client_pid, this->name, uuid_utoa(pargfid),
+                 bname, child_idx, priv->children[child_idx]->name, child_idx,
                  uuid_utoa_r(replies[child_idx].poststat.ia_gfid, g1), src_idx,
                  priv->children[src_idx]->name, src_idx,
                  uuid_utoa_r(replies[src_idx].poststat.ia_gfid, g2));
@@ -512,7 +513,8 @@ afr_selfheal_restore_time(call_frame_t *frame, xlator_t *this, inode_t *inode,
 
     AFR_ONLIST(healed_sinks, frame, afr_sh_generic_fop_cbk, setattr, &loc,
                &replies[source].poststat,
-               (GF_SET_ATTR_ATIME | GF_SET_ATTR_MTIME), NULL);
+               (GF_SET_ATTR_ATIME | GF_SET_ATTR_MTIME | GF_SET_ATTR_CTIME),
+               NULL);
 
     loc_wipe(&loc);
 
@@ -909,7 +911,7 @@ afr_dict_contains_heal_op(call_frame_t *frame)
 
     local = frame->local;
     xdata_req = local->xdata_req;
-    ret = dict_get_int32(xdata_req, "heal-op", &heal_op);
+    ret = dict_get_int32_sizen(xdata_req, "heal-op", &heal_op);
     if (ret)
         return _gf_false;
     if (local->xdata_rsp == NULL) {
@@ -917,8 +919,8 @@ afr_dict_contains_heal_op(call_frame_t *frame)
         if (!local->xdata_rsp)
             return _gf_true;
     }
-    ret = dict_set_str(local->xdata_rsp, "sh-fail-msg",
-                       "File not in split-brain");
+    ret = dict_set_sizen_str_sizen(local->xdata_rsp, "sh-fail-msg",
+                                   SFILE_NOT_IN_SPLIT_BRAIN);
 
     return _gf_true;
 }
@@ -972,7 +974,8 @@ afr_mark_split_brain_source_sinks_by_heal_op(
     xdata_rsp = local->xdata_rsp;
 
     if (!afr_can_decide_split_brain_source_sinks(replies, priv->child_count)) {
-        ret = dict_set_str(xdata_rsp, "sh-fail-msg", SBRAIN_HEAL_NO_GO_MSG);
+        ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                       SBRAIN_HEAL_NO_GO_MSG);
         ret = -1;
         goto out;
     }
@@ -983,16 +986,16 @@ afr_mark_split_brain_source_sinks_by_heal_op(
     switch (heal_op) {
         case GF_SHD_OP_SBRAIN_HEAL_FROM_BIGGER_FILE:
             if (type == AFR_METADATA_TRANSACTION) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg",
-                                   "Use source-brick option to"
-                                   " heal metadata split-brain");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SUSE_SOURCE_BRICK_TO_HEAL);
                 if (!ret)
                     ret = -1;
                 goto out;
             }
             afr_mark_largest_file_as_source(this, sources, replies);
             if (AFR_COUNT(sources, priv->child_count) != 1) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg", "No bigger file");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SNO_BIGGER_FILE);
                 if (!ret)
                     ret = -1;
                 goto out;
@@ -1000,36 +1003,36 @@ afr_mark_split_brain_source_sinks_by_heal_op(
             break;
         case GF_SHD_OP_SBRAIN_HEAL_FROM_LATEST_MTIME:
             if (type == AFR_METADATA_TRANSACTION) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg",
-                                   "Use source-brick option to"
-                                   " heal metadata split-brain");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SUSE_SOURCE_BRICK_TO_HEAL);
                 if (!ret)
                     ret = -1;
                 goto out;
             }
             afr_mark_latest_mtime_file_as_source(this, sources, replies);
             if (AFR_COUNT(sources, priv->child_count) != 1) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg",
-                                   "No difference in mtime");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SNO_DIFF_IN_MTIME);
                 if (!ret)
                     ret = -1;
                 goto out;
             }
             break;
         case GF_SHD_OP_SBRAIN_HEAL_FROM_BRICK:
-            ret = dict_get_str(xdata_req, "child-name", &name);
+            ret = dict_get_str_sizen(xdata_req, "child-name", &name);
             if (ret)
                 goto out;
             source = afr_get_child_index_from_name(this, name);
             if (source < 0) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg",
-                                   "Invalid brick name");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SINVALID_BRICK_NAME);
                 if (!ret)
                     ret = -1;
                 goto out;
             }
             if (locked_on[source] != 1) {
-                ret = dict_set_str(xdata_rsp, "sh-fail-msg", "Brick is not up");
+                ret = dict_set_sizen_str_sizen(xdata_rsp, "sh-fail-msg",
+                                               SBRICK_IS_NOT_UP);
                 if (!ret)
                     ret = -1;
                 goto out;
@@ -1170,7 +1173,8 @@ afr_sh_fav_by_ctime(xlator_t *this, struct afr_reply *replies, inode_t *inode)
 }
 
 /*
- * afr_sh_fav_by_size: Choose favorite child by size.
+ * afr_sh_fav_by_size: Choose favorite child by size
+ * when not all files are of zero size.
  */
 int
 afr_sh_fav_by_size(xlator_t *this, struct afr_reply *replies, inode_t *inode)
@@ -1181,18 +1185,31 @@ afr_sh_fav_by_size(xlator_t *this, struct afr_reply *replies, inode_t *inode)
     uint64_t cmp_sz = 0;
 
     priv = this->private;
-
     for (i = 0; i < priv->child_count; i++) {
-        if (replies[i].valid == 1) {
-            gf_msg_debug(this->name, 0,
-                         "Child:%s file size = %" PRIu64 " for gfid %s",
-                         priv->children[i]->name, replies[i].poststat.ia_size,
-                         uuid_utoa(inode->gfid));
-            if (replies[i].poststat.ia_size > cmp_sz) {
-                cmp_sz = replies[i].poststat.ia_size;
-                fav_child = i;
-            }
+        if (!replies[i].valid) {
+            continue;
+        }
+        gf_msg_debug(this->name, 0,
+                     "Child:%s file size = %" PRIu64 " for gfid %s",
+                     priv->children[i]->name, replies[i].poststat.ia_size,
+                     uuid_utoa(inode->gfid));
+        if (replies[i].poststat.ia_type == IA_IFDIR) {
+            gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SBRAIN_FAV_CHILD_POLICY,
+                   "Cannot perform selfheal on %s. "
+                   "Size policy is not applicable to directories.",
+                   uuid_utoa(inode->gfid));
+            break;
         }
+        if (replies[i].poststat.ia_size > cmp_sz) {
+            cmp_sz = replies[i].poststat.ia_size;
+            fav_child = i;
+        } else if (replies[i].poststat.ia_size == cmp_sz) {
+            fav_child = -1;
+        }
+    }
+    if (fav_child == -1) {
+        gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SPLIT_BRAIN,
+               "No bigger file");
     }
     return fav_child;
 }
@@ -1259,7 +1276,10 @@ afr_mark_split_brain_source_sinks_by_policy(
     priv = this->private;
 
     fav_child = afr_sh_get_fav_by_policy(this, replies, inode, &policy_str);
-    if (fav_child > priv->child_count - 1) {
+    if (fav_child == -1) {
+        gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SBRAIN_FAV_CHILD_POLICY,
+               "No child selected by favorite-child policy.");
+    } else if (fav_child > priv->child_count - 1) {
         gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SBRAIN_FAV_CHILD_POLICY,
                "Invalid child (%d) "
                "selected by policy %s.",
@@ -1390,7 +1410,7 @@ afr_mark_split_brain_source_sinks(
     if (source >= 0)
         return source;
 
-    ret = dict_get_int32(xdata_req, "heal-op", &heal_op);
+    ret = dict_get_int32_sizen(xdata_req, "heal-op", &heal_op);
     if (ret)
         goto autoheal;
 
@@ -1406,7 +1426,7 @@ autoheal:
             frame, this, inode, sources, sinks, healed_sinks, locked_on,
             replies, type);
         if (source != -1) {
-            ret = dict_set_int32(xdata_req, "fav-child-policy", 1);
+            ret = dict_set_int32_sizen(xdata_req, "fav-child-policy", 1);
             if (ret)
                 return -1;
         }
@@ -1437,7 +1457,7 @@ _afr_fav_child_reset_sink_xattrs(call_frame_t *frame, xlator_t *this,
     priv = this->private;
     local = frame->local;
 
-    if (!dict_get(local->xdata_req, "fav-child-policy"))
+    if (!dict_get_sizen(local->xdata_req, "fav-child-policy"))
         return 0;
 
     xdata = dict_new();
@@ -1555,7 +1575,6 @@ afr_selfheal_find_direction(call_frame_t *frame, xlator_t *this,
     char *accused = NULL;      /* Accused others without any self-accusal */
     char *pending = NULL;      /* Have pending operations on others */
     char *self_accused = NULL; /* Accused itself */
-    int min_participants = -1;
 
     priv = this->private;
 
@@ -1579,12 +1598,7 @@ afr_selfheal_find_direction(call_frame_t *frame, xlator_t *this,
         }
     }
 
-    if (type == AFR_DATA_TRANSACTION) {
-        min_participants = priv->child_count;
-    } else {
-        min_participants = AFR_SH_MIN_PARTICIPANTS;
-    }
-    if (afr_success_count(replies, priv->child_count) < min_participants) {
+    if (afr_success_count(replies, priv->child_count) < priv->child_count) {
         /* Treat this just like locks not being acquired */
         return -ENOTCONN;
     }
@@ -1645,7 +1659,7 @@ afr_selfheal_find_direction(call_frame_t *frame, xlator_t *this,
         }
     }
 
-    if (type == AFR_DATA_TRANSACTION)
+    if (type == AFR_DATA_TRANSACTION || type == AFR_METADATA_TRANSACTION)
         afr_selfheal_post_op_failure_accounting(priv, accused, sources,
                                                 locked_on);
 
@@ -1753,11 +1767,9 @@ afr_selfheal_discover_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     if (xdata) {
         local->replies[i].xdata = dict_ref(xdata);
         ret = dict_get_int8(xdata, "link-count", &need_heal);
-        local->replies[i].need_heal = need_heal;
-    } else {
-        local->replies[i].need_heal = need_heal;
     }
 
+    local->replies[i].need_heal = need_heal;
     syncbarrier_wake(&local->barrier);
 
     return 0;
@@ -1813,10 +1825,41 @@ afr_selfheal_unlocked_lookup_on(call_frame_t *frame, inode_t *parent,
     return inode;
 }
 
+static int
+afr_set_multi_dom_lock_count_request(xlator_t *this, dict_t *dict)
+{
+    int ret = 0;
+    afr_private_t *priv = NULL;
+    char *key1 = NULL;
+    char *key2 = NULL;
+
+    priv = this->private;
+    key1 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+                   strlen(this->name));
+    key2 = alloca0(strlen(GLUSTERFS_INODELK_DOM_PREFIX) + 2 +
+                   strlen(priv->sh_domain));
+
+    ret = dict_set_uint32(dict, GLUSTERFS_MULTIPLE_DOM_LK_CNT_REQUESTS, 1);
+    if (ret)
+        return ret;
+
+    sprintf(key1, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, this->name);
+    ret = dict_set_uint32(dict, key1, 1);
+    if (ret)
+        return ret;
+
+    sprintf(key2, "%s:%s", GLUSTERFS_INODELK_DOM_PREFIX, priv->sh_domain);
+    ret = dict_set_uint32(dict, key2, 1);
+    if (ret)
+        return ret;
+
+    return 0;
+}
+
 int
 afr_selfheal_unlocked_discover_on(call_frame_t *frame, inode_t *inode,
                                   uuid_t gfid, struct afr_reply *replies,
-                                  unsigned char *discover_on)
+                                  unsigned char *discover_on, dict_t *dict)
 {
     loc_t loc = {
         0,
@@ -1831,12 +1874,19 @@ afr_selfheal_unlocked_discover_on(call_frame_t *frame, inode_t *inode,
     xattr_req = dict_new();
     if (!xattr_req)
         return -ENOMEM;
+    if (dict)
+        dict_copy(dict, xattr_req);
 
     if (afr_xattr_req_prepare(frame->this, xattr_req) != 0) {
         dict_unref(xattr_req);
         return -ENOMEM;
     }
 
+    if (afr_set_multi_dom_lock_count_request(frame->this, xattr_req)) {
+        dict_unref(xattr_req);
+        return -1;
+    }
+
     loc.inode = inode_ref(inode);
     gf_uuid_copy(loc.gfid, gfid);
 
@@ -1855,12 +1905,16 @@ int
 afr_selfheal_unlocked_discover(call_frame_t *frame, inode_t *inode, uuid_t gfid,
                                struct afr_reply *replies)
 {
-    afr_private_t *priv = NULL;
+    afr_local_t *local = NULL;
+    dict_t *dict = NULL;
 
-    priv = frame->this->private;
+    local = frame->local;
+
+    if (local->xattr_req)
+        dict = local->xattr_req;
 
     return afr_selfheal_unlocked_discover_on(frame, inode, gfid, replies,
-                                             priv->child_up);
+                                             local->child_up, dict);
 }
 
 unsigned int
@@ -2227,7 +2281,8 @@ int
 afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
                               inode_t **link_inode, gf_boolean_t *data_selfheal,
                               gf_boolean_t *metadata_selfheal,
-                              gf_boolean_t *entry_selfheal)
+                              gf_boolean_t *entry_selfheal,
+                              struct afr_reply *replies_dst)
 {
     afr_private_t *priv = NULL;
     inode_t *inode = NULL;
@@ -2286,11 +2341,13 @@ afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
                    priv->children[i]->name,
                    uuid_utoa(replies[i].poststat.ia_gfid));
             gf_event(EVENT_AFR_SPLIT_BRAIN,
+                     "client-pid=%d;"
                      "subvol=%s;"
                      "type=file;gfid=%s;"
                      "ia_type-%d=%s;ia_type-%d=%s",
-                     this->name, uuid_utoa(replies[i].poststat.ia_gfid),
-                     first_idx, gf_inode_type_to_str(first.ia_type), i,
+                     this->ctx->cmd_args.client_pid, this->name,
+                     uuid_utoa(replies[i].poststat.ia_gfid), first_idx,
+                     gf_inode_type_to_str(first.ia_type), i,
                      gf_inode_type_to_str(replies[i].poststat.ia_type));
             ret = -EIO;
             goto out;
@@ -2361,6 +2418,8 @@ afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
 
     ret = 0;
 out:
+    if (replies && replies_dst)
+        afr_replies_copy(replies_dst, replies, priv->child_count);
     if (inode)
         inode_unref(inode);
     if (replies)
@@ -2400,8 +2459,11 @@ afr_frame_create(xlator_t *this, int32_t *op_errno)
     pid_t pid = GF_CLIENT_PID_SELF_HEALD;
 
     frame = create_frame(this, this->ctx->pool);
-    if (!frame)
+    if (!frame) {
+        if (op_errno)
+            *op_errno = ENOMEM;
         return NULL;
+    }
 
     local = AFR_FRAME_INIT(frame, (*op_errno));
     if (!local) {
@@ -2472,17 +2534,12 @@ afr_selfheal_do(call_frame_t *frame, xlator_t *this, uuid_t gfid)
     gf_boolean_t metadata_selfheal = _gf_false;
     gf_boolean_t entry_selfheal = _gf_false;
     afr_private_t *priv = NULL;
-    gf_boolean_t dataheal_enabled = _gf_false;
 
     priv = this->private;
 
-    ret = gf_string2boolean(priv->data_self_heal, &dataheal_enabled);
-    if (ret)
-        goto out;
-
     ret = afr_selfheal_unlocked_inspect(frame, this, gfid, &inode,
                                         &data_selfheal, &metadata_selfheal,
-                                        &entry_selfheal);
+                                        &entry_selfheal, NULL);
     if (ret)
         goto out;
 
@@ -2499,7 +2556,7 @@ afr_selfheal_do(call_frame_t *frame, xlator_t *this, uuid_t gfid)
         }
     }
 
-    if (data_selfheal && dataheal_enabled)
+    if (data_selfheal && priv->data_self_heal)
         data_ret = afr_selfheal_data(frame, this, fd);
 
     if (metadata_selfheal && priv->metadata_self_heal)
@@ -2693,3 +2750,185 @@ afr_choose_source_by_policy(afr_private_t *priv, unsigned char *sources,
 out:
     return source;
 }
+
+static int
+afr_anon_inode_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                         int32_t op_ret, int32_t op_errno, inode_t *inode,
+                         struct iatt *buf, struct iatt *preparent,
+                         struct iatt *postparent, dict_t *xdata)
+{
+    afr_local_t *local = frame->local;
+    int i = (long)cookie;
+
+    local->replies[i].valid = 1;
+    local->replies[i].op_ret = op_ret;
+    local->replies[i].op_errno = op_errno;
+    if (op_ret == 0) {
+        local->op_ret = 0;
+        local->replies[i].poststat = *buf;
+        local->replies[i].preparent = *preparent;
+        local->replies[i].postparent = *postparent;
+    }
+    if (xdata) {
+        local->replies[i].xdata = dict_ref(xdata);
+    }
+
+    syncbarrier_wake(&local->barrier);
+    return 0;
+}
+
+int
+afr_anon_inode_create(xlator_t *this, int child, inode_t **linked_inode)
+{
+    call_frame_t *frame = NULL;
+    afr_local_t *local = NULL;
+    afr_private_t *priv = this->private;
+    unsigned char *mkdir_on = alloca0(priv->child_count);
+    unsigned char *lookup_on = alloca0(priv->child_count);
+    loc_t loc = {0};
+    int32_t op_errno = 0;
+    int32_t child_op_errno = 0;
+    struct iatt iatt = {0};
+    dict_t *xdata = NULL;
+    uuid_t anon_inode_gfid = {0};
+    int mkdir_count = 0;
+    int i = 0;
+
+    /*Try to mkdir everywhere and return success if the dir exists on 'child'
+     */
+
+    if (!priv->use_anon_inode) {
+        op_errno = EINVAL;
+        goto out;
+    }
+
+    frame = afr_frame_create(this, &op_errno);
+    if (op_errno) {
+        goto out;
+    }
+    local = frame->local;
+    if (!local->child_up[child]) {
+        /*Other bricks may need mkdir so don't error out yet*/
+        child_op_errno = ENOTCONN;
+    }
+    gf_uuid_parse(priv->anon_gfid_str, anon_inode_gfid);
+    for (i = 0; i < priv->child_count; i++) {
+        if (!local->child_up[i])
+            continue;
+
+        if (priv->anon_inode[i]) {
+            mkdir_on[i] = 0;
+        } else {
+            mkdir_on[i] = 1;
+            mkdir_count++;
+        }
+    }
+
+    if (mkdir_count == 0) {
+        *linked_inode = inode_find(this->itable, anon_inode_gfid);
+        if (*linked_inode) {
+            op_errno = 0;
+            goto out;
+        }
+    }
+
+    loc.parent = inode_ref(this->itable->root);
+    loc.name = priv->anon_inode_name;
+    loc.inode = inode_new(this->itable);
+    if (!loc.inode) {
+        op_errno = ENOMEM;
+        goto out;
+    }
+
+    xdata = dict_new();
+    if (!xdata) {
+        op_errno = ENOMEM;
+        goto out;
+    }
+
+    op_errno = -dict_set_gfuuid(xdata, "gfid-req", anon_inode_gfid, _gf_true);
+    if (op_errno) {
+        goto out;
+    }
+
+    if (mkdir_count == 0) {
+        memcpy(lookup_on, local->child_up, priv->child_count);
+        goto lookup;
+    }
+
+    AFR_ONLIST(mkdir_on, frame, afr_anon_inode_mkdir_cbk, mkdir, &loc, 0755, 0,
+               xdata);
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (!mkdir_on[i]) {
+            continue;
+        }
+
+        if (local->replies[i].op_ret == 0) {
+            priv->anon_inode[i] = 1;
+            iatt = local->replies[i].poststat;
+        } else if (local->replies[i].op_ret < 0 &&
+                   local->replies[i].op_errno == EEXIST) {
+            lookup_on[i] = 1;
+        } else if (i == child) {
+            child_op_errno = local->replies[i].op_errno;
+        }
+    }
+
+    if (AFR_COUNT(lookup_on, priv->child_count) == 0) {
+        goto link;
+    }
+
+lookup:
+    AFR_ONLIST(lookup_on, frame, afr_selfheal_discover_cbk, lookup, &loc,
+               xdata);
+    for (i = 0; i < priv->child_count; i++) {
+        if (!lookup_on[i]) {
+            continue;
+        }
+
+        if (local->replies[i].op_ret == 0) {
+            if (gf_uuid_compare(anon_inode_gfid,
+                                local->replies[i].poststat.ia_gfid) == 0) {
+                priv->anon_inode[i] = 1;
+                iatt = local->replies[i].poststat;
+            } else {
+                if (i == child)
+                    child_op_errno = EINVAL;
+                gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_DATA,
+                       "%s has gfid: %s", priv->anon_inode_name,
+                       uuid_utoa(local->replies[i].poststat.ia_gfid));
+            }
+        } else if (i == child) {
+            child_op_errno = local->replies[i].op_errno;
+        }
+    }
+link:
+    if (!gf_uuid_is_null(iatt.ia_gfid)) {
+        *linked_inode = inode_link(loc.inode, loc.parent, loc.name, &iatt);
+        if (*linked_inode) {
+            op_errno = 0;
+            inode_lookup(*linked_inode);
+        } else {
+            op_errno = ENOMEM;
+        }
+        goto out;
+    }
+
+out:
+    if (xdata)
+        dict_unref(xdata);
+    loc_wipe(&loc);
+    /*child_op_errno takes precedence*/
+    if (child_op_errno == 0) {
+        child_op_errno = op_errno;
+    }
+
+    if (child_op_errno && *linked_inode) {
+        inode_unref(*linked_inode);
+        *linked_inode = NULL;
+    }
+    if (frame)
+        AFR_STACK_DESTROY(frame);
+    return -child_op_errno;
+}
diff --git a/xlators/cluster/afr/src/afr-self-heal-data.c b/xlators/cluster/afr/src/afr-self-heal-data.c
index 8cd79c73aa4..37bcc2b3f9e 100644
--- a/xlators/cluster/afr/src/afr-self-heal-data.c
+++ b/xlators/cluster/afr/src/afr-self-heal-data.c
@@ -10,15 +10,10 @@
 
 #include "afr.h"
 #include "afr-self-heal.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 #include "protocol-common.h"
 #include "afr-messages.h"
-#include "events.h"
-
-enum {
-    AFR_SELFHEAL_DATA_FULL = 0,
-    AFR_SELFHEAL_DATA_DIFF,
-};
+#include <glusterfs/events.h>
 
 #define HAS_HOLES(i) ((i->ia_blocks * 512) < (i->ia_size))
 static int
@@ -73,7 +68,7 @@ __afr_can_skip_data_block_heal(call_frame_t *frame, xlator_t *this, fd_t *fd,
     xdata = dict_new();
     if (!xdata)
         goto out;
-    if (dict_set_int32(xdata, "check-zero-filled", 1)) {
+    if (dict_set_int32_sizen(xdata, "check-zero-filled", 1)) {
         dict_unref(xdata);
         goto out;
     }
@@ -230,24 +225,40 @@ __afr_selfheal_data_read_write(call_frame_t *frame, xlator_t *this, fd_t *fd,
     return ret;
 }
 
+static gf_boolean_t
+afr_source_sinks_locked(xlator_t *this, unsigned char *locked_on, int source,
+                        unsigned char *healed_sinks)
+{
+    afr_private_t *priv = this->private;
+    int i = 0;
+
+    if (!locked_on[source])
+        return _gf_false;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (healed_sinks[i] && locked_on[i])
+            return _gf_true;
+    }
+
+    return _gf_false;
+}
+
 static int
 afr_selfheal_data_block(call_frame_t *frame, xlator_t *this, fd_t *fd,
                         int source, unsigned char *healed_sinks, off_t offset,
                         size_t size, int type, struct afr_reply *replies)
 {
     int ret = -1;
-    int sink_count = 0;
     afr_private_t *priv = NULL;
     unsigned char *data_lock = NULL;
 
     priv = this->private;
-    sink_count = AFR_COUNT(healed_sinks, priv->child_count);
     data_lock = alloca0(priv->child_count);
 
     ret = afr_selfheal_inodelk(frame, this, fd->inode, this->name, offset, size,
                                data_lock);
     {
-        if (ret < sink_count) {
+        if (!afr_source_sinks_locked(this, data_lock, source, healed_sinks)) {
             ret = -ENOTCONN;
             goto unlock;
         }
@@ -301,7 +312,7 @@ afr_data_self_heal_type_get(afr_private_t *priv, unsigned char *healed_sinks,
     int type = AFR_SELFHEAL_DATA_FULL;
     int i = 0;
 
-    if (priv->data_self_heal_algorithm == NULL) {
+    if (priv->data_self_heal_algorithm == AFR_SELFHEAL_DATA_DYNAMIC) {
         type = AFR_SELFHEAL_DATA_FULL;
         for (i = 0; i < priv->child_count; i++) {
             if (!healed_sinks[i] && i != source)
@@ -311,10 +322,8 @@ afr_data_self_heal_type_get(afr_private_t *priv, unsigned char *healed_sinks,
                 break;
             }
         }
-    } else if (strcmp(priv->data_self_heal_algorithm, "full") == 0) {
-        type = AFR_SELFHEAL_DATA_FULL;
-    } else if (strcmp(priv->data_self_heal_algorithm, "diff") == 0) {
-        type = AFR_SELFHEAL_DATA_DIFF;
+    } else {
+        type = priv->data_self_heal_algorithm;
     }
     return type;
 }
@@ -331,6 +340,9 @@ afr_selfheal_data_do(call_frame_t *frame, xlator_t *this, fd_t *fd, int source,
     call_frame_t *iter_frame = NULL;
     unsigned char arbiter_sink_status = 0;
 
+    gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_SELF_HEAL_INFO,
+           "performing data selfheal on %s", uuid_utoa(fd->inode->gfid));
+
     priv = this->private;
     if (priv->arbiter_count) {
         arbiter_sink_status = healed_sinks[ARBITER_BRICK_INDEX];
@@ -382,17 +394,18 @@ __afr_selfheal_truncate_sinks(call_frame_t *frame, xlator_t *this, fd_t *fd,
 {
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
-    unsigned char arbiter_sink_status = 0;
     int i = 0;
 
     local = frame->local;
     priv = this->private;
 
-    if (priv->arbiter_count) {
-        arbiter_sink_status = healed_sinks[ARBITER_BRICK_INDEX];
-        healed_sinks[ARBITER_BRICK_INDEX] = 0;
-    }
-
+    /* This will send truncate on the arbiter brick as well if it is marked as
+     * sink. If changelog is enabled on the volume it captures truncate as a
+     * data transactions on the arbiter brick. This will help geo-rep to
+     * properly sync the data from master to slave if arbiter is the ACTIVE
+     * brick during syncing and which had got some entries healed for data as
+     * part of self heal.
+     */
     AFR_ONLIST(healed_sinks, frame, afr_sh_generic_fop_cbk, ftruncate, fd, size,
                NULL);
 
@@ -403,8 +416,6 @@ __afr_selfheal_truncate_sinks(call_frame_t *frame, xlator_t *this, fd_t *fd,
             */
             healed_sinks[i] = 0;
 
-    if (arbiter_sink_status)
-        healed_sinks[ARBITER_BRICK_INDEX] = arbiter_sink_status;
     return 0;
 }
 
@@ -545,9 +556,11 @@ __afr_selfheal_data_finalize_source(
             replies, AFR_DATA_TRANSACTION);
         if (source < 0) {
             gf_event(EVENT_AFR_SPLIT_BRAIN,
+                     "client-pid=%d;"
                      "subvol=%s;type=data;"
                      "file=%s",
-                     this->name, uuid_utoa(inode->gfid));
+                     this->ctx->cmd_args.client_pid, this->name,
+                     uuid_utoa(inode->gfid));
             return -EIO;
         }
 
@@ -700,19 +713,18 @@ __afr_selfheal_data(call_frame_t *frame, xlator_t *this, fd_t *fd,
             goto unlock;
         }
 
-        if (priv->arbiter_count &&
-            AFR_COUNT(healed_sinks, priv->child_count) == 1 &&
-            healed_sinks[ARBITER_BRICK_INDEX]) {
-            is_arbiter_the_only_sink = _gf_true;
-            goto restore_time;
-        }
-
         ret = __afr_selfheal_truncate_sinks(
             frame, this, fd, healed_sinks,
             locked_replies[source].poststat.ia_size);
         if (ret < 0)
             goto unlock;
 
+        if (priv->arbiter_count &&
+            AFR_COUNT(healed_sinks, priv->child_count) == 1 &&
+            healed_sinks[ARBITER_BRICK_INDEX]) {
+            is_arbiter_the_only_sink = _gf_true;
+            goto restore_time;
+        }
         ret = 0;
     }
 unlock:
diff --git a/xlators/cluster/afr/src/afr-self-heal-entry.c b/xlators/cluster/afr/src/afr-self-heal-entry.c
index fb6952c10fd..64893f441e3 100644
--- a/xlators/cluster/afr/src/afr-self-heal-entry.c
+++ b/xlators/cluster/afr/src/afr-self-heal-entry.c
@@ -10,60 +10,176 @@
 
 #include "afr.h"
 #include "afr-self-heal.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 #include "afr-transaction.h"
 #include "afr-messages.h"
-#include "syncop-utils.h"
-#include "events.h"
+#include <glusterfs/syncop-utils.h>
+#include <glusterfs/events.h>
 
-static int
-afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
-                          inode_t *inode, int child, struct afr_reply *replies)
+int
+afr_selfheal_entry_anon_inode(xlator_t *this, inode_t *dir, const char *name,
+                              inode_t *inode, int child,
+                              struct afr_reply *replies,
+                              gf_boolean_t *anon_inode)
 {
     afr_private_t *priv = NULL;
+    afr_local_t *local = NULL;
     xlator_t *subvol = NULL;
     int ret = 0;
+    int i = 0;
+    char g[64] = {0};
+    unsigned char *lookup_success = NULL;
+    call_frame_t *frame = NULL;
+    loc_t loc2 = {
+        0,
+    };
     loc_t loc = {
         0,
     };
-    char g[64];
 
     priv = this->private;
-
     subvol = priv->children[child];
+    lookup_success = alloca0(priv->child_count);
+    uuid_utoa_r(replies[child].poststat.ia_gfid, g);
+    loc.inode = inode_new(inode->table);
+    if (!loc.inode) {
+        ret = -ENOMEM;
+        goto out;
+    }
+
+    if (replies[child].poststat.ia_type == IA_IFDIR) {
+        /* This directory may have sub-directory hierarchy which may need to
+         * be preserved for subsequent heals. So unconditionally move the
+         * directory to anonymous-inode directory*/
+        *anon_inode = _gf_true;
+        goto anon_inode;
+    }
+
+    frame = afr_frame_create(this, &ret);
+    if (!frame) {
+        ret = -ret;
+        goto out;
+    }
+    local = frame->local;
+    gf_uuid_copy(loc.gfid, replies[child].poststat.ia_gfid);
+    AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup, &loc,
+               NULL);
+    for (i = 0; i < priv->child_count; i++) {
+        if (local->replies[i].op_ret == 0) {
+            lookup_success[i] = 1;
+        } else if (local->replies[i].op_errno != ENOENT &&
+                   local->replies[i].op_errno != ESTALE) {
+            ret = -local->replies[i].op_errno;
+        }
+    }
+
+    if (priv->quorum_count) {
+        if (afr_has_quorum(lookup_success, this, NULL)) {
+            *anon_inode = _gf_true;
+        }
+    } else if (AFR_COUNT(lookup_success, priv->child_count) > 1) {
+        *anon_inode = _gf_true;
+    } else if (ret) {
+        goto out;
+    }
+
+anon_inode:
+    if (!*anon_inode) {
+        ret = 0;
+        goto out;
+    }
 
     loc.parent = inode_ref(dir);
     gf_uuid_copy(loc.pargfid, dir->gfid);
     loc.name = name;
-    loc.inode = inode_ref(inode);
 
-    if (replies[child].valid && replies[child].op_ret == 0) {
-        switch (replies[child].poststat.ia_type) {
-            case IA_IFDIR:
-                gf_msg(this->name, GF_LOG_WARNING, 0,
-                       AFR_MSG_EXPUNGING_FILE_OR_DIR,
-                       "expunging dir %s/%s (%s) on %s", uuid_utoa(dir->gfid),
-                       name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
-                       subvol->name);
-                ret = syncop_rmdir(subvol, &loc, 1, NULL, NULL);
-                break;
-            default:
-                gf_msg(this->name, GF_LOG_WARNING, 0,
-                       AFR_MSG_EXPUNGING_FILE_OR_DIR,
-                       "expunging file %s/%s (%s) on %s", uuid_utoa(dir->gfid),
-                       name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
-                       subvol->name);
-                ret = syncop_unlink(subvol, &loc, NULL, NULL);
-                break;
-        }
+    ret = afr_anon_inode_create(this, child, &loc2.parent);
+    if (ret < 0)
+        goto out;
+
+    loc2.name = g;
+    ret = syncop_rename(subvol, &loc, &loc2, NULL, NULL);
+    if (ret < 0) {
+        gf_msg(this->name, GF_LOG_WARNING, -ret, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+               "Rename to %s dir %s/%s (%s) on %s failed",
+               priv->anon_inode_name, uuid_utoa(dir->gfid), name, g,
+               subvol->name);
+    } else {
+        gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+               "Rename to %s dir %s/%s (%s) on %s successful",
+               priv->anon_inode_name, uuid_utoa(dir->gfid), name, g,
+               subvol->name);
     }
 
+out:
     loc_wipe(&loc);
+    loc_wipe(&loc2);
+    if (frame) {
+        AFR_STACK_DESTROY(frame);
+    }
 
     return ret;
 }
 
 int
+afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
+                          inode_t *inode, int child, struct afr_reply *replies)
+{
+    char g[64] = {0};
+    afr_private_t *priv = NULL;
+    xlator_t *subvol = NULL;
+    int ret = 0;
+    loc_t loc = {
+        0,
+    };
+    gf_boolean_t anon_inode = _gf_false;
+
+    priv = this->private;
+    subvol = priv->children[child];
+
+    if ((!replies[child].valid) || (replies[child].op_ret < 0)) {
+        /*Nothing to do*/
+        ret = 0;
+        goto out;
+    }
+
+    if (priv->use_anon_inode) {
+        ret = afr_selfheal_entry_anon_inode(this, dir, name, inode, child,
+                                            replies, &anon_inode);
+        if (ret < 0 || anon_inode)
+            goto out;
+    }
+
+    loc.parent = inode_ref(dir);
+    loc.inode = inode_new(inode->table);
+    if (!loc.inode) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    loc.name = name;
+    switch (replies[child].poststat.ia_type) {
+        case IA_IFDIR:
+            gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+                   "expunging dir %s/%s (%s) on %s", uuid_utoa(dir->gfid), name,
+                   uuid_utoa_r(replies[child].poststat.ia_gfid, g),
+                   subvol->name);
+            ret = syncop_rmdir(subvol, &loc, 1, NULL, NULL);
+            break;
+        default:
+            gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_EXPUNGING_FILE_OR_DIR,
+                   "expunging file %s/%s (%s) on %s", uuid_utoa(dir->gfid),
+                   name, uuid_utoa_r(replies[child].poststat.ia_gfid, g),
+                   subvol->name);
+            ret = syncop_unlink(subvol, &loc, NULL, NULL);
+            break;
+    }
+
+out:
+    loc_wipe(&loc);
+    return ret;
+}
+
+int
 afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
                             unsigned char *sources, inode_t *dir,
                             const char *name, inode_t *inode,
@@ -76,6 +192,9 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
     loc_t srcloc = {
         0,
     };
+    loc_t anonloc = {
+        0,
+    };
     xlator_t *this = frame->this;
     afr_private_t *priv = NULL;
     dict_t *xdata = NULL;
@@ -86,15 +205,18 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
         0,
     };
     unsigned char *newentry = NULL;
+    char iatt_uuid_str[64] = {0};
+    char dir_uuid_str[64] = {0};
 
     priv = this->private;
     iatt = &replies[source].poststat;
+    uuid_utoa_r(iatt->ia_gfid, iatt_uuid_str);
     if (iatt->ia_type == IA_INVAL || gf_uuid_is_null(iatt->ia_gfid)) {
         gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_SELF_HEAL_FAILED,
                "Invalid ia_type (%d) or gfid(%s). source brick=%d, "
                "pargfid=%s, name=%s",
-               iatt->ia_type, uuid_utoa(iatt->ia_gfid), source,
-               uuid_utoa(dir->gfid), name);
+               iatt->ia_type, iatt_uuid_str, source,
+               uuid_utoa_r(dir->gfid, dir_uuid_str), name);
         ret = -EINVAL;
         goto out;
     }
@@ -119,14 +241,24 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
 
     srcloc.inode = inode_ref(inode);
     gf_uuid_copy(srcloc.gfid, iatt->ia_gfid);
-    if (iatt->ia_type != IA_IFDIR)
-        ret = syncop_lookup(priv->children[dst], &srcloc, 0, 0, 0, 0);
-    if (iatt->ia_type == IA_IFDIR || ret == -ENOENT || ret == -ESTALE) {
+    ret = syncop_lookup(priv->children[dst], &srcloc, 0, 0, 0, 0);
+    if (ret == -ENOENT || ret == -ESTALE) {
         newentry[dst] = 1;
         ret = afr_selfheal_newentry_mark(frame, this, inode, source, replies,
                                          sources, newentry);
         if (ret)
             goto out;
+    } else if (ret == 0 && iatt->ia_type == IA_IFDIR && priv->use_anon_inode) {
+        // Try rename from hidden directory
+        ret = afr_anon_inode_create(this, dst, &anonloc.parent);
+        if (ret < 0)
+            goto out;
+        anonloc.inode = inode_ref(inode);
+        anonloc.name = iatt_uuid_str;
+        ret = syncop_rename(priv->children[dst], &anonloc, &loc, NULL, NULL);
+        if (ret == -ENOENT || ret == -ESTALE)
+            ret = -1; /*This sets 'mismatch' to true*/
+        goto out;
     }
 
     mode = st_mode_from_ia(iatt->ia_prot, iatt->ia_type);
@@ -149,7 +281,7 @@ afr_selfheal_recreate_entry(call_frame_t *frame, int dst, int source,
             }
             break;
         default:
-            ret = dict_set_int32(xdata, GLUSTERFS_INTERNAL_FOP_KEY, 1);
+            ret = dict_set_int32_sizen(xdata, GLUSTERFS_INTERNAL_FOP_KEY, 1);
             if (ret)
                 goto out;
             ret = syncop_mknod(
@@ -165,6 +297,7 @@ out:
     GF_FREE(linkname);
     loc_wipe(&loc);
     loc_wipe(&srcloc);
+    loc_wipe(&anonloc);
     return ret;
 }
 
@@ -246,6 +379,19 @@ afr_selfheal_detect_gfid_and_type_mismatch(xlator_t *this,
         if (replies[i].op_ret != 0)
             continue;
 
+        if (gf_uuid_is_null(replies[i].poststat.ia_gfid))
+            continue;
+
+        if (replies[i].poststat.ia_type == IA_INVAL)
+            continue;
+
+        if (ia_type == IA_INVAL || gf_uuid_is_null(gfid)) {
+            src_idx = i;
+            ia_type = replies[src_idx].poststat.ia_type;
+            gfid = &replies[src_idx].poststat.ia_gfid;
+            continue;
+        }
+
         if (gf_uuid_compare(gfid, replies[i].poststat.ia_gfid) &&
             (ia_type == replies[i].poststat.ia_type)) {
             ret = afr_gfid_split_brain_source(this, replies, inode, pargfid,
@@ -269,11 +415,12 @@ afr_selfheal_detect_gfid_and_type_mismatch(xlator_t *this,
                    gf_inode_type_to_str(replies[src_idx].poststat.ia_type),
                    priv->children[src_idx]->name);
             gf_event(EVENT_AFR_SPLIT_BRAIN,
+                     "client-pid=%d;"
                      "subvol=%s;type=file;"
                      "file=<gfid:%s>/%s>;count=2;child-%d=%s;type-"
                      "%d=%s;child-%d=%s;type-%d=%s",
-                     this->name, uuid_utoa(pargfid), bname, i,
-                     priv->children[i]->name, i,
+                     this->ctx->cmd_args.client_pid, this->name,
+                     uuid_utoa(pargfid), bname, i, priv->children[i]->name, i,
                      gf_inode_type_to_str(replies[i].poststat.ia_type), src_idx,
                      priv->children[src_idx]->name, src_idx,
                      gf_inode_type_to_str(replies[src_idx].poststat.ia_type));
@@ -465,6 +612,7 @@ __afr_selfheal_entry_finalize_source(xlator_t *this, unsigned char *sources,
     afr_private_t *priv = NULL;
     int source = -1;
     int sources_count = 0;
+    int i = 0;
 
     priv = this->private;
 
@@ -478,6 +626,20 @@ __afr_selfheal_entry_finalize_source(xlator_t *this, unsigned char *sources,
     }
 
     source = afr_choose_source_by_policy(priv, sources, AFR_ENTRY_TRANSACTION);
+
+    /*If the selected source does not blame any other brick, then mark
+     * everything as sink to trigger conservative merge.
+     */
+    if (source != -1 && !AFR_COUNT(healed_sinks, priv->child_count)) {
+        for (i = 0; i < priv->child_count; i++) {
+            if (locked_on[i]) {
+                sources[i] = 0;
+                healed_sinks[i] = 1;
+            }
+        }
+        return -1;
+    }
+
     return source;
 }
 
@@ -548,10 +710,15 @@ afr_selfheal_entry_dirent(call_frame_t *frame, xlator_t *this, fd_t *fd,
 
     priv = this->private;
 
+    if (afr_is_private_directory(priv, fd->inode->gfid, name,
+                                 GF_CLIENT_PID_SELF_HEALD)) {
+        return 0;
+    }
+
     xattr = dict_new();
     if (!xattr)
         return -ENOMEM;
-    ret = dict_set_int32(xattr, GF_GFIDLESS_LOOKUP, 1);
+    ret = dict_set_int32_sizen(xattr, GF_GFIDLESS_LOOKUP, 1);
     if (ret) {
         dict_unref(xattr);
         return -1;
@@ -568,7 +735,7 @@ afr_selfheal_entry_dirent(call_frame_t *frame, xlator_t *this, fd_t *fd,
     ret = afr_selfheal_entrylk(frame, this, fd->inode, this->name, NULL,
                                locked_on);
     {
-        if (ret < AFR_SH_MIN_PARTICIPANTS) {
+        if (ret < priv->child_count) {
             gf_msg_debug(this->name, 0,
                          "%s: Skipping "
                          "entry self-heal as only %d sub-volumes "
@@ -596,7 +763,7 @@ afr_selfheal_entry_dirent(call_frame_t *frame, xlator_t *this, fd_t *fd,
                                           replies);
 
         if ((ret == 0) && (priv->esh_granular) && parent_idx_inode) {
-            ret = afr_shd_index_purge(subvol, parent_idx_inode, name,
+            ret = afr_shd_entry_purge(subvol, parent_idx_inode, name,
                                       inode->ia_type);
             /* Why is ret force-set to 0? We do not care about
              * index purge failing for full heal as it is quite
@@ -726,10 +893,6 @@ afr_selfheal_entry_do_subvol(call_frame_t *frame, xlator_t *this, fd_t *fd,
             if (!strcmp(entry->d_name, ".") || !strcmp(entry->d_name, ".."))
                 continue;
 
-            if (__is_root_gfid(fd->inode->gfid) &&
-                !strcmp(entry->d_name, GF_REPLICATE_TRASH_DIR))
-                continue;
-
             ret = afr_selfheal_entry_dirent(iter_frame, this, fd, entry->d_name,
                                             loc.inode, subvol,
                                             local->need_full_crawl);
@@ -792,7 +955,7 @@ afr_selfheal_entry_granular_dirent(xlator_t *subvol, gf_dirent_t *entry,
         /* The name indices under the pgfid index dir are guaranteed
          * to be regular files. Hence the hardcoding.
          */
-        afr_shd_index_purge(subvol, parent->inode, entry->d_name, IA_IFREG);
+        afr_shd_entry_purge(subvol, parent->inode, entry->d_name, IA_IFREG);
         ret = 0;
         goto out;
     }
@@ -831,6 +994,8 @@ afr_selfheal_entry_granular(call_frame_t *frame, xlator_t *this, fd_t *fd,
     subvol = priv->children[subvol_idx];
 
     args.frame = afr_copy_frame(frame);
+    if (!args.frame)
+        goto out;
     args.xl = this;
     /* args.heal_fd represents the fd associated with the original directory
      * on which entry heal is being attempted.
@@ -849,9 +1014,10 @@ afr_selfheal_entry_granular(call_frame_t *frame, xlator_t *this, fd_t *fd,
          * do not treat heal as failure.
          */
         if (is_src)
-            return -errno;
+            ret = -errno;
         else
-            return 0;
+            ret = 0;
+        goto out;
     }
 
     ret = syncop_dir_scan(subvol, &loc, GF_CLIENT_PID_SELF_HEALD, &args,
@@ -861,7 +1027,9 @@ afr_selfheal_entry_granular(call_frame_t *frame, xlator_t *this, fd_t *fd,
 
     if (args.mismatch == _gf_true)
         ret = -1;
-
+out:
+    if (args.frame)
+        AFR_STACK_DESTROY(args.frame);
     return ret;
 }
 
@@ -957,7 +1125,7 @@ __afr_selfheal_entry(call_frame_t *frame, xlator_t *this, fd_t *fd,
     ret = afr_selfheal_entrylk(frame, this, fd->inode, this->name, NULL,
                                data_lock);
     {
-        if (ret < AFR_SH_MIN_PARTICIPANTS) {
+        if (ret < priv->child_count) {
             gf_msg_debug(this->name, 0,
                          "%s: Skipping "
                          "entry self-heal as only %d sub-volumes could "
@@ -1013,6 +1181,8 @@ unlock:
             goto postop_unlock;
         }
 
+        afr_selfheal_restore_time(frame, this, fd->inode, source, healed_sinks,
+                                  locked_replies);
         ret = afr_selfheal_undo_pending(
             frame, this, fd->inode, sources, sinks, healed_sinks, undid_pending,
             AFR_ENTRY_TRANSACTION, locked_replies, postop_lock);
@@ -1079,7 +1249,7 @@ afr_selfheal_entry(call_frame_t *frame, xlator_t *this, inode_t *inode)
     ret = afr_selfheal_tie_breaker_entrylk(frame, this, inode, priv->sh_domain,
                                            NULL, locked_on);
     {
-        if (ret < AFR_SH_MIN_PARTICIPANTS) {
+        if (ret < priv->child_count) {
             gf_msg_debug(this->name, 0,
                          "%s: Skipping "
                          "entry self-heal as only %d sub-volumes could "
diff --git a/xlators/cluster/afr/src/afr-self-heal-metadata.c b/xlators/cluster/afr/src/afr-self-heal-metadata.c
index ea2a7bfd52f..03f43bad16e 100644
--- a/xlators/cluster/afr/src/afr-self-heal-metadata.c
+++ b/xlators/cluster/afr/src/afr-self-heal-metadata.c
@@ -10,9 +10,9 @@
 
 #include "afr.h"
 #include "afr-self-heal.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 #include "protocol-common.h"
-#include "events.h"
+#include <glusterfs/events.h>
 
 #define AFR_HEAL_ATTR (GF_SET_ATTR_UID | GF_SET_ATTR_GID | GF_SET_ATTR_MODE)
 
@@ -190,6 +190,59 @@ out:
     return ret;
 }
 
+static int
+__afr_selfheal_metadata_mark_pending_xattrs(call_frame_t *frame, xlator_t *this,
+                                            inode_t *inode,
+                                            struct afr_reply *replies,
+                                            unsigned char *sources)
+{
+    int ret = 0;
+    int i = 0;
+    int m_idx = 0;
+    afr_private_t *priv = NULL;
+    int raw[AFR_NUM_CHANGE_LOGS] = {0};
+    dict_t *xattr = NULL;
+
+    priv = this->private;
+    m_idx = afr_index_for_transaction_type(AFR_METADATA_TRANSACTION);
+    raw[m_idx] = 1;
+
+    xattr = dict_new();
+    if (!xattr)
+        return -ENOMEM;
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (sources[i])
+            continue;
+        ret = dict_set_static_bin(xattr, priv->pending_key[i], raw,
+                                  sizeof(int) * AFR_NUM_CHANGE_LOGS);
+        if (ret) {
+            ret = -1;
+            goto out;
+        }
+    }
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (!sources[i])
+            continue;
+        ret = afr_selfheal_post_op(frame, this, inode, i, xattr, NULL);
+        if (ret < 0) {
+            gf_msg(this->name, GF_LOG_INFO, -ret, AFR_MSG_SELF_HEAL_INFO,
+                   "Failed to set pending metadata xattr on child %d for %s", i,
+                   uuid_utoa(inode->gfid));
+            goto out;
+        }
+    }
+
+    afr_replies_wipe(replies, priv->child_count);
+    ret = afr_selfheal_unlocked_discover(frame, inode, inode->gfid, replies);
+
+out:
+    if (xattr)
+        dict_unref(xattr);
+    return ret;
+}
+
 /*
  * Look for mismatching uid/gid or mode or user xattrs even if
  * AFR xattrs don't say so, and pick one arbitrarily as winner. */
@@ -210,6 +263,7 @@ __afr_selfheal_metadata_finalize_source(call_frame_t *frame, xlator_t *this,
     };
     int source = -1;
     int sources_count = 0;
+    int ret = 0;
 
     priv = this->private;
 
@@ -242,9 +296,11 @@ __afr_selfheal_metadata_finalize_source(call_frame_t *frame, xlator_t *this,
 
         if (!priv->metadata_splitbrain_forced_heal) {
             gf_event(EVENT_AFR_SPLIT_BRAIN,
+                     "client-pid=%d;"
                      "subvol=%s;"
                      "type=metadata;file=%s",
-                     this->name, uuid_utoa(inode->gfid));
+                     this->ctx->cmd_args.client_pid, this->name,
+                     uuid_utoa(inode->gfid));
             return -EIO;
         }
 
@@ -298,7 +354,13 @@ __afr_selfheal_metadata_finalize_source(call_frame_t *frame, xlator_t *this,
             healed_sinks[i] = 1;
         }
     }
-
+    if ((sources_count == priv->child_count) && (source > -1) &&
+        (AFR_COUNT(healed_sinks, priv->child_count) != 0)) {
+        ret = __afr_selfheal_metadata_mark_pending_xattrs(frame, this, inode,
+                                                          replies, sources);
+        if (ret < 0)
+            return ret;
+    }
 out:
     afr_mark_active_sinks(this, sources, locked_on, healed_sinks);
     return source;
@@ -396,7 +458,7 @@ afr_selfheal_metadata(call_frame_t *frame, xlator_t *this, inode_t *inode)
     ret = afr_selfheal_inodelk(frame, this, inode, this->name, LLONG_MAX - 1, 0,
                                data_lock);
     {
-        if (ret < AFR_SH_MIN_PARTICIPANTS) {
+        if (ret < priv->child_count) {
             ret = -ENOTCONN;
             goto unlock;
         }
@@ -419,12 +481,8 @@ afr_selfheal_metadata(call_frame_t *frame, xlator_t *this, inode_t *inode)
         if (ret)
             goto unlock;
 
-        /* Restore atime/mtime for files that don't need data heal as
-         * restoring timestamps happens only as a part of data-heal.
-         */
-        if (!IA_ISREG(locked_replies[source].poststat.ia_type))
-            afr_selfheal_restore_time(frame, this, inode, source, healed_sinks,
-                                      locked_replies);
+        afr_selfheal_restore_time(frame, this, inode, source, healed_sinks,
+                                  locked_replies);
 
         ret = afr_selfheal_undo_pending(
             frame, this, inode, sources, sinks, healed_sinks, undid_pending,
diff --git a/xlators/cluster/afr/src/afr-self-heal-name.c b/xlators/cluster/afr/src/afr-self-heal-name.c
index aa20ad1b835..834aac86d48 100644
--- a/xlators/cluster/afr/src/afr-self-heal-name.c
+++ b/xlators/cluster/afr/src/afr-self-heal-name.c
@@ -8,7 +8,7 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "events.h"
+#include <glusterfs/events.h>
 #include "afr.h"
 #include "afr-self-heal.h"
 #include "afr-messages.h"
@@ -98,21 +98,12 @@ __afr_selfheal_name_expunge(xlator_t *this, inode_t *parent, uuid_t pargfid,
                             const char *bname, inode_t *inode,
                             struct afr_reply *replies)
 {
-    loc_t loc = {
-        0,
-    };
     int i = 0;
     afr_private_t *priv = NULL;
-    char g[64];
     int ret = 0;
 
     priv = this->private;
 
-    loc.parent = inode_ref(parent);
-    gf_uuid_copy(loc.pargfid, pargfid);
-    loc.name = bname;
-    loc.inode = inode_ref(inode);
-
     for (i = 0; i < priv->child_count; i++) {
         if (!replies[i].valid)
             continue;
@@ -120,30 +111,10 @@ __afr_selfheal_name_expunge(xlator_t *this, inode_t *parent, uuid_t pargfid,
         if (replies[i].op_ret)
             continue;
 
-        switch (replies[i].poststat.ia_type) {
-            case IA_IFDIR:
-                gf_msg(this->name, GF_LOG_WARNING, 0,
-                       AFR_MSG_EXPUNGING_FILE_OR_DIR,
-                       "expunging dir %s/%s (%s) on %s", uuid_utoa(pargfid),
-                       bname, uuid_utoa_r(replies[i].poststat.ia_gfid, g),
-                       priv->children[i]->name);
-
-                ret |= syncop_rmdir(priv->children[i], &loc, 1, NULL, NULL);
-                break;
-            default:
-                gf_msg(this->name, GF_LOG_WARNING, 0,
-                       AFR_MSG_EXPUNGING_FILE_OR_DIR,
-                       "expunging file %s/%s (%s) on %s", uuid_utoa(pargfid),
-                       bname, uuid_utoa_r(replies[i].poststat.ia_gfid, g),
-                       priv->children[i]->name);
-
-                ret |= syncop_unlink(priv->children[i], &loc, NULL, NULL);
-                break;
-        }
+        ret |= afr_selfheal_entry_delete(this, parent, bname, inode, i,
+                                         replies);
     }
 
-    loc_wipe(&loc);
-
     return ret;
 }
 
@@ -222,13 +193,14 @@ afr_selfheal_name_type_mismatch_check(xlator_t *this, struct afr_reply *replies,
                        gf_inode_type_to_str(inode_type),
                        priv->children[type_idx]->name);
                 gf_event(EVENT_AFR_SPLIT_BRAIN,
+                         "client-pid=%d;"
                          "subvol=%s;type=file;"
                          "file=<gfid:%s>/%s;count=2;"
                          "child-%d=%s;type-%d=%s;child-%d=%s;"
                          "type-%d=%s",
-                         this->name, uuid_utoa(pargfid), bname, i,
-                         priv->children[i]->name, i,
-                         gf_inode_type_to_str(inode_type1), type_idx,
+                         this->ctx->cmd_args.client_pid, this->name,
+                         uuid_utoa(pargfid), bname, i, priv->children[i]->name,
+                         i, gf_inode_type_to_str(inode_type1), type_idx,
                          priv->children[type_idx]->name, type_idx,
                          gf_inode_type_to_str(inode_type));
                 return -EIO;
@@ -277,9 +249,8 @@ afr_selfheal_name_gfid_mismatch_check(xlator_t *this, struct afr_reply *replies,
                                                   bname, gfid_idx_iter, i,
                                                   locked_on, gfid_idx, xdata);
                 if (!ret && *gfid_idx >= 0) {
-                    ret = dict_set_str(xdata, "gfid-heal-msg",
-                                       "GFID split-brain "
-                                       "resolved");
+                    ret = dict_set_sizen_str_sizen(xdata, "gfid-heal-msg",
+                                                   "GFID split-brain resolved");
                     if (ret)
                         gf_msg(this->name, GF_LOG_ERROR, 0,
                                AFR_MSG_DICT_SET_FAILED,
@@ -381,7 +352,7 @@ __afr_selfheal_name_do(call_frame_t *frame, xlator_t *this, inode_t *parent,
     ret = __afr_selfheal_assign_gfid(this, parent, pargfid, bname, inode,
                                      replies, gfid, locked_on, source, sources,
                                      is_gfid_absent, &gfid_idx);
-    if (ret)
+    if (ret || (gfid_idx < 0))
         return ret;
 
     ret = __afr_selfheal_name_impunge(frame, this, parent, pargfid, bname,
@@ -496,7 +467,7 @@ afr_selfheal_name_do(call_frame_t *frame, xlator_t *this, inode_t *parent,
     if (!xattr)
         return -ENOMEM;
 
-    ret = dict_set_int32(xattr, GF_GFIDLESS_LOOKUP, 1);
+    ret = dict_set_int32_sizen(xattr, GF_GFIDLESS_LOOKUP, 1);
     if (ret) {
         dict_unref(xattr);
         return -1;
@@ -514,7 +485,7 @@ afr_selfheal_name_do(call_frame_t *frame, xlator_t *this, inode_t *parent,
     ret = afr_selfheal_entrylk(frame, this, parent, this->name, bname,
                                locked_on);
     {
-        if (ret < AFR_SH_MIN_PARTICIPANTS) {
+        if (ret < priv->child_count) {
             ret = -ENOTCONN;
             goto unlock;
         }
@@ -560,13 +531,15 @@ afr_selfheal_name_unlocked_inspect(call_frame_t *frame, xlator_t *this,
     struct afr_reply *replies = NULL;
     inode_t *inode = NULL;
     int first_idx = -1;
+    afr_local_t *local = NULL;
 
     priv = this->private;
+    local = frame->local;
 
     replies = alloca0(sizeof(*replies) * priv->child_count);
 
     inode = afr_selfheal_unlocked_lookup_on(frame, parent, bname, replies,
-                                            priv->child_up, NULL);
+                                            local->child_up, NULL);
     if (!inode)
         return -ENOMEM;
 
diff --git a/xlators/cluster/afr/src/afr-self-heal.h b/xlators/cluster/afr/src/afr-self-heal.h
index 41adbe05dc8..48e6dbcfb18 100644
--- a/xlators/cluster/afr/src/afr-self-heal.h
+++ b/xlators/cluster/afr/src/afr-self-heal.h
@@ -11,8 +11,6 @@
 #ifndef _AFR_SELFHEAL_H
 #define _AFR_SELFHEAL_H
 
-#define AFR_SH_MIN_PARTICIPANTS 2
-
 /* Perform fop on all UP subvolumes and wait for all callbacks to return */
 
 #define AFR_ONALL(frame, rfn, fop, args...)                                    \
@@ -20,9 +18,8 @@
         afr_local_t *__local = frame->local;                                   \
         afr_private_t *__priv = frame->this->private;                          \
         int __i = 0, __count = 0;                                              \
-        unsigned char *__child_up = NULL;                                      \
+        unsigned char *__child_up = alloca(__priv->child_count);               \
                                                                                \
-        __child_up = alloca0(__priv->child_count);                             \
         memcpy(__child_up, __priv->child_up,                                   \
                sizeof(*__child_up) * __priv->child_count);                     \
         __count = AFR_COUNT(__child_up, __priv->child_count);                  \
@@ -48,13 +45,16 @@
         afr_local_t *__local = frame->local;                                   \
         afr_private_t *__priv = frame->this->private;                          \
         int __i = 0;                                                           \
-        int __count = AFR_COUNT(list, __priv->child_count);                    \
+        int __count = 0;                                                       \
+        unsigned char *__list = alloca(__priv->child_count);                   \
                                                                                \
+        memcpy(__list, list, sizeof(*__list) * __priv->child_count);           \
+        __count = AFR_COUNT(__list, __priv->child_count);                      \
         __local->barrier.waitfor = __count;                                    \
         afr_local_replies_wipe(__local, __priv);                               \
                                                                                \
         for (__i = 0; __i < __priv->child_count; __i++) {                      \
-            if (!list[__i])                                                    \
+            if (!__list[__i])                                                  \
                 continue;                                                      \
             STACK_WIND_COOKIE(frame, rfn, (void *)(long)__i,                   \
                               __priv->children[__i],                           \
@@ -83,9 +83,9 @@
 
 #define ALLOC_MATRIX(n, type)                                                  \
     ({                                                                         \
-        type **__ptr = NULL;                                                   \
         int __i;                                                               \
-        __ptr = alloca0(n * sizeof(type *));                                   \
+        type **__ptr = alloca(n * sizeof(type *));                             \
+                                                                               \
         for (__i = 0; __i < n; __i++)                                          \
             __ptr[__i] = alloca0(n * sizeof(type));                            \
         __ptr;                                                                 \
@@ -97,6 +97,27 @@
 #define SBRAIN_HEAL_NO_GO_MSG                                                  \
     "Failed to obtain replies from all bricks of "                             \
     "the replica (are they up?). Cannot resolve split-brain."
+#define SFILE_NOT_IN_SPLIT_BRAIN "File not in split-brain"
+#define SNO_BIGGER_FILE "No bigger file"
+#define SNO_DIFF_IN_MTIME "No difference in mtime"
+#define SUSE_SOURCE_BRICK_TO_HEAL                                              \
+    "Use source-brick option to heal metadata"                                 \
+    " split-brain"
+#define SINVALID_BRICK_NAME "Invalid brick name"
+#define SBRICK_IS_NOT_UP "Brick is not up"
+#define SBRICK_NOT_CONNECTED "Brick is not connected"
+#define SLESS_THAN2_BRICKS_in_REP "< 2 bricks in replica are up"
+#define SBRICK_IS_REMOTE "Brick is remote"
+#define SSTARTED_SELF_HEAL "Started self-heal"
+#define SOP_NOT_SUPPORTED "Operation Not Supported"
+#define SFILE_NOT_UNDER_DATA                                                   \
+    "The file is not under data or metadata "                                  \
+    "split-brain"
+#define SFILE_NOT_IN_SPLIT_BRAIN "File not in split-brain"
+#define SALL_BRICKS_UP_TO_RESOLVE                                              \
+    "All the bricks should be up to resolve the"                               \
+    " gfid split brain"
+#define SERROR_GETTING_SRC_BRICK "Error getting the source brick"
 int
 afr_selfheal(xlator_t *this, uuid_t gfid);
 
@@ -166,7 +187,7 @@ afr_selfheal_unlocked_discover(call_frame_t *frame, inode_t *inode, uuid_t gfid,
 int
 afr_selfheal_unlocked_discover_on(call_frame_t *frame, inode_t *inode,
                                   uuid_t gfid, struct afr_reply *replies,
-                                  unsigned char *discover_on);
+                                  unsigned char *discover_on, dict_t *dict);
 inode_t *
 afr_selfheal_unlocked_lookup_on(call_frame_t *frame, inode_t *parent,
                                 const char *name, struct afr_reply *replies,
@@ -305,7 +326,8 @@ int
 afr_selfheal_unlocked_inspect(call_frame_t *frame, xlator_t *this, uuid_t gfid,
                               inode_t **link_inode, gf_boolean_t *data_selfheal,
                               gf_boolean_t *metadata_selfheal,
-                              gf_boolean_t *entry_selfheal);
+                              gf_boolean_t *entry_selfheal,
+                              struct afr_reply *replies);
 
 int
 afr_selfheal_do(call_frame_t *frame, xlator_t *this, uuid_t gfid);
@@ -347,4 +369,9 @@ gf_boolean_t
 afr_is_file_empty_on_all_children(afr_private_t *priv,
                                   struct afr_reply *replies);
 
+int
+afr_selfheal_entry_delete(xlator_t *this, inode_t *dir, const char *name,
+                          inode_t *inode, int child, struct afr_reply *replies);
+int
+afr_anon_inode_create(xlator_t *this, int child, inode_t **linked_inode);
 #endif /* !_AFR_SELFHEAL_H */
diff --git a/xlators/cluster/afr/src/afr-self-heald.c b/xlators/cluster/afr/src/afr-self-heald.c
index e2de77a9c45..109fd4b7421 100644
--- a/xlators/cluster/afr/src/afr-self-heald.c
+++ b/xlators/cluster/afr/src/afr-self-heald.c
@@ -12,11 +12,10 @@
 #include "afr-self-heal.h"
 #include "afr-self-heald.h"
 #include "protocol-common.h"
-#include "syncop-utils.h"
+#include <glusterfs/syncop-utils.h>
 #include "afr-messages.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 
-#define SHD_INODE_LRU_LIMIT 2048
 #define AFR_EH_SPLIT_BRAIN_LIMIT 1024
 #define AFR_STATISTICS_HISTORY_SIZE 50
 
@@ -95,7 +94,7 @@ __afr_shd_healer_wait(struct subvol_healer *healer)
     priv = healer->this->private;
 
 disabled_loop:
-    wait_till.tv_sec = time(NULL) + priv->shd.timeout;
+    wait_till.tv_sec = gf_time() + priv->shd.timeout;
 
     while (!healer->rerun) {
         ret = pthread_cond_timedwait(&healer->cond, &healer->mutex, &wait_till);
@@ -223,7 +222,7 @@ out:
 }
 
 int
-afr_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name,
+afr_shd_entry_purge(xlator_t *subvol, inode_t *inode, char *name,
                     ia_type_t type)
 {
     int ret = 0;
@@ -372,8 +371,9 @@ afr_shd_sweep_prepare(struct subvol_healer *healer)
     event->split_brain_count = 0;
     event->heal_failed_count = 0;
 
-    time(&event->start_time);
+    event->start_time = gf_time();
     event->end_time = 0;
+    _mask_cancellation();
 }
 
 void
@@ -386,8 +386,8 @@ afr_shd_sweep_done(struct subvol_healer *healer)
     event = &healer->crawl_event;
     shd = &(((afr_private_t *)healer->this->private)->shd);
 
-    time(&event->end_time);
-    history = memdup(event, sizeof(*event));
+    event->end_time = gf_time();
+    history = gf_memdup(event, sizeof(*event));
     event->start_time = 0;
 
     if (!history)
@@ -395,6 +395,7 @@ afr_shd_sweep_done(struct subvol_healer *healer)
 
     if (eh_save_history(shd->statistics[healer->subvol], history) < 0)
         GF_FREE(history);
+    _unmask_cancellation();
 }
 
 int
@@ -423,7 +424,7 @@ afr_shd_index_heal(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
     ret = afr_shd_selfheal(healer, healer->subvol, gfid);
 
     if (ret == -ENOENT || ret == -ESTALE)
-        afr_shd_index_purge(subvol, parent->inode, entry->d_name, val);
+        afr_shd_entry_purge(subvol, parent->inode, entry->d_name, val);
 
     if (ret == 2)
         /* If bricks crashed in pre-op after creating indices/xattrop
@@ -465,7 +466,7 @@ afr_shd_index_sweep(struct subvol_healer *healer, char *vgfid)
     }
 
     xdata = dict_new();
-    if (!xdata || dict_set_int32(xdata, "get-gfid-type", 1)) {
+    if (!xdata || dict_set_int32_sizen(xdata, "get-gfid-type", 1)) {
         ret = -ENOMEM;
         goto out;
     }
@@ -523,6 +524,11 @@ afr_shd_full_heal(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
     afr_private_t *priv = NULL;
 
     priv = this->private;
+
+    if (this->cleanup_starting) {
+        return -ENOTCONN;
+    }
+
     if (!priv->shd.enabled)
         return -EBUSY;
 
@@ -591,7 +597,9 @@ _afr_shd_ta_get_xattrs(xlator_t *this, loc_t *loc, dict_t **xdata)
 {
     afr_private_t *priv = NULL;
     dict_t *xattr = NULL;
-    int *raw = NULL;
+    int raw[AFR_NUM_CHANGE_LOGS] = {
+        0,
+    };
     int ret = -1;
     int i = 0;
 
@@ -603,18 +611,11 @@ _afr_shd_ta_get_xattrs(xlator_t *this, loc_t *loc, dict_t **xdata)
                "Failed to create dict.");
         goto out;
     }
-
     for (i = 0; i < priv->child_count; i++) {
-        raw = GF_CALLOC(AFR_NUM_CHANGE_LOGS, sizeof(int), gf_afr_mt_int32_t);
-        if (!raw)
-            goto out;
-
-        ret = dict_set_bin(xattr, priv->pending_key[i], raw,
-                           AFR_NUM_CHANGE_LOGS * sizeof(int));
-        if (ret) {
-            GF_FREE(raw);
+        ret = dict_set_static_bin(xattr, priv->pending_key[i], &raw,
+                                  AFR_NUM_CHANGE_LOGS * sizeof(int));
+        if (ret)
             goto out;
-        }
     }
 
     ret = syncop_xattrop(priv->children[THIN_ARBITER_BRICK_INDEX], loc,
@@ -641,6 +642,7 @@ afr_shd_ta_get_xattrs(xlator_t *this, loc_t *loc, struct subvol_healer *healer,
     if (afr_shd_fill_ta_loc(this, loc)) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to populate thin-arbiter loc for: %s.", loc->name);
+        ret = -1;
         goto out;
     }
 
@@ -722,9 +724,9 @@ afr_shd_ta_unset_xattrs(xlator_t *this, loc_t *loc, dict_t **xdata, int healer)
         }
 
         ret = dict_set_bin(xattr, priv->pending_key[i], raw,
-                           AFR_NUM_CHANGE_LOGS * sizeof (int));
+                           AFR_NUM_CHANGE_LOGS * sizeof(int));
         if (ret) {
-            GF_FREE (raw);
+            GF_FREE(raw);
             goto out;
         }
 
@@ -799,6 +801,218 @@ afr_bricks_available_for_heal(afr_private_t *priv)
     return _gf_true;
 }
 
+static gf_boolean_t
+afr_shd_ta_needs_heal(xlator_t *this, struct subvol_healer *healer)
+{
+    dict_t *xdata = NULL;
+    afr_private_t *priv = NULL;
+    loc_t loc = {
+        0,
+    };
+    int ret = -1;
+    int i = 0;
+    gf_boolean_t need_heal = _gf_false;
+
+    priv = this->private;
+
+    ret = afr_shd_fill_ta_loc(this, &loc);
+    if (ret) {
+        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
+               "Failed to populate thin-arbiter loc for: %s.", loc.name);
+        healer->rerun = 1;
+        goto out;
+    }
+
+    if (_afr_shd_ta_get_xattrs(this, &loc, &xdata)) {
+        healer->rerun = 1;
+        goto out;
+    }
+
+    for (i = 0; i < priv->child_count; i++) {
+        if (afr_ta_dict_contains_pending_xattr(xdata, priv, i)) {
+            need_heal = _gf_true;
+            break;
+        }
+    }
+
+out:
+    if (xdata)
+        dict_unref(xdata);
+    loc_wipe(&loc);
+
+    return need_heal;
+}
+
+static int
+afr_shd_anon_inode_cleaner(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
+                           void *data)
+{
+    struct subvol_healer *healer = data;
+    afr_private_t *priv = healer->this->private;
+    call_frame_t *frame = NULL;
+    afr_local_t *local = NULL;
+    int ret = 0;
+    loc_t loc = {0};
+    int count = 0;
+    int i = 0;
+    int op_errno = 0;
+    struct iatt *iatt = NULL;
+    gf_boolean_t multiple_links = _gf_false;
+    unsigned char *gfid_present = alloca0(priv->child_count);
+    unsigned char *entry_present = alloca0(priv->child_count);
+    char *type = "file";
+
+    frame = afr_frame_create(healer->this, &ret);
+    if (!frame) {
+        ret = -ret;
+        goto out;
+    }
+    local = frame->local;
+    if (AFR_COUNT(local->child_up, priv->child_count) != priv->child_count) {
+        gf_msg_debug(healer->this->name, 0,
+                     "Not all bricks are up. Skipping "
+                     "cleanup of %s on %s",
+                     entry->d_name, subvol->name);
+        ret = 0;
+        goto out;
+    }
+
+    loc.inode = inode_new(parent->inode->table);
+    if (!loc.inode) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    ret = gf_uuid_parse(entry->d_name, loc.gfid);
+    if (ret) {
+        ret = 0;
+        goto out;
+    }
+    AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup, &loc,
+               NULL);
+    for (i = 0; i < priv->child_count; i++) {
+        if (local->replies[i].op_ret == 0) {
+            count++;
+            gfid_present[i] = 1;
+            iatt = &local->replies[i].poststat;
+            if (iatt->ia_type == IA_IFDIR) {
+                type = "dir";
+            }
+
+            if (i == healer->subvol) {
+                if (local->replies[i].poststat.ia_nlink > 1) {
+                    multiple_links = _gf_true;
+                }
+            }
+        } else if (local->replies[i].op_errno != ENOENT &&
+                   local->replies[i].op_errno != ESTALE) {
+            /*We don't have complete view. Skip the entry*/
+            gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+                         "Skipping cleanup of %s on %s", entry->d_name,
+                         subvol->name);
+            ret = 0;
+            goto out;
+        }
+    }
+
+    /*Inode is deleted from subvol*/
+    if (count == 1 || (iatt->ia_type != IA_IFDIR && multiple_links)) {
+        gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+               AFR_MSG_EXPUNGING_FILE_OR_DIR, "expunging %s %s/%s on %s", type,
+               priv->anon_inode_name, entry->d_name, subvol->name);
+        ret = afr_shd_entry_purge(subvol, parent->inode, entry->d_name,
+                                  iatt->ia_type);
+        if (ret == -ENOENT || ret == -ESTALE)
+            ret = 0;
+    } else if (count > 1) {
+        loc_wipe(&loc);
+        loc.parent = inode_ref(parent->inode);
+        loc.name = entry->d_name;
+        loc.inode = inode_new(parent->inode->table);
+        if (!loc.inode) {
+            ret = -ENOMEM;
+            goto out;
+        }
+        AFR_ONLIST(local->child_up, frame, afr_selfheal_discover_cbk, lookup,
+                   &loc, NULL);
+        count = 0;
+        for (i = 0; i < priv->child_count; i++) {
+            if (local->replies[i].op_ret == 0) {
+                count++;
+                entry_present[i] = 1;
+                iatt = &local->replies[i].poststat;
+            } else if (local->replies[i].op_errno != ENOENT &&
+                       local->replies[i].op_errno != ESTALE) {
+                /*We don't have complete view. Skip the entry*/
+                gf_msg_debug(healer->this->name, local->replies[i].op_errno,
+                             "Skipping cleanup of %s on %s", entry->d_name,
+                             subvol->name);
+                ret = 0;
+                goto out;
+            }
+        }
+        for (i = 0; i < priv->child_count; i++) {
+            if (gfid_present[i] && !entry_present[i]) {
+                /*Entry is not anonymous on at least one subvol*/
+                gf_msg_debug(healer->this->name, 0,
+                             "Valid entry present on %s "
+                             "Skipping cleanup of %s on %s",
+                             priv->children[i]->name, entry->d_name,
+                             subvol->name);
+                ret = 0;
+                goto out;
+            }
+        }
+
+        gf_msg(healer->this->name, GF_LOG_WARNING, 0,
+               AFR_MSG_EXPUNGING_FILE_OR_DIR,
+               "expunging %s %s/%s on all subvols", type, priv->anon_inode_name,
+               entry->d_name);
+        ret = 0;
+        for (i = 0; i < priv->child_count; i++) {
+            op_errno = -afr_shd_entry_purge(priv->children[i], loc.parent,
+                                            entry->d_name, iatt->ia_type);
+            if (op_errno != ENOENT && op_errno != ESTALE) {
+                ret |= -op_errno;
+            }
+        }
+    }
+
+out:
+    if (frame)
+        AFR_STACK_DESTROY(frame);
+    loc_wipe(&loc);
+    return ret;
+}
+
+static void
+afr_cleanup_anon_inode_dir(struct subvol_healer *healer)
+{
+    int ret = 0;
+    call_frame_t *frame = NULL;
+    afr_private_t *priv = healer->this->private;
+    loc_t loc = {0};
+
+    ret = afr_anon_inode_create(healer->this, healer->subvol, &loc.inode);
+    if (ret)
+        goto out;
+
+    frame = afr_frame_create(healer->this, &ret);
+    if (!frame) {
+        ret = -ret;
+        goto out;
+    }
+
+    ret = syncop_mt_dir_scan(frame, priv->children[healer->subvol], &loc,
+                             GF_CLIENT_PID_SELF_HEALD, healer,
+                             afr_shd_anon_inode_cleaner, NULL,
+                             priv->shd.max_threads, priv->shd.wait_qlength);
+out:
+    if (frame)
+        AFR_STACK_DESTROY(frame);
+    loc_wipe(&loc);
+    return;
+}
+
 void *
 afr_shd_index_healer(void *data)
 {
@@ -825,7 +1039,8 @@ afr_shd_index_healer(void *data)
         priv->local[healer->subvol] = healer->local;
 
         if (priv->thin_arbiter_count) {
-            afr_shd_ta_get_xattrs(this, &loc, healer, &pre_crawl_xdata);
+            if (afr_shd_ta_needs_heal(this, healer))
+                afr_shd_ta_get_xattrs(this, &loc, healer, &pre_crawl_xdata);
         }
 
         do {
@@ -855,9 +1070,17 @@ afr_shd_index_healer(void *data)
             sleep(1);
         } while (ret > 0);
 
-        if (pre_crawl_xdata && !healer->crawl_event.heal_failed_count) {
+        if (ret == 0) {
+            afr_cleanup_anon_inode_dir(healer);
+        }
+
+        if (ret == 0 && pre_crawl_xdata &&
+            !healer->crawl_event.heal_failed_count) {
             afr_shd_ta_check_and_unset_xattrs(this, &loc, healer,
                                               pre_crawl_xdata);
+        }
+
+        if (pre_crawl_xdata) {
             dict_unref(pre_crawl_xdata);
             pre_crawl_xdata = NULL;
         }
@@ -975,7 +1198,9 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
 {
     int ret = 0;
     uint64_t count = 0;
-    char key[256] = {0};
+    char key[128] = {0};
+    int keylen = 0;
+    char suffix[64] = {0};
     int xl_id = 0;
     uint64_t healed_count = 0;
     uint64_t split_brain_count = 0;
@@ -1010,8 +1235,8 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
     snprintf(key, sizeof(key), "statistics-%d-%d-count", xl_id, child);
     ret = dict_get_uint64(output, key, &count);
 
-    snprintf(key, sizeof(key), "statistics_healed_cnt-%d-%d-%" PRIu64, xl_id,
-             child, count);
+    snprintf(suffix, sizeof(suffix), "%d-%d-%" PRIu64, xl_id, child, count);
+    snprintf(key, sizeof(key), "statistics_healed_cnt-%s", suffix);
     ret = dict_set_uint64(output, key, healed_count);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
@@ -1019,8 +1244,7 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
         goto out;
     }
 
-    snprintf(key, sizeof(key), "statistics_sb_cnt-%d-%d-%" PRIu64, xl_id, child,
-             count);
+    snprintf(key, sizeof(key), "statistics_sb_cnt-%s", suffix);
     ret = dict_set_uint64(output, key, split_brain_count);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
@@ -1028,17 +1252,15 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
         goto out;
     }
 
-    snprintf(key, sizeof(key), "statistics_crawl_type-%d-%d-%" PRIu64, xl_id,
-             child, count);
-    ret = dict_set_str(output, key, crawl_type);
+    keylen = snprintf(key, sizeof(key), "statistics_crawl_type-%s", suffix);
+    ret = dict_set_strn(output, key, keylen, crawl_type);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                "Could not add statistics_crawl_type to output");
         goto out;
     }
 
-    snprintf(key, sizeof(key), "statistics_heal_failed_cnt-%d-%d-%" PRIu64,
-             xl_id, child, count);
+    snprintf(key, sizeof(key), "statistics_heal_failed_cnt-%s", suffix);
     ret = dict_set_uint64(output, key, heal_failed_count);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
@@ -1046,9 +1268,8 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
         goto out;
     }
 
-    snprintf(key, sizeof(key), "statistics_strt_time-%d-%d-%" PRIu64, xl_id,
-             child, count);
-    ret = dict_set_dynstr(output, key, start_time_str);
+    keylen = snprintf(key, sizeof(key), "statistics_strt_time-%s", suffix);
+    ret = dict_set_dynstrn(output, key, keylen, start_time_str);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                "Could not add statistics_crawl_start_time to output");
@@ -1062,11 +1283,10 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
     else
         progress = 0;
 
-    snprintf(key, sizeof(key), "statistics_end_time-%d-%d-%" PRIu64, xl_id,
-             child, count);
+    keylen = snprintf(key, sizeof(key), "statistics_end_time-%s", suffix);
     if (!end_time_str)
         end_time_str = gf_strdup("Could not determine the end time");
-    ret = dict_set_dynstr(output, key, end_time_str);
+    ret = dict_set_dynstrn(output, key, keylen, end_time_str);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                "Could not add statistics_crawl_end_time to output");
@@ -1075,10 +1295,9 @@ afr_shd_dict_add_crawl_event(xlator_t *this, dict_t *output,
         end_time_str = NULL;
     }
 
-    snprintf(key, sizeof(key), "statistics_inprogress-%d-%d-%" PRIu64, xl_id,
-             child, count);
+    keylen = snprintf(key, sizeof(key), "statistics_inprogress-%s", suffix);
 
-    ret = dict_set_int32(output, key, progress);
+    ret = dict_set_int32n(output, key, keylen, progress);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
                "Could not add statistics_inprogress to output");
@@ -1104,7 +1323,9 @@ afr_shd_dict_add_path(xlator_t *this, dict_t *output, int child, char *path,
 {
     int ret = -1;
     uint64_t count = 0;
-    char key[256] = {0};
+    char key[64] = {0};
+    int keylen = 0;
+    char xl_id_child_str[32] = {0};
     int xl_id = 0;
 
     ret = dict_get_int32(output, this->name, &xl_id);
@@ -1114,11 +1335,12 @@ afr_shd_dict_add_path(xlator_t *this, dict_t *output, int child, char *path,
         goto out;
     }
 
-    snprintf(key, sizeof(key), "%d-%d-count", xl_id, child);
+    snprintf(xl_id_child_str, sizeof(xl_id_child_str), "%d-%d", xl_id, child);
+    snprintf(key, sizeof(key), "%s-count", xl_id_child_str);
     ret = dict_get_uint64(output, key, &count);
 
-    snprintf(key, sizeof(key), "%d-%d-%" PRIu64, xl_id, child, count);
-    ret = dict_set_dynstr(output, key, path);
+    keylen = snprintf(key, sizeof(key), "%s-%" PRIu64, xl_id_child_str, count);
+    ret = dict_set_dynstrn(output, key, keylen, path);
 
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
@@ -1127,7 +1349,7 @@ afr_shd_dict_add_path(xlator_t *this, dict_t *output, int child, char *path,
     }
 
     if (tv) {
-        snprintf(key, sizeof(key), "%d-%d-%" PRIu64 "-time", xl_id, child,
+        snprintf(key, sizeof(key), "%s-%" PRIu64 "-time", xl_id_child_str,
                  count);
         ret = dict_set_uint32(output, key, tv->tv_sec);
         if (ret) {
@@ -1137,7 +1359,7 @@ afr_shd_dict_add_path(xlator_t *this, dict_t *output, int child, char *path,
         }
     }
 
-    snprintf(key, sizeof(key), "%d-%d-count", xl_id, child);
+    snprintf(key, sizeof(key), "%s-count", xl_id_child_str);
 
     ret = dict_set_uint64(output, key, count + 1);
     if (ret) {
@@ -1210,10 +1432,6 @@ afr_selfheal_daemon_init(xlator_t *this)
     priv = this->private;
     shd = &priv->shd;
 
-    this->itable = inode_table_new(SHD_INODE_LRU_LIMIT, this);
-    if (!this->itable)
-        goto out;
-
     shd->index_healers = GF_CALLOC(sizeof(*shd->index_healers),
                                    priv->child_count,
                                    gf_afr_mt_subvol_healer_t);
@@ -1264,12 +1482,18 @@ out:
     return ret;
 }
 
-int
-afr_selfheal_childup(xlator_t *this, int subvol)
+void
+afr_selfheal_childup(xlator_t *this, afr_private_t *priv)
 {
-    afr_shd_index_healer_spawn(this, subvol);
+    int subvol = 0;
 
-    return 0;
+    if (!priv->shd.iamshd)
+        return;
+    for (subvol = 0; subvol < priv->child_count; subvol++)
+        if (priv->child_up[subvol])
+            afr_shd_index_healer_spawn(this, subvol);
+
+    return;
 }
 
 int
@@ -1319,41 +1543,78 @@ afr_xl_op(xlator_t *this, dict_t *input, dict_t *output)
     struct subvol_healer *healer = NULL;
     int i = 0;
     char key[64];
+    int keylen = 0;
+    int this_name_len = 0;
     int op_ret = 0;
     uint64_t cnt = 0;
 
+#define AFR_SET_DICT_AND_LOG(name, output, key, keylen, dict_str,              \
+                             dict_str_len)                                     \
+    {                                                                          \
+        int ret;                                                               \
+                                                                               \
+        ret = dict_set_nstrn(output, key, keylen, dict_str, dict_str_len);     \
+        if (ret) {                                                             \
+            gf_smsg(name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,         \
+                    "key=%s", key, "value=%s", dict_str, NULL);                \
+        }                                                                      \
+    }
+
     priv = this->private;
     shd = &priv->shd;
 
-    ret = dict_get_int32(input, "xl-op", (int32_t *)&op);
-    if (ret)
+    ret = dict_get_int32_sizen(input, "xl-op", (int32_t *)&op);
+    if (ret) {
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_GET_FAILED,
+                "key=xl-op", NULL);
         goto out;
-    ret = dict_get_int32(input, this->name, &xl_id);
-    if (ret)
+    }
+    this_name_len = strlen(this->name);
+    ret = dict_get_int32n(input, this->name, this_name_len, &xl_id);
+    if (ret) {
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_GET_FAILED,
+                "key=%s", this->name, NULL);
         goto out;
-    ret = dict_set_int32(output, this->name, xl_id);
-    if (ret)
+    }
+    ret = dict_set_int32n(output, this->name, this_name_len, xl_id);
+    if (ret) {
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_DICT_SET_FAILED,
+                "key=%s", this->name, NULL);
         goto out;
+    }
     switch (op) {
         case GF_SHD_OP_HEAL_INDEX:
             op_ret = 0;
 
             for (i = 0; i < priv->child_count; i++) {
                 healer = &shd->index_healers[i];
-                snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
+                keylen = snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
 
                 if (!priv->child_up[i]) {
-                    ret = dict_set_str(output, key, "Brick is not connected");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SBRICK_NOT_CONNECTED,
+                                         SLEN(SBRICK_NOT_CONNECTED));
                     op_ret = -1;
                 } else if (AFR_COUNT(priv->child_up, priv->child_count) < 2) {
-                    ret = dict_set_str(output, key,
-                                       "< 2 bricks in replica are up");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SLESS_THAN2_BRICKS_in_REP,
+                                         SLEN(SLESS_THAN2_BRICKS_in_REP));
                     op_ret = -1;
                 } else if (!afr_shd_is_subvol_local(this, healer->subvol)) {
-                    ret = dict_set_str(output, key, "Brick is remote");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SBRICK_IS_REMOTE,
+                                         SLEN(SBRICK_IS_REMOTE));
                 } else {
-                    ret = dict_set_str(output, key, "Started self-heal");
-                    afr_shd_index_healer_spawn(this, i);
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SSTARTED_SELF_HEAL,
+                                         SLEN(SSTARTED_SELF_HEAL));
+
+                    ret = afr_shd_index_healer_spawn(this, i);
+
+                    if (ret) {
+                        gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                                AFR_MSG_HEALER_SPAWN_FAILED, NULL);
+                    }
                 }
             }
             break;
@@ -1362,18 +1623,31 @@ afr_xl_op(xlator_t *this, dict_t *input, dict_t *output)
 
             for (i = 0; i < priv->child_count; i++) {
                 healer = &shd->full_healers[i];
-                snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
+                keylen = snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
 
                 if (!priv->child_up[i]) {
-                    ret = dict_set_str(output, key, "Brick is not connected");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SBRICK_NOT_CONNECTED,
+                                         SLEN(SBRICK_NOT_CONNECTED));
                 } else if (AFR_COUNT(priv->child_up, priv->child_count) < 2) {
-                    ret = dict_set_str(output, key,
-                                       "< 2 bricks in replica are up");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SLESS_THAN2_BRICKS_in_REP,
+                                         SLEN(SLESS_THAN2_BRICKS_in_REP));
                 } else if (!afr_shd_is_subvol_local(this, healer->subvol)) {
-                    ret = dict_set_str(output, key, "Brick is remote");
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SBRICK_IS_REMOTE,
+                                         SLEN(SBRICK_IS_REMOTE));
                 } else {
-                    ret = dict_set_str(output, key, "Started self-heal");
-                    afr_shd_full_healer_spawn(this, i);
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SSTARTED_SELF_HEAL,
+                                         SLEN(SSTARTED_SELF_HEAL));
+
+                    ret = afr_shd_full_healer_spawn(this, i);
+
+                    if (ret) {
+                        gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                                AFR_MSG_HEALER_SPAWN_FAILED, NULL);
+                    }
                     op_ret = 0;
                 }
             }
@@ -1381,25 +1655,25 @@ afr_xl_op(xlator_t *this, dict_t *input, dict_t *output)
         case GF_SHD_OP_INDEX_SUMMARY:
             /* this case has been handled in glfs-heal.c */
             break;
-        case GF_SHD_OP_HEALED_FILES:
-        case GF_SHD_OP_HEAL_FAILED_FILES:
-            for (i = 0; i < priv->child_count; i++) {
-                snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
-                ret = dict_set_str(output, key,
-                                   "Operation Not "
-                                   "Supported");
-            }
-            break;
         case GF_SHD_OP_SPLIT_BRAIN_FILES:
             eh_dump(shd->split_brain, output, afr_add_shd_event);
             break;
         case GF_SHD_OP_STATISTICS:
             for (i = 0; i < priv->child_count; i++) {
                 eh_dump(shd->statistics[i], output, afr_add_crawl_event);
-                afr_shd_dict_add_crawl_event(
+                ret = afr_shd_dict_add_crawl_event(
                     this, output, &shd->index_healers[i].crawl_event);
-                afr_shd_dict_add_crawl_event(this, output,
-                                             &shd->full_healers[i].crawl_event);
+                if (ret) {
+                    gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                            AFR_MSG_ADD_CRAWL_EVENT_FAILED, NULL);
+                }
+
+                ret = afr_shd_dict_add_crawl_event(
+                    this, output, &shd->full_healers[i].crawl_event);
+                if (ret) {
+                    gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                            AFR_MSG_ADD_CRAWL_EVENT_FAILED, NULL);
+                }
             }
             break;
         case GF_SHD_OP_STATISTICS_HEAL_COUNT:
@@ -1408,14 +1682,21 @@ afr_xl_op(xlator_t *this, dict_t *input, dict_t *output)
 
             for (i = 0; i < priv->child_count; i++) {
                 if (!priv->child_up[i]) {
-                    snprintf(key, sizeof(key), "%d-%d-status", xl_id, i);
-                    ret = dict_set_str(output, key, "Brick is not connected");
+                    keylen = snprintf(key, sizeof(key), "%d-%d-status", xl_id,
+                                      i);
+                    AFR_SET_DICT_AND_LOG(this->name, output, key, keylen,
+                                         SBRICK_NOT_CONNECTED,
+                                         SLEN(SBRICK_NOT_CONNECTED));
                 } else {
                     snprintf(key, sizeof(key), "%d-%d-hardlinks", xl_id, i);
                     ret = afr_shd_get_index_count(this, i, &cnt);
                     if (ret == 0) {
                         ret = dict_set_uint64(output, key, cnt);
                     }
+                    if (ret) {
+                        gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                                AFR_MSG_DICT_SET_FAILED, NULL);
+                    }
                     op_ret = 0;
                 }
             }
@@ -1423,11 +1704,13 @@ afr_xl_op(xlator_t *this, dict_t *input, dict_t *output)
             break;
 
         default:
-            gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_ARG,
-                   "Unknown set op %d", op);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, AFR_MSG_INVALID_ARG, "op=%d",
+                    op, NULL);
             break;
     }
 out:
-    dict_del(output, this->name);
+    dict_deln(output, this->name, this_name_len);
     return op_ret;
+
+#undef AFR_SET_DICT_AND_LOG
 }
diff --git a/xlators/cluster/afr/src/afr-self-heald.h b/xlators/cluster/afr/src/afr-self-heald.h
index 7de7c431460..18db728ea7b 100644
--- a/xlators/cluster/afr/src/afr-self-heald.h
+++ b/xlators/cluster/afr/src/afr-self-heald.h
@@ -14,12 +14,11 @@
 #include <pthread.h>
 
 typedef struct {
-    int child;
     char *path;
+    int child;
 } shd_event_t;
 
 typedef struct {
-    int child;
     uint64_t healed_count;
     uint64_t split_brain_count;
     uint64_t heal_failed_count;
@@ -31,38 +30,36 @@ typedef struct {
        cralwer is in progress */
     time_t end_time;
     char *crawl_type;
+    int child;
 } crawl_event_t;
 
 struct subvol_healer {
     xlator_t *this;
-    int subvol;
-    gf_boolean_t local;
-    gf_boolean_t running;
-    gf_boolean_t rerun;
     crawl_event_t crawl_event;
     pthread_mutex_t mutex;
     pthread_cond_t cond;
     pthread_t thread;
+    int subvol;
+    gf_boolean_t local;
+    gf_boolean_t running;
+    gf_boolean_t rerun;
 };
 
 typedef struct {
-    gf_boolean_t iamshd;
-    gf_boolean_t enabled;
-    int timeout;
     struct subvol_healer *index_healers;
     struct subvol_healer *full_healers;
 
     eh_t *split_brain;
     eh_t **statistics;
+    int timeout;
     uint32_t max_threads;
     uint32_t wait_qlength;
     uint32_t halo_max_latency_msec;
+    gf_boolean_t iamshd;
+    gf_boolean_t enabled;
 } afr_self_heald_t;
 
 int
-afr_selfheal_childup(xlator_t *this, int subvol);
-
-int
 afr_selfheal_daemon_init(xlator_t *this);
 
 int
@@ -73,6 +70,6 @@ afr_shd_gfid_to_path(xlator_t *this, xlator_t *subvol, uuid_t gfid,
                      char **path_p);
 
 int
-afr_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name,
+afr_shd_entry_purge(xlator_t *subvol, inode_t *inode, char *name,
                     ia_type_t type);
 #endif /* !_AFR_SELF_HEALD_H */
diff --git a/xlators/cluster/afr/src/afr-transaction.c b/xlators/cluster/afr/src/afr-transaction.c
index fb78c198d9c..a51f79b1f43 100644
--- a/xlators/cluster/afr/src/afr-transaction.c
+++ b/xlators/cluster/afr/src/afr-transaction.c
@@ -8,10 +8,10 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "dict.h"
-#include "byte-order.h"
-#include "common-utils.h"
-#include "timer.h"
+#include <glusterfs/dict.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/timer.h>
 
 #include "afr.h"
 #include "afr-transaction.h"
@@ -28,6 +28,12 @@ typedef enum {
 static void
 afr_lock_resume_shared(struct list_head *list);
 
+static void
+afr_post_op_handle_success(call_frame_t *frame, xlator_t *this);
+
+static void
+afr_post_op_handle_failure(call_frame_t *frame, xlator_t *this, int op_errno);
+
 void
 __afr_transaction_wake_shared(afr_local_t *local, struct list_head *shared);
 
@@ -68,6 +74,14 @@ afr_changelog_post_op_done(call_frame_t *frame, xlator_t *this);
 static void
 afr_changelog_post_op_fail(call_frame_t *frame, xlator_t *this, int op_errno);
 
+void
+afr_ta_locked_priv_invalidate(afr_private_t *priv)
+{
+    priv->ta_bad_child_index = AFR_CHILD_UNKNOWN;
+    priv->release_ta_notify_dom_lock = _gf_false;
+    priv->ta_notify_dom_lock_offset = 0;
+}
+
 static void
 afr_ta_process_waitq(xlator_t *this)
 {
@@ -110,9 +124,9 @@ afr_release_notify_lock_for_ta(void *opaque)
 
     this = (xlator_t *)opaque;
     priv = this->private;
-    ret = afr_fill_ta_loc(this, &loc);
+    ret = afr_fill_ta_loc(this, &loc, _gf_true);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, AFR_MSG_THIN_ARB,
+        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to populate loc for thin-arbiter.");
         goto out;
     }
@@ -121,17 +135,16 @@ afr_release_notify_lock_for_ta(void *opaque)
     flock.l_len = 1;
     ret = syncop_inodelk(priv->children[THIN_ARBITER_BRICK_INDEX],
                          AFR_TA_DOM_NOTIFY, &loc, F_SETLK, &flock, NULL, NULL);
-    if (!ret) {
-        LOCK(&priv->lock);
-        priv->ta_bad_child_index = AFR_CHILD_UNKNOWN;
-        priv->release_ta_notify_dom_lock = _gf_false;
-        priv->ta_notify_dom_lock_offset = 0;
-        UNLOCK(&priv->lock);
-    } else {
+    if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to unlock AFR_TA_DOM_NOTIFY lock.");
     }
 
+    LOCK(&priv->lock);
+    {
+        afr_ta_locked_priv_invalidate(priv);
+    }
+    UNLOCK(&priv->lock);
 out:
     loc_wipe(&loc);
     return ret;
@@ -242,7 +255,7 @@ afr_changelog_has_quorum(afr_local_t *local, xlator_t *this)
         }
     }
 
-    if (afr_has_quorum(success_children, this)) {
+    if (afr_has_quorum(success_children, this, NULL)) {
         return _gf_true;
     }
 
@@ -366,6 +379,8 @@ afr_transaction_done(call_frame_t *frame, xlator_t *this)
     }
     local->transaction.unwind(frame, this);
 
+    GF_ASSERT(list_empty(&local->transaction.owner_list));
+    GF_ASSERT(list_empty(&local->transaction.wait_list));
     AFR_STACK_DESTROY(frame);
 
     return 0;
@@ -387,7 +402,7 @@ afr_lock_fail_shared(afr_local_t *local, struct list_head *list)
 }
 
 static void
-afr_handle_lock_acquire_failure(afr_local_t *local, gf_boolean_t locked)
+afr_handle_lock_acquire_failure(afr_local_t *local)
 {
     struct list_head shared;
     afr_lock_t *lock = NULL;
@@ -408,13 +423,8 @@ afr_handle_lock_acquire_failure(afr_local_t *local, gf_boolean_t locked)
     afr_lock_fail_shared(local, &shared);
     local->transaction.do_eager_unlock = _gf_true;
 out:
-    if (locked) {
-        local->internal_lock.lock_cbk = afr_transaction_done;
-        afr_unlock(local->transaction.frame, local->transaction.frame->this);
-    } else {
-        afr_transaction_done(local->transaction.frame,
-                             local->transaction.frame->this);
-    }
+    local->internal_lock.lock_cbk = afr_transaction_done;
+    afr_unlock(local->transaction.frame, local->transaction.frame->this);
 }
 
 call_frame_t *
@@ -511,42 +521,6 @@ afr_compute_pre_op_sources(call_frame_t *frame, xlator_t *this)
                 local->transaction.pre_op_sources[j] = 0;
 }
 
-gf_boolean_t
-afr_has_arbiter_fop_cbk_quorum(call_frame_t *frame)
-{
-    afr_local_t *local = NULL;
-    afr_private_t *priv = NULL;
-    xlator_t *this = NULL;
-    gf_boolean_t fop_failed = _gf_false;
-    unsigned char *pre_op_sources = NULL;
-    int i = 0;
-
-    local = frame->local;
-    this = frame->this;
-    priv = this->private;
-    pre_op_sources = local->transaction.pre_op_sources;
-
-    /* If the fop failed on the brick, it is not a source. */
-    for (i = 0; i < priv->child_count; i++)
-        if (local->transaction.failed_subvols[i])
-            pre_op_sources[i] = 0;
-
-    switch (AFR_COUNT(pre_op_sources, priv->child_count)) {
-        case 1:
-            if (pre_op_sources[ARBITER_BRICK_INDEX])
-                fop_failed = _gf_true;
-            break;
-        case 0:
-            fop_failed = _gf_true;
-            break;
-    }
-
-    if (fop_failed)
-        return _gf_false;
-
-    return _gf_true;
-}
-
 void
 afr_txn_arbitrate_fop(call_frame_t *frame, xlator_t *this)
 {
@@ -613,7 +587,7 @@ afr_transaction_perform_fop(call_frame_t *frame, xlator_t *this)
     failure_count = AFR_COUNT(local->transaction.failed_subvols,
                               priv->child_count);
     if (failure_count == priv->child_count) {
-        afr_handle_lock_acquire_failure(local, _gf_true);
+        afr_handle_lock_acquire_failure(local);
         return 0;
     } else {
         lock = &local->inode_ctx->lock[local->transaction.type];
@@ -660,8 +634,9 @@ afr_set_pending_dict(afr_private_t *priv, dict_t *xattr, int **pending)
 
     return ret;
 }
+
 static void
-afr_ta_dom_lock_check_and_release(afr_local_t *local, xlator_t *this)
+afr_ta_dom_lock_check_and_release(afr_ta_fop_state_t fop_state, xlator_t *this)
 {
     afr_private_t *priv = this->private;
     unsigned int inmem_count = 0;
@@ -671,17 +646,25 @@ afr_ta_dom_lock_check_and_release(afr_local_t *local, xlator_t *this)
     LOCK(&priv->lock);
     {
         /*Once we get notify lock release upcall notification,
-         if two fop states are non empty/non zero, we will
-         not release lock.
-         1 - If anything in memory txn
-         2 - If anything in onwire or onwireq
+         if any of the fop state counters are non-zero, we will
+         not release the lock.
          */
-        if (local->fop_state == TA_INFO_IN_MEMORY_SUCCESS) {
-            inmem_count = --priv->ta_in_mem_txn_count;
-        } else {
-            inmem_count = priv->ta_in_mem_txn_count;
-        }
         onwire_count = priv->ta_on_wire_txn_count;
+        inmem_count = priv->ta_in_mem_txn_count;
+        switch (fop_state) {
+            case TA_GET_INFO_FROM_TA_FILE:
+                onwire_count = --priv->ta_on_wire_txn_count;
+                break;
+            case TA_INFO_IN_MEMORY_SUCCESS:
+            case TA_INFO_IN_MEMORY_FAILED:
+                inmem_count = --priv->ta_in_mem_txn_count;
+                break;
+            case TA_WAIT_FOR_NOTIFY_LOCK_REL:
+                GF_ASSERT(0);
+                break;
+            case TA_SUCCESS:
+                break;
+        }
         release = priv->release_ta_notify_dom_lock;
     }
     UNLOCK(&priv->lock);
@@ -693,7 +676,7 @@ afr_ta_dom_lock_check_and_release(afr_local_t *local, xlator_t *this)
 }
 
 static void
-afr_ta_process_onwireq(afr_local_t *local, xlator_t *this)
+afr_ta_process_onwireq(afr_ta_fop_state_t fop_state, xlator_t *this)
 {
     afr_private_t *priv = this->private;
     afr_local_t *entry = NULL;
@@ -706,15 +689,6 @@ afr_ta_process_onwireq(afr_local_t *local, xlator_t *this)
 
     LOCK(&priv->lock);
     {
-        if (--priv->ta_on_wire_txn_count == 0) {
-            UNLOCK(&priv->lock);
-            /*Only one write fop came and after taking notify
-             *lock and before doing xattrop, it has received
-             *lock contention upcall, so this is the only place
-             *to find this out and release the lock*/
-            afr_ta_dom_lock_check_and_release(local, this);
-            return;
-        }
         bad_child = priv->ta_bad_child_index;
         if (bad_child == AFR_CHILD_UNKNOWN) {
             /*The previous on-wire ta_post_op was a failure. Just dequeue
@@ -735,13 +709,10 @@ afr_ta_process_onwireq(afr_local_t *local, xlator_t *this)
         while (!list_empty(&onwireq)) {
             entry = list_entry(onwireq.next, afr_local_t, ta_onwireq);
             list_del_init(&entry->ta_onwireq);
-            LOCK(&priv->lock);
-            --priv->ta_on_wire_txn_count;
-            UNLOCK(&priv->lock);
             if (entry->ta_failed_subvol == bad_child) {
-                afr_changelog_post_op_do(entry->transaction.frame, this);
+                afr_post_op_handle_success(entry->transaction.frame, this);
             } else {
-                afr_changelog_post_op_fail(entry->transaction.frame, this, EIO);
+                afr_post_op_handle_failure(entry->transaction.frame, this, EIO);
             }
         }
     }
@@ -760,7 +731,7 @@ afr_changelog_post_op_done(call_frame_t *frame, xlator_t *this)
 
     if (priv->thin_arbiter_count) {
         /*fop should not come here with TA_WAIT_FOR_NOTIFY_LOCK_REL state */
-        afr_ta_dom_lock_check_and_release(frame->local, this);
+        afr_ta_dom_lock_check_and_release(local->fop_state, this);
     }
 
     /* Fail the FOP if post-op did not succeed on quorum no. of bricks. */
@@ -797,21 +768,9 @@ afr_changelog_post_op_fail(call_frame_t *frame, xlator_t *this, int op_errno)
 unsigned char *
 afr_locked_nodes_get(afr_transaction_type type, afr_internal_lock_t *int_lock)
 {
-    unsigned char *locked_nodes = NULL;
-    switch (type) {
-        case AFR_DATA_TRANSACTION:
-        case AFR_METADATA_TRANSACTION:
-            locked_nodes = int_lock->locked_nodes;
-            break;
-
-        case AFR_ENTRY_TRANSACTION:
-        case AFR_ENTRY_RENAME_TRANSACTION:
-            /*Because same set of subvols participate in all lockee
-             * entities*/
-            locked_nodes = int_lock->lockee[0].locked_nodes;
-            break;
-    }
-    return locked_nodes;
+    /*Because same set of subvols participate in all lockee
+     * entities*/
+    return int_lock->lockee[0].locked_nodes;
 }
 
 int
@@ -869,7 +828,7 @@ afr_handle_symmetric_errors(call_frame_t *frame, xlator_t *this)
 }
 
 gf_boolean_t
-afr_has_quorum(unsigned char *subvols, xlator_t *this)
+afr_has_quorum(unsigned char *subvols, xlator_t *this, call_frame_t *frame)
 {
     unsigned int quorum_count = 0;
     afr_private_t *priv = NULL;
@@ -878,6 +837,9 @@ afr_has_quorum(unsigned char *subvols, xlator_t *this)
     priv = this->private;
     up_children_count = AFR_COUNT(subvols, priv->child_count);
 
+    if (afr_lookup_has_quorum(frame, up_children_count))
+        return _gf_true;
+
     if (priv->quorum_count == AFR_QUORUM_AUTO) {
         /*
          * Special case for auto-quorum with an even number of nodes.
@@ -932,7 +894,7 @@ afr_has_fop_quorum(call_frame_t *frame)
 
     locked_nodes = afr_locked_nodes_get(local->transaction.type,
                                         &local->internal_lock);
-    return afr_has_quorum(locked_nodes, this);
+    return afr_has_quorum(locked_nodes, this, NULL);
 }
 
 static gf_boolean_t
@@ -950,7 +912,7 @@ afr_has_fop_cbk_quorum(call_frame_t *frame)
                 success[i] = 1;
     }
 
-    return afr_has_quorum(success, this);
+    return afr_has_quorum(success, this, NULL);
 }
 
 gf_boolean_t
@@ -973,12 +935,8 @@ afr_need_dirty_marking(call_frame_t *frame, xlator_t *this)
         priv->child_count)
         return _gf_false;
 
-    if (priv->arbiter_count) {
-        if (!afr_has_arbiter_fop_cbk_quorum(frame))
-            need_dirty = _gf_true;
-    } else if (!afr_has_fop_cbk_quorum(frame)) {
+    if (!afr_has_fop_cbk_quorum(frame))
         need_dirty = _gf_true;
-    }
 
     return need_dirty;
 }
@@ -1028,12 +986,8 @@ afr_handle_quorum(call_frame_t *frame, xlator_t *this)
      * no split-brain with the fix. The problem is eliminated completely.
      */
 
-    if (priv->arbiter_count) {
-        if (afr_has_arbiter_fop_cbk_quorum(frame))
-            return;
-    } else if (afr_has_fop_cbk_quorum(frame)) {
+    if (afr_has_fop_cbk_quorum(frame))
         return;
-    }
 
     if (afr_need_dirty_marking(frame, this))
         goto set_response;
@@ -1075,7 +1029,7 @@ set_response:
 }
 
 int
-afr_fill_ta_loc(xlator_t *this, loc_t *loc)
+afr_fill_ta_loc(xlator_t *this, loc_t *loc, gf_boolean_t is_gfid_based_fop)
 {
     afr_private_t *priv = NULL;
 
@@ -1083,6 +1037,11 @@ afr_fill_ta_loc(xlator_t *this, loc_t *loc)
     loc->parent = inode_ref(priv->root_inode);
     gf_uuid_copy(loc->pargfid, loc->parent->gfid);
     loc->name = priv->pending_key[THIN_ARBITER_BRICK_INDEX];
+    if (is_gfid_based_fop && gf_uuid_is_null(priv->ta_gfid)) {
+        /* Except afr_ta_id_file_check() which is path based, all other gluster
+         * FOPS need gfid.*/
+        return -EINVAL;
+    }
     gf_uuid_copy(loc->gfid, priv->ta_gfid);
     loc->inode = inode_new(loc->parent->table);
     if (!loc->inode) {
@@ -1092,99 +1051,76 @@ afr_fill_ta_loc(xlator_t *this, loc_t *loc)
     return 0;
 }
 
-int
-afr_changelog_thin_arbiter_post_op(xlator_t *this, afr_local_t *local)
+static int
+afr_ta_post_op_done(int ret, call_frame_t *frame, void *opaque)
 {
-    int ret = 0;
-    afr_private_t *priv = NULL;
-    dict_t *xattr = NULL;
-    int failed_count = 0;
-    struct gf_flock flock = {
-        0,
-    };
-    loc_t loc = {
-        0,
-    };
-    int i = 0;
-
-    priv = this->private;
-    if (!priv->thin_arbiter_count)
-        return 0;
-
-    failed_count = AFR_COUNT(local->transaction.failed_subvols,
-                             priv->child_count);
-    if (!failed_count)
-        return 0;
+    xlator_t *this = NULL;
+    afr_local_t *local = NULL;
+    call_frame_t *txn_frame = NULL;
+    afr_ta_fop_state_t fop_state;
 
-    GF_ASSERT(failed_count == 1);
-    ret = afr_fill_ta_loc(this, &loc);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
-               "Failed to populate thin-arbiter loc for: %s.", loc.name);
-        goto out;
-    }
+    local = (afr_local_t *)opaque;
+    fop_state = local->fop_state;
+    txn_frame = local->transaction.frame;
+    this = frame->this;
 
-    xattr = dict_new();
-    if (!xattr) {
-        ret = -ENOMEM;
-        goto out;
-    }
-    for (i = 0; i < priv->child_count; i++) {
-        ret = dict_set_static_bin(xattr, priv->pending_key[i],
-                                  local->pending[i],
-                                  AFR_NUM_CHANGE_LOGS * sizeof(int));
-        if (ret)
-            goto out;
+    if (ret == 0) {
+        /*Mark pending xattrs on the up data brick.*/
+        afr_post_op_handle_success(txn_frame, this);
+    } else {
+        afr_post_op_handle_failure(txn_frame, this, -ret);
     }
 
-    flock.l_type = F_WRLCK;
-    flock.l_start = 0;
-    flock.l_len = 0;
-
-    /*TODO: Convert to two domain locking. */
-    ret = syncop_inodelk(priv->children[THIN_ARBITER_BRICK_INDEX],
-                         AFR_TA_DOM_NOTIFY, &loc, F_SETLKW, &flock, NULL, NULL);
-    if (ret)
-        goto out;
+    STACK_DESTROY(frame->root);
+    afr_ta_process_onwireq(fop_state, this);
 
-    ret = syncop_xattrop(priv->children[THIN_ARBITER_BRICK_INDEX], &loc,
-                         GF_XATTROP_ADD_ARRAY, xattr, NULL, NULL, NULL);
+    return 0;
+}
 
-    if (ret == -EINVAL) {
-        gf_msg(this->name, GF_LOG_INFO, -ret, AFR_MSG_THIN_ARB,
-               "Thin-arbiter has denied post-op on %s for gfid %s.",
-               priv->pending_key[THIN_ARBITER_BRICK_INDEX],
-               uuid_utoa(local->inode->gfid));
+int **
+afr_set_changelog_xattr(afr_private_t *priv, unsigned char *pending,
+                        dict_t *xattr, afr_local_t *local)
+{
+    int **changelog = NULL;
+    int idx = 0;
+    int ret = 0;
+    int i;
 
-    } else if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
-               "Post-op on thin-arbiter id file %s failed for gfid %s.",
-               priv->pending_key[THIN_ARBITER_BRICK_INDEX],
-               uuid_utoa(local->inode->gfid));
+    if (local->is_new_entry == _gf_true) {
+        changelog = afr_mark_pending_changelog(priv, pending, xattr,
+                                               local->cont.dir_fop.buf.ia_type);
+    } else {
+        idx = afr_index_for_transaction_type(local->transaction.type);
+        changelog = afr_matrix_create(priv->child_count, AFR_NUM_CHANGE_LOGS);
+        if (!changelog) {
+            goto out;
+        }
+        for (i = 0; i < priv->child_count; i++) {
+            if (local->transaction.failed_subvols[i])
+                changelog[i][idx] = hton32(1);
+        }
+        ret = afr_set_pending_dict(priv, xattr, changelog);
+        if (ret < 0) {
+            afr_matrix_cleanup(changelog, priv->child_count);
+            return NULL;
+        }
     }
-    flock.l_type = F_UNLCK;
-    syncop_inodelk(priv->children[THIN_ARBITER_BRICK_INDEX], AFR_TA_DOM_NOTIFY,
-                   &loc, F_SETLK, &flock, NULL, NULL);
-out:
-    if (xattr)
-        dict_unref(xattr);
 
-    return ret;
+out:
+    return changelog;
 }
 
-static int
-afr_ta_post_op_done(int ret, call_frame_t *frame, void *opaque)
+static void
+afr_ta_locked_xattrop_validate(afr_private_t *priv, afr_local_t *local,
+                               gf_boolean_t *valid)
 {
-    xlator_t *this = NULL;
-    afr_local_t *local = NULL;
-
-    local = (afr_local_t *)opaque;
-    this = frame->this;
-
-    STACK_DESTROY(frame->root);
-    afr_ta_process_onwireq(local, this);
-
-    return 0;
+    if (priv->ta_event_gen > local->ta_event_gen) {
+        /* We can't trust the ta's response anymore.*/
+        afr_ta_locked_priv_invalidate(priv);
+        *valid = _gf_false;
+        return;
+    }
+    return;
 }
 
 static int
@@ -1193,27 +1129,25 @@ afr_ta_post_op_do(void *opaque)
     afr_local_t *local = NULL;
     afr_private_t *priv = NULL;
     xlator_t *this = NULL;
-    call_frame_t *txn_frame = NULL;
     dict_t *xattr = NULL;
-    int **pending = NULL;
+    unsigned char *pending = NULL;
+    int **changelog = NULL;
     int failed_subvol = -1;
     int success_subvol = -1;
     loc_t loc = {
         0,
     };
-    int idx = 0;
     int i = 0;
     int ret = 0;
+    gf_boolean_t valid = _gf_true;
 
     local = (afr_local_t *)opaque;
-    txn_frame = local->transaction.frame;
-    this = txn_frame->this;
+    this = local->transaction.frame->this;
     priv = this->private;
-    idx = afr_index_for_transaction_type(local->transaction.type);
 
-    ret = afr_fill_ta_loc(this, &loc);
+    ret = afr_fill_ta_loc(this, &loc, _gf_true);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, AFR_MSG_THIN_ARB,
+        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
                "Failed to populate loc for thin-arbiter.");
         goto out;
     }
@@ -1224,23 +1158,23 @@ afr_ta_post_op_do(void *opaque)
         goto out;
     }
 
-    pending = afr_matrix_create(priv->child_count, AFR_NUM_CHANGE_LOGS);
-    if (!pending) {
-        ret = -ENOMEM;
-        goto out;
-    }
+    pending = alloca0(priv->child_count);
+
     for (i = 0; i < priv->child_count; i++) {
         if (local->transaction.failed_subvols[i]) {
-            pending[i][idx] = hton32(1);
+            pending[i] = 1;
             failed_subvol = i;
         } else {
             success_subvol = i;
         }
     }
 
-    ret = afr_set_pending_dict(priv, xattr, pending);
-    if (ret < 0)
+    changelog = afr_set_changelog_xattr(priv, pending, xattr, local);
+
+    if (!changelog) {
+        ret = -ENOMEM;
         goto out;
+    }
 
     ret = afr_ta_post_op_lock(this, &loc);
     if (ret)
@@ -1248,22 +1182,31 @@ afr_ta_post_op_do(void *opaque)
 
     ret = syncop_xattrop(priv->children[THIN_ARBITER_BRICK_INDEX], &loc,
                          GF_XATTROP_ADD_ARRAY, xattr, NULL, NULL, NULL);
+    if (ret) {
+        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
+               "Post-op on thin-arbiter id file %s failed for gfid %s.",
+               priv->pending_key[THIN_ARBITER_BRICK_INDEX],
+               uuid_utoa(local->inode->gfid));
+    }
     LOCK(&priv->lock);
     {
         if (ret == 0) {
             priv->ta_bad_child_index = failed_subvol;
         } else if (ret == -EINVAL) {
             priv->ta_bad_child_index = success_subvol;
+            ret = -EIO; /* TA failed the fop. Return EIO to application. */
         }
+
+        afr_ta_locked_xattrop_validate(priv, local, &valid);
     }
     UNLOCK(&priv->lock);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, AFR_MSG_THIN_ARB,
-               "Post-op on thin-arbiter id file %s failed for gfid %s.",
+    if (valid == _gf_false) {
+        gf_msg(this->name, GF_LOG_ERROR, EIO, AFR_MSG_THIN_ARB,
+               "Post-op on thin-arbiter id file %s for gfid %s invalidated due "
+               "to event-gen mismatch.",
                priv->pending_key[THIN_ARBITER_BRICK_INDEX],
                uuid_utoa(local->inode->gfid));
-        if (ret == -EINVAL)
-            ret = -EIO; /* TA failed the fop. Return EIO to application. */
+        ret = -EIO;
     }
 
     afr_ta_post_op_unlock(this, &loc);
@@ -1271,17 +1214,11 @@ out:
     if (xattr)
         dict_unref(xattr);
 
-    if (pending)
-        afr_matrix_cleanup(pending, priv->child_count);
+    if (changelog)
+        afr_matrix_cleanup(changelog, priv->child_count);
 
     loc_wipe(&loc);
 
-    if (ret == 0) {
-        /*Mark pending xattrs on the up data brick.*/
-        afr_changelog_post_op_do(local->transaction.frame, this);
-    } else {
-        afr_changelog_post_op_fail(local->transaction.frame, this, -ret);
-    }
     return ret;
 }
 
@@ -1340,6 +1277,7 @@ afr_ta_set_fop_state(afr_private_t *priv, afr_local_t *local,
             /* Post-op on TA not needed as the fop succeeded only on the
              * in-memory bad data brick and not the good one. Fail the fop.*/
             local->fop_state = TA_INFO_IN_MEMORY_FAILED;
+            priv->ta_in_mem_txn_count++;
         }
     }
     UNLOCK(&priv->lock);
@@ -1359,6 +1297,28 @@ afr_ta_fill_failed_subvol(afr_private_t *priv, afr_local_t *local)
 }
 
 static void
+afr_post_op_handle_success(call_frame_t *frame, xlator_t *this)
+{
+    afr_local_t *local = NULL;
+
+    local = frame->local;
+    if (local->is_new_entry == _gf_true) {
+        afr_mark_new_entry_changelog(frame, this);
+    }
+    afr_changelog_post_op_do(frame, this);
+
+    return;
+}
+
+static void
+afr_post_op_handle_failure(call_frame_t *frame, xlator_t *this, int op_errno)
+{
+    afr_changelog_post_op_fail(frame, this, op_errno);
+
+    return;
+}
+
+static void
 afr_ta_decide_post_op_state(call_frame_t *frame, xlator_t *this)
 {
     afr_private_t *priv = NULL;
@@ -1380,10 +1340,12 @@ afr_ta_decide_post_op_state(call_frame_t *frame, xlator_t *this)
             /*Post releasing the notify lock, we will act on this queue*/
             break;
         case TA_INFO_IN_MEMORY_SUCCESS:
-            afr_changelog_post_op_do(frame, this);
+            afr_post_op_handle_success(frame, this);
             break;
         case TA_INFO_IN_MEMORY_FAILED:
-            afr_changelog_post_op_fail(frame, this, EIO);
+            afr_post_op_handle_failure(frame, this, EIO);
+            break;
+        default:
             break;
     }
     return;
@@ -1713,6 +1675,7 @@ afr_changelog_populate_xdata(call_frame_t *frame, afr_xattrop_type_t op,
     int i = 0;
     int ret = 0;
     char *key = NULL;
+    int keylen = 0;
     const char *name = NULL;
     dict_t *xdata1 = NULL;
     dict_t *xdata2 = NULL;
@@ -1770,7 +1733,8 @@ afr_changelog_populate_xdata(call_frame_t *frame, afr_xattrop_type_t op,
     }
 
     if (need_entry_key_set) {
-        ret = dict_set_str(xdata1, key, (char *)name);
+        keylen = strlen(key);
+        ret = dict_set_strn(xdata1, key, keylen, (char *)name);
         if (ret)
             gf_msg(THIS->name, GF_LOG_ERROR, 0, AFR_MSG_DICT_SET_FAILED,
                    "%s/%s: Could not set %s key during xattrop",
@@ -1780,7 +1744,8 @@ afr_changelog_populate_xdata(call_frame_t *frame, afr_xattrop_type_t op,
             if (!xdata2)
                 goto out;
 
-            ret = dict_set_str(xdata2, key, (char *)local->newloc.name);
+            ret = dict_set_strn(xdata2, key, keylen,
+                                (char *)local->newloc.name);
             if (ret)
                 gf_msg(THIS->name, GF_LOG_ERROR, 0, AFR_MSG_DICT_SET_FAILED,
                        "%s/%s: Could not set %s key during "
@@ -2042,7 +2007,7 @@ err:
     local->op_ret = -1;
     local->op_errno = op_errno;
 
-    afr_handle_lock_acquire_failure(local, _gf_true);
+    afr_handle_lock_acquire_failure(local);
 
     if (xdata_req)
         dict_unref(xdata_req);
@@ -2051,7 +2016,7 @@ err:
 }
 
 int
-afr_post_nonblocking_inodelk_cbk(call_frame_t *frame, xlator_t *this)
+afr_post_nonblocking_lock_cbk(call_frame_t *frame, xlator_t *this)
 {
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
@@ -2062,36 +2027,12 @@ afr_post_nonblocking_inodelk_cbk(call_frame_t *frame, xlator_t *this)
     /* Initiate blocking locks if non-blocking has failed */
     if (int_lock->lock_op_ret < 0) {
         gf_msg_debug(this->name, 0,
-                     "Non blocking inodelks failed. Proceeding to blocking");
+                     "Non blocking locks failed. Proceeding to blocking");
         int_lock->lock_cbk = afr_internal_lock_finish;
         afr_blocking_lock(frame, this);
     } else {
         gf_msg_debug(this->name, 0,
-                     "Non blocking inodelks done. Proceeding to FOP");
-        afr_internal_lock_finish(frame, this);
-    }
-
-    return 0;
-}
-
-int
-afr_post_nonblocking_entrylk_cbk(call_frame_t *frame, xlator_t *this)
-{
-    afr_internal_lock_t *int_lock = NULL;
-    afr_local_t *local = NULL;
-
-    local = frame->local;
-    int_lock = &local->internal_lock;
-
-    /* Initiate blocking locks if non-blocking has failed */
-    if (int_lock->lock_op_ret < 0) {
-        gf_msg_debug(this->name, 0,
-                     "Non blocking entrylks failed. Proceeding to blocking");
-        int_lock->lock_cbk = afr_internal_lock_finish;
-        afr_blocking_lock(frame, this);
-    } else {
-        gf_msg_debug(this->name, 0,
-                     "Non blocking entrylks done. Proceeding to FOP");
+                     "Non blocking locks done. Proceeding to FOP");
 
         afr_internal_lock_finish(frame, this);
     }
@@ -2109,7 +2050,7 @@ afr_post_blocking_rename_cbk(call_frame_t *frame, xlator_t *this)
     int_lock = &local->internal_lock;
 
     if (int_lock->lock_op_ret < 0) {
-        gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_BLOCKING_LKS_FAILED,
+        gf_msg(this->name, GF_LOG_INFO, 0, AFR_MSG_INTERNAL_LKS_FAILED,
                "Blocking entrylks failed.");
 
         afr_transaction_done(frame, this);
@@ -2140,25 +2081,26 @@ afr_post_lower_unlock_cbk(call_frame_t *frame, xlator_t *this)
 }
 
 int
-afr_set_transaction_flock(xlator_t *this, afr_local_t *local)
+afr_set_transaction_flock(xlator_t *this, afr_local_t *local,
+                          afr_lockee_t *lockee)
 {
-    afr_internal_lock_t *int_lock = NULL;
     afr_private_t *priv = NULL;
+    struct gf_flock *flock = NULL;
 
-    int_lock = &local->internal_lock;
     priv = this->private;
+    flock = &lockee->flock;
 
     if ((priv->arbiter_count || local->transaction.eager_lock_on ||
          priv->full_lock) &&
         local->transaction.type == AFR_DATA_TRANSACTION) {
         /*Lock entire file to avoid network split brains.*/
-        int_lock->flock.l_len = 0;
-        int_lock->flock.l_start = 0;
+        flock->l_len = 0;
+        flock->l_start = 0;
     } else {
-        int_lock->flock.l_len = local->transaction.len;
-        int_lock->flock.l_start = local->transaction.start;
+        flock->l_len = local->transaction.len;
+        flock->l_start = local->transaction.start;
     }
-    int_lock->flock.l_type = F_WRLCK;
+    flock->l_type = F_WRLCK;
 
     return 0;
 }
@@ -2168,26 +2110,21 @@ afr_lock(call_frame_t *frame, xlator_t *this)
 {
     afr_internal_lock_t *int_lock = NULL;
     afr_local_t *local = NULL;
+    int i = 0;
 
     local = frame->local;
     int_lock = &local->internal_lock;
 
+    int_lock->lock_cbk = afr_post_nonblocking_lock_cbk;
     int_lock->domain = this->name;
 
     switch (local->transaction.type) {
         case AFR_DATA_TRANSACTION:
         case AFR_METADATA_TRANSACTION:
-            afr_set_transaction_flock(this, local);
-
-            int_lock->lock_cbk = afr_post_nonblocking_inodelk_cbk;
-
-            afr_nonblocking_inodelk(frame, this);
-            break;
-
-        case AFR_ENTRY_RENAME_TRANSACTION:
+            for (i = 0; i < int_lock->lockee_count; i++) {
+                afr_set_transaction_flock(this, local, &int_lock->lockee[i]);
+            }
 
-            int_lock->lock_cbk = afr_post_nonblocking_entrylk_cbk;
-            afr_nonblocking_entrylk(frame, this);
             break;
 
         case AFR_ENTRY_TRANSACTION:
@@ -2196,11 +2133,11 @@ afr_lock(call_frame_t *frame, xlator_t *this)
                 int_lock->lk_loc = &local->transaction.parent_loc;
             else
                 GF_ASSERT(local->fd);
-
-            int_lock->lock_cbk = afr_post_nonblocking_entrylk_cbk;
-            afr_nonblocking_entrylk(frame, this);
+            break;
+        case AFR_ENTRY_RENAME_TRANSACTION:
             break;
     }
+    afr_lock_nonblocking(frame, this);
 
     return 0;
 }
@@ -2271,17 +2208,19 @@ afr_has_lock_conflict(afr_local_t *local, gf_boolean_t waitlist_check)
 /* }}} */
 static void
 afr_copy_inodelk_vars(afr_internal_lock_t *dst, afr_internal_lock_t *src,
-                      xlator_t *this)
+                      xlator_t *this, int lockee_num)
 {
     afr_private_t *priv = this->private;
+    afr_lockee_t *sl = &src->lockee[lockee_num];
+    afr_lockee_t *dl = &dst->lockee[lockee_num];
 
     dst->domain = src->domain;
-    dst->flock.l_len = src->flock.l_len;
-    dst->flock.l_start = src->flock.l_start;
-    dst->flock.l_type = src->flock.l_type;
-    dst->lock_count = src->lock_count;
-    memcpy(dst->locked_nodes, src->locked_nodes,
-           priv->child_count * sizeof(*dst->locked_nodes));
+    dl->flock.l_len = sl->flock.l_len;
+    dl->flock.l_start = sl->flock.l_start;
+    dl->flock.l_type = sl->flock.l_type;
+    dl->locked_count = sl->locked_count;
+    memcpy(dl->locked_nodes, sl->locked_nodes,
+           priv->child_count * sizeof(*dl->locked_nodes));
 }
 
 void
@@ -2302,7 +2241,7 @@ __afr_transaction_wake_shared(afr_local_t *local, struct list_head *shared)
         if (conflict && !list_empty(&lock->owners))
             return;
         afr_copy_inodelk_vars(&each->internal_lock, &local->internal_lock,
-                              each->transaction.frame->this);
+                              each->transaction.frame->this, 0);
         list_move_tail(&each->transaction.wait_list, shared);
         list_add_tail(&each->transaction.owner_list, &lock->owners);
     }
@@ -2337,7 +2276,7 @@ afr_internal_lock_finish(call_frame_t *frame, xlator_t *this)
     } else {
         lock = &local->inode_ctx->lock[local->transaction.type];
         if (local->internal_lock.lock_op_ret < 0) {
-            afr_handle_lock_acquire_failure(local, _gf_false);
+            afr_handle_lock_acquire_failure(local);
         } else {
             lock->event_generation = local->event_generation;
             afr_changelog_pre_op(frame, this);
@@ -2408,8 +2347,13 @@ afr_is_delayed_changelog_post_op_needed(call_frame_t *frame, xlator_t *this,
         goto out;
     }
 
-    if ((local->op != GF_FOP_WRITE) && (local->op != GF_FOP_FXATTROP)) {
-        /*Only allow writes but shard does [f]xattrops on writes, so
+    if (local->transaction.disable_delayed_post_op) {
+        goto out;
+    }
+
+    if ((local->op != GF_FOP_WRITE) && (local->op != GF_FOP_FXATTROP) &&
+        (local->op != GF_FOP_FSYNC)) {
+        /*Only allow writes/fsyncs but shard does [f]xattrops on writes, so
          * they are fine too*/
         goto out;
     }
@@ -2536,8 +2480,10 @@ afr_changelog_fsync(call_frame_t *frame, xlator_t *this)
     local->call_count = call_count;
 
     xdata = dict_new();
-    if (xdata)
-        ret = dict_set_int32(xdata, "batch-fsync", 1);
+    if (xdata) {
+        ret = dict_set_int32_sizen(xdata, "batch-fsync", 1);
+        ret = dict_set_str(xdata, GLUSTERFS_INTERNAL_FOP_KEY, "yes");
+    }
 
     for (i = 0; i < priv->child_count; i++) {
         if (!local->transaction.pre_op[i])
@@ -2779,7 +2725,7 @@ __afr_eager_lock_handle(afr_local_t *local, gf_boolean_t *take_lock,
             *timer_local = list_entry(lock->post_op.next, afr_local_t,
                                       transaction.owner_list);
             afr_copy_inodelk_vars(&local->internal_lock,
-                                  &(*timer_local)->internal_lock, this);
+                                  &(*timer_local)->internal_lock, this, 0);
             lock->delay_timer = NULL;
             *do_pre_op = _gf_true;
             list_add_tail(&local->transaction.owner_list, &lock->owners);
@@ -2796,7 +2742,7 @@ __afr_eager_lock_handle(afr_local_t *local, gf_boolean_t *take_lock,
         owner_local = list_entry(lock->owners.next, afr_local_t,
                                  transaction.owner_list);
         afr_copy_inodelk_vars(&local->internal_lock,
-                              &owner_local->internal_lock, this);
+                              &owner_local->internal_lock, this, 0);
         *take_lock = _gf_false;
         *do_pre_op = _gf_true;
     }
@@ -2866,6 +2812,62 @@ fail:
 }
 
 int
+afr_transaction_lockee_init(call_frame_t *frame)
+{
+    afr_local_t *local = frame->local;
+    afr_internal_lock_t *int_lock = &local->internal_lock;
+    afr_private_t *priv = frame->this->private;
+    int ret = 0;
+
+    switch (local->transaction.type) {
+        case AFR_DATA_TRANSACTION:
+        case AFR_METADATA_TRANSACTION:
+            ret = afr_add_inode_lockee(local, priv->child_count);
+            break;
+
+        case AFR_ENTRY_TRANSACTION:
+        case AFR_ENTRY_RENAME_TRANSACTION:
+            ret = afr_add_entry_lockee(local, &local->transaction.parent_loc,
+                                       local->transaction.basename,
+                                       priv->child_count);
+            if (ret) {
+                goto out;
+            }
+            if (local->op == GF_FOP_RENAME) {
+                ret = afr_add_entry_lockee(
+                    local, &local->transaction.new_parent_loc,
+                    local->transaction.new_basename, priv->child_count);
+                if (ret) {
+                    goto out;
+                }
+
+                if (local->newloc.inode &&
+                    IA_ISDIR(local->newloc.inode->ia_type)) {
+                    ret = afr_add_entry_lockee(local, &local->newloc, NULL,
+                                               priv->child_count);
+                    if (ret) {
+                        goto out;
+                    }
+                }
+            } else if (local->op == GF_FOP_RMDIR) {
+                ret = afr_add_entry_lockee(local, &local->loc, NULL,
+                                           priv->child_count);
+                if (ret) {
+                    goto out;
+                }
+            }
+
+            if (int_lock->lockee_count > 1) {
+                qsort(int_lock->lockee, int_lock->lockee_count,
+                      sizeof(*int_lock->lockee), afr_entry_lockee_cmp);
+            }
+            break;
+    }
+out:
+    return ret;
+}
+
+int
 afr_transaction(call_frame_t *frame, xlator_t *this, afr_transaction_type type)
 {
     afr_local_t *local = NULL;
@@ -2879,7 +2881,7 @@ afr_transaction(call_frame_t *frame, xlator_t *this, afr_transaction_type type)
 
     local->transaction.type = type;
 
-    if (priv->quorum_count && !afr_has_quorum(local->child_up, this)) {
+    if (priv->quorum_count && !afr_has_quorum(local->child_up, this, NULL)) {
         ret = -afr_quorum_errno(priv);
         goto out;
     }
@@ -2898,6 +2900,10 @@ afr_transaction(call_frame_t *frame, xlator_t *this, afr_transaction_type type)
     if (ret < 0)
         goto out;
 
+    ret = afr_transaction_lockee_init(frame);
+    if (ret)
+        goto out;
+
     if (type != AFR_METADATA_TRANSACTION) {
         goto txn_start;
     }
diff --git a/xlators/cluster/afr/src/afr-transaction.h b/xlators/cluster/afr/src/afr-transaction.h
index 35a922544bc..beefa26f4a6 100644
--- a/xlators/cluster/afr/src/afr-transaction.h
+++ b/xlators/cluster/afr/src/afr-transaction.h
@@ -48,7 +48,7 @@ afr_pending_read_decrement(afr_private_t *priv, int child_index);
 call_frame_t *
 afr_transaction_detach_fop_frame(call_frame_t *frame);
 gf_boolean_t
-afr_has_quorum(unsigned char *subvols, xlator_t *this);
+afr_has_quorum(unsigned char *subvols, xlator_t *this, call_frame_t *frame);
 gf_boolean_t
 afr_needs_changelog_update(afr_local_t *local);
 void
diff --git a/xlators/cluster/afr/src/afr.c b/xlators/cluster/afr/src/afr.c
index 5d5e536ff60..df7366f0a65 100644
--- a/xlators/cluster/afr/src/afr.c
+++ b/xlators/cluster/afr/src/afr.c
@@ -87,7 +87,7 @@ static void
 fix_quorum_options(xlator_t *this, afr_private_t *priv, char *qtype,
                    dict_t *options)
 {
-    if (dict_get(options, "quorum-type") == NULL) {
+    if (dict_get_sizen(options, "quorum-type") == NULL) {
         /* If user doesn't configure anything enable auto-quorum if the
          * replica has more than two subvolumes */
         if (priv->child_count > 2)
@@ -120,23 +120,62 @@ afr_set_favorite_child_policy(afr_private_t *priv, char *policy)
 
     return 0;
 }
+
+static void
+set_data_self_heal_algorithm(afr_private_t *priv, char *algo)
+{
+    if (!algo) {
+        priv->data_self_heal_algorithm = AFR_SELFHEAL_DATA_DYNAMIC;
+    } else if (strcmp(algo, "full") == 0) {
+        priv->data_self_heal_algorithm = AFR_SELFHEAL_DATA_FULL;
+    } else if (strcmp(algo, "diff") == 0) {
+        priv->data_self_heal_algorithm = AFR_SELFHEAL_DATA_DIFF;
+    } else {
+        priv->data_self_heal_algorithm = AFR_SELFHEAL_DATA_DYNAMIC;
+    }
+}
+
+void
+afr_handle_anon_inode_options(afr_private_t *priv, dict_t *options)
+{
+    char *volfile_id_str = NULL;
+    uuid_t anon_inode_gfid = {0};
+
+    /*If volume id is not present don't enable anything*/
+    if (dict_get_str(options, "volume-id", &volfile_id_str))
+        return;
+    GF_ASSERT(strlen(AFR_ANON_DIR_PREFIX) + strlen(volfile_id_str) <= NAME_MAX);
+    /*anon_inode_name is not supposed to change once assigned*/
+    if (!priv->anon_inode_name[0]) {
+        snprintf(priv->anon_inode_name, sizeof(priv->anon_inode_name), "%s-%s",
+                 AFR_ANON_DIR_PREFIX, volfile_id_str);
+        gf_uuid_parse(volfile_id_str, anon_inode_gfid);
+        /*Flip a bit to make sure volfile-id and anon-gfid are not same*/
+        anon_inode_gfid[0] ^= 1;
+        uuid_utoa_r(anon_inode_gfid, priv->anon_gfid_str);
+    }
+}
+
 int
 reconfigure(xlator_t *this, dict_t *options)
 {
     afr_private_t *priv = NULL;
     xlator_t *read_subvol = NULL;
     int read_subvol_index = -1;
+    int timeout_old = 0;
     int ret = -1;
     int index = -1;
     char *qtype = NULL;
     char *fav_child_policy = NULL;
+    char *data_self_heal = NULL;
+    char *data_self_heal_algorithm = NULL;
+    char *locking_scheme = NULL;
     gf_boolean_t consistent_io = _gf_false;
     gf_boolean_t choose_local_old = _gf_false;
+    gf_boolean_t enabled_old = _gf_false;
 
     priv = this->private;
 
-    GF_OPTION_RECONF("afr-dirty-xattr", priv->afr_dirty, options, str, out);
-
     GF_OPTION_RECONF("metadata-splitbrain-forced-heal",
                      priv->metadata_splitbrain_forced_heal, options, bool, out);
 
@@ -149,7 +188,9 @@ reconfigure(xlator_t *this, dict_t *options)
     GF_OPTION_RECONF("metadata-self-heal", priv->metadata_self_heal, options,
                      bool, out);
 
-    GF_OPTION_RECONF("data-self-heal", priv->data_self_heal, options, str, out);
+    GF_OPTION_RECONF("data-self-heal", data_self_heal, options, str, out);
+    if (gf_string2boolean(data_self_heal, &priv->data_self_heal) == -1)
+        goto out;
 
     GF_OPTION_RECONF("entry-self-heal", priv->entry_self_heal, options, bool,
                      out);
@@ -157,8 +198,9 @@ reconfigure(xlator_t *this, dict_t *options)
     GF_OPTION_RECONF("data-self-heal-window-size",
                      priv->data_self_heal_window_size, options, uint32, out);
 
-    GF_OPTION_RECONF("data-self-heal-algorithm", priv->data_self_heal_algorithm,
+    GF_OPTION_RECONF("data-self-heal-algorithm", data_self_heal_algorithm,
                      options, str, out);
+    set_data_self_heal_algorithm(priv, data_self_heal_algorithm);
 
     GF_OPTION_RECONF("halo-enabled", priv->halo_enabled, options, bool, out);
 
@@ -214,16 +256,19 @@ reconfigure(xlator_t *this, dict_t *options)
     }
 
     GF_OPTION_RECONF("pre-op-compat", priv->pre_op_compat, options, bool, out);
-    GF_OPTION_RECONF("locking-scheme", priv->locking_scheme, options, str, out);
+    GF_OPTION_RECONF("locking-scheme", locking_scheme, options, str, out);
+    priv->granular_locks = (strcmp(locking_scheme, "granular") == 0);
     GF_OPTION_RECONF("full-lock", priv->full_lock, options, bool, out);
     GF_OPTION_RECONF("granular-entry-heal", priv->esh_granular, options, bool,
                      out);
 
     GF_OPTION_RECONF("eager-lock", priv->eager_lock, options, bool, out);
+    GF_OPTION_RECONF("optimistic-change-log", priv->optimistic_change_log,
+                     options, bool, out);
     GF_OPTION_RECONF("quorum-type", qtype, options, str, out);
     GF_OPTION_RECONF("quorum-count", priv->quorum_count, options, uint32, out);
     fix_quorum_options(this, priv, qtype, options);
-    if (priv->quorum_count && !afr_has_quorum(priv->child_up, this))
+    if (priv->quorum_count && !afr_has_quorum(priv->child_up, this, NULL))
         gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_QUORUM_FAIL,
                "Client-quorum is not met");
 
@@ -236,11 +281,13 @@ reconfigure(xlator_t *this, dict_t *options)
     GF_OPTION_RECONF("ensure-durability", priv->ensure_durability, options,
                      bool, out);
 
+    enabled_old = priv->shd.enabled;
     GF_OPTION_RECONF("self-heal-daemon", priv->shd.enabled, options, bool, out);
 
     GF_OPTION_RECONF("iam-self-heal-daemon", priv->shd.iamshd, options, bool,
                      out);
 
+    timeout_old = priv->shd.timeout;
     GF_OPTION_RECONF("heal-timeout", priv->shd.timeout, options, int32, out);
 
     GF_OPTION_RECONF("consistent-metadata", priv->consistent_metadata, options,
@@ -264,6 +311,16 @@ reconfigure(xlator_t *this, dict_t *options)
         consistent_io = _gf_false;
     priv->consistent_io = consistent_io;
 
+    afr_handle_anon_inode_options(priv, options);
+
+    GF_OPTION_RECONF("use-anonymous-inode", priv->use_anon_inode, options, bool,
+                     out);
+    if (priv->shd.enabled) {
+        if ((priv->shd.enabled != enabled_old) ||
+            (timeout_old != priv->shd.timeout))
+            afr_selfheal_childup(this, priv);
+    }
+
     ret = 0;
 out:
     return ret;
@@ -349,7 +406,7 @@ afr_ta_init(afr_private_t *priv)
     priv->release_ta_notify_dom_lock = _gf_false;
     INIT_LIST_HEAD(&priv->ta_waitq);
     INIT_LIST_HEAD(&priv->ta_onwireq);
-    *priv->ta_gfid = 0;
+    gf_uuid_clear(priv->ta_gfid);
 }
 
 int32_t
@@ -366,6 +423,9 @@ init(xlator_t *this)
     char *qtype = NULL;
     char *fav_child_policy = NULL;
     char *thin_arbiter = NULL;
+    char *data_self_heal = NULL;
+    char *locking_scheme = NULL;
+    char *data_self_heal_algorithm = NULL;
 
     if (!this->children) {
         gf_msg(this->name, GF_LOG_ERROR, 0, AFR_MSG_CHILD_MISCONFIGURED,
@@ -385,6 +445,8 @@ init(xlator_t *this)
         goto out;
 
     priv = this->private;
+    INIT_LIST_HEAD(&priv->saved_locks);
+    INIT_LIST_HEAD(&priv->lk_healq);
     LOCK_INIT(&priv->lock);
 
     child_count = xlator_subvolume_count(this);
@@ -448,10 +510,13 @@ init(xlator_t *this)
 
     GF_OPTION_INIT("heal-wait-queue-length", priv->heal_wait_qlen, uint32, out);
 
-    GF_OPTION_INIT("data-self-heal", priv->data_self_heal, str, out);
+    GF_OPTION_INIT("data-self-heal", data_self_heal, str, out);
+    if (gf_string2boolean(data_self_heal, &priv->data_self_heal) == -1)
+        goto out;
 
-    GF_OPTION_INIT("data-self-heal-algorithm", priv->data_self_heal_algorithm,
-                   str, out);
+    GF_OPTION_INIT("data-self-heal-algorithm", data_self_heal_algorithm, str,
+                   out);
+    set_data_self_heal_algorithm(priv, data_self_heal_algorithm);
 
     GF_OPTION_INIT("data-self-heal-window-size",
                    priv->data_self_heal_window_size, uint32, out);
@@ -479,7 +544,8 @@ init(xlator_t *this)
                    out);
 
     GF_OPTION_INIT("pre-op-compat", priv->pre_op_compat, bool, out);
-    GF_OPTION_INIT("locking-scheme", priv->locking_scheme, str, out);
+    GF_OPTION_INIT("locking-scheme", locking_scheme, str, out);
+    priv->granular_locks = (strcmp(locking_scheme, "granular") == 0);
     GF_OPTION_INIT("full-lock", priv->full_lock, bool, out);
     GF_OPTION_INIT("granular-entry-heal", priv->esh_granular, bool, out);
 
@@ -500,7 +566,9 @@ init(xlator_t *this)
 
     GF_OPTION_INIT("consistent-metadata", priv->consistent_metadata, bool, out);
     GF_OPTION_INIT("consistent-io", priv->consistent_io, bool, out);
+    afr_handle_anon_inode_options(priv, this->options);
 
+    GF_OPTION_INIT("use-anonymous-inode", priv->use_anon_inode, bool, out);
     if (priv->quorum_count != 0)
         priv->consistent_io = _gf_false;
 
@@ -512,13 +580,19 @@ init(xlator_t *this)
         goto out;
     }
 
+    priv->anon_inode = GF_CALLOC(sizeof(unsigned char), child_count,
+                                 gf_afr_mt_char);
+
     priv->child_up = GF_CALLOC(sizeof(unsigned char), child_count,
                                gf_afr_mt_char);
 
     priv->child_latency = GF_MALLOC(sizeof(*priv->child_latency) * child_count,
                                     gf_afr_mt_child_latency_t);
+    priv->halo_child_up = GF_CALLOC(sizeof(unsigned char), child_count,
+                                    gf_afr_mt_char);
 
-    if (!priv->child_up || !priv->child_latency) {
+    if (!priv->child_up || !priv->child_latency || !priv->halo_child_up ||
+        !priv->anon_inode) {
         ret = -ENOMEM;
         goto out;
     }
@@ -559,12 +633,20 @@ init(xlator_t *this)
         goto out;
     }
 
-    ret = afr_selfheal_daemon_init(this);
-    if (ret) {
+    this->itable = inode_table_new(SHD_INODE_LRU_LIMIT, this);
+    if (!this->itable) {
         ret = -ENOMEM;
         goto out;
     }
 
+    if (priv->shd.iamshd) {
+        ret = afr_selfheal_daemon_init(this);
+        if (ret) {
+            ret = -ENOMEM;
+            goto out;
+        }
+    }
+
     /* keep more local here as we may need them for self-heal etc */
     this->local_pool = mem_pool_new(afr_local_t, 512);
     if (!this->local_pool) {
@@ -578,24 +660,91 @@ init(xlator_t *this)
 out:
     return ret;
 }
+void
+afr_destroy_healer_object(xlator_t *this, struct subvol_healer *healer)
+{
+    int ret = -1;
 
-int
+    if (!healer)
+        return;
+
+    if (healer->running) {
+        /*
+         * If there are any resources to cleanup, We need
+         * to do that gracefully using pthread_cleanup_push
+         */
+        ret = gf_thread_cleanup_xint(healer->thread);
+        if (ret)
+            gf_msg(this->name, GF_LOG_WARNING, 0, AFR_MSG_SELF_HEAL_FAILED,
+                   "Failed to clean up healer threads.");
+        healer->thread = 0;
+    }
+    pthread_cond_destroy(&healer->cond);
+    pthread_mutex_destroy(&healer->mutex);
+}
+
+void
+afr_selfheal_daemon_fini(xlator_t *this)
+{
+    struct subvol_healer *healer = NULL;
+    afr_self_heald_t *shd = NULL;
+    afr_private_t *priv = NULL;
+    int i = 0;
+
+    priv = this->private;
+    if (!priv)
+        return;
+
+    shd = &priv->shd;
+    if (!shd->iamshd)
+        return;
+
+    for (i = 0; i < priv->child_count; i++) {
+        healer = &shd->index_healers[i];
+        afr_destroy_healer_object(this, healer);
+
+        healer = &shd->full_healers[i];
+        afr_destroy_healer_object(this, healer);
+
+        if (shd->statistics[i])
+            eh_destroy(shd->statistics[i]);
+    }
+    GF_FREE(shd->index_healers);
+    GF_FREE(shd->full_healers);
+    GF_FREE(shd->statistics);
+    if (shd->split_brain)
+        eh_destroy(shd->split_brain);
+}
+void
 fini(xlator_t *this)
 {
     afr_private_t *priv = NULL;
 
     priv = this->private;
+
+    afr_selfheal_daemon_fini(this);
+    GF_ASSERT(list_empty(&priv->saved_locks));
+
     LOCK(&priv->lock);
     if (priv->timer != NULL) {
         gf_timer_call_cancel(this->ctx, priv->timer);
         priv->timer = NULL;
     }
     UNLOCK(&priv->lock);
+
+    if (this->local_pool != NULL) {
+        mem_pool_destroy(this->local_pool);
+        this->local_pool = NULL;
+    }
+
     this->private = NULL;
     afr_priv_destroy(priv);
-    // if (this->itable);//I don't see any destroy func
+    if (this->itable) {
+        inode_table_destroy(this->itable);
+        this->itable = NULL;
+    }
 
-    return 0;
+    return;
 }
 
 struct xlator_fops fops = {
@@ -619,6 +768,7 @@ struct xlator_fops fops = {
     .getxattr = afr_getxattr,
     .fgetxattr = afr_fgetxattr,
     .readv = afr_readv,
+    .seek = afr_seek,
 
     /* inode write */
     .writev = afr_writev,
@@ -691,7 +841,7 @@ struct volume_options options[] = {
     {.key = {"read-hash-mode"},
      .type = GF_OPTION_TYPE_INT,
      .min = 0,
-     .max = 3,
+     .max = 5,
      .default_value = "1",
      .op_version = {2},
      .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE | OPT_FLAG_DOC,
@@ -704,7 +854,10 @@ struct volume_options options[] = {
          "1 = hash by GFID of file (all clients use "
          "same subvolume).\n"
          "2 = hash by GFID of file and client PID.\n"
-         "3 = brick having the least outstanding read requests."},
+         "3 = brick having the least outstanding read requests.\n"
+         "4 = brick having the least network ping latency.\n"
+         "5 = Hybrid mode between 3 and 4, ie least value among "
+         "network-latency multiplied by outstanding-read-requests."},
     {
         .key = {"choose-local"},
         .type = GF_OPTION_TYPE_BOOL,
@@ -797,7 +950,7 @@ struct volume_options options[] = {
      .type = GF_OPTION_TYPE_STR,
      .value = {"1", "on", "yes", "true", "enable", "0", "off", "no", "false",
                "disable", "open"},
-     .default_value = "on",
+     .default_value = "off",
      .op_version = {1},
      .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE | OPT_FLAG_DOC,
      .tags = {"replicate"},
@@ -834,7 +987,7 @@ struct volume_options options[] = {
                     "process would be applied simultaneously."},
     {.key = {"metadata-self-heal"},
      .type = GF_OPTION_TYPE_BOOL,
-     .default_value = "on",
+     .default_value = "off",
      .op_version = {1},
      .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE | OPT_FLAG_DOC,
      .tags = {"replicate"},
@@ -844,7 +997,7 @@ struct volume_options options[] = {
                     "the file/directory."},
     {.key = {"entry-self-heal"},
      .type = GF_OPTION_TYPE_BOOL,
-     .default_value = "on",
+     .default_value = "off",
      .op_version = {1},
      .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE | OPT_FLAG_DOC,
      .tags = {"replicate"},
@@ -1164,5 +1317,28 @@ struct volume_options options[] = {
      .tags = {"replicate"},
      .description = "This option exists only for backward compatibility "
                     "and configuring it doesn't have any effect"},
+    {.key = {"use-anonymous-inode"},
+     .type = GF_OPTION_TYPE_BOOL,
+     .default_value = "no",
+     .op_version = {GD_OP_VERSION_8_0},
+     .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE,
+     .tags = {"replicate"},
+     .description = "Setting this option heals directory renames efficiently"},
+
     {.key = {NULL}},
 };
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .notify = notify,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1}, /* Present from the initial version */
+    .dumpops = &dumpops,
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "replicate",
+    .category = GF_MAINTAINED,
+};
diff --git a/xlators/cluster/afr/src/afr.h b/xlators/cluster/afr/src/afr.h
index 6f8015380f0..d62f9a9caf2 100644
--- a/xlators/cluster/afr/src/afr.h
+++ b/xlators/cluster/afr/src/afr.h
@@ -11,17 +11,18 @@
 #ifndef __AFR_H__
 #define __AFR_H__
 
-#include "call-stub.h"
-#include "compat-errno.h"
+#include <glusterfs/call-stub.h>
+#include <glusterfs/compat-errno.h>
 #include "afr-mem-types.h"
 
 #include "libxlator.h"
-#include "timer.h"
-#include "syncop.h"
+#include <glusterfs/timer.h>
+#include <glusterfs/syncop.h>
 
 #include "afr-self-heald.h"
 #include "afr-messages.h"
 
+#define SHD_INODE_LRU_LIMIT 1
 #define AFR_PATHINFO_HEADER "REPLICATE:"
 #define AFR_SH_READDIR_SIZE_KEY "self-heal-readdir-size"
 #define AFR_SH_DATA_DOMAIN_FMT "%s:self-heal"
@@ -38,7 +39,10 @@
 #define AFR_TA_DOM_NOTIFY "afr.ta.dom-notify"
 #define AFR_TA_DOM_MODIFY "afr.ta.dom-modify"
 
+#define AFR_LK_HEAL_DOM "afr.lock-heal.domain"
+
 #define AFR_HALO_MAX_LATENCY 99999
+#define AFR_ANON_DIR_PREFIX ".glusterfs-anonymous-inode"
 
 #define PFLAG_PENDING (1 << 0)
 #define PFLAG_SBRAIN (1 << 1)
@@ -94,6 +98,25 @@ typedef int (*afr_changelog_resume_t)(call_frame_t *frame, xlator_t *this);
                    gf_fop_list[local->op], uuid_utoa(local->inode->gfid));     \
     } while (0)
 
+#define AFR_ERROR_OUT_IF_FDCTX_INVALID(__fd, __this, __error, __label)         \
+    do {                                                                       \
+        afr_fd_ctx_t *__fd_ctx = NULL;                                         \
+        __fd_ctx = afr_fd_ctx_get(__fd, __this);                               \
+        if (__fd_ctx && __fd_ctx->is_fd_bad) {                                 \
+            __error = EBADF;                                                   \
+            goto __label;                                                      \
+        }                                                                      \
+    } while (0)
+
+typedef enum {
+    AFR_READ_POLICY_FIRST_UP,
+    AFR_READ_POLICY_GFID_HASH,
+    AFR_READ_POLICY_GFID_PID_HASH,
+    AFR_READ_POLICY_LESS_LOAD,
+    AFR_READ_POLICY_LEAST_LATENCY,
+    AFR_READ_POLICY_LOAD_LATENCY_HYBRID,
+} afr_read_hash_mode_t;
+
 typedef enum {
     AFR_FAV_CHILD_NONE,
     AFR_FAV_CHILD_BY_SIZE,
@@ -104,6 +127,12 @@ typedef enum {
 } afr_favorite_child_policy;
 
 typedef enum {
+    AFR_SELFHEAL_DATA_FULL = 0,
+    AFR_SELFHEAL_DATA_DIFF,
+    AFR_SELFHEAL_DATA_DYNAMIC,
+} afr_data_self_heal_type_t;
+
+typedef enum {
     AFR_CHILD_UNKNOWN = -1,
     AFR_CHILD_ZERO,
     AFR_CHILD_ONE,
@@ -119,13 +148,27 @@ typedef enum {
                                   *on BAD brick - Success*/
     TA_INFO_IN_MEMORY_FAILED,    /*Bad brick info is in memory and fop failed
                                   *on GOOD brick - Failed*/
+    TA_SUCCESS,                  /*FOP succeeded on both data bricks.*/
 } afr_ta_fop_state_t;
 
 struct afr_nfsd {
-    gf_boolean_t iamnfsd;
     uint32_t halo_max_latency_msec;
+    gf_boolean_t iamnfsd;
 };
 
+typedef struct _afr_lk_heal_info {
+    fd_t *fd;
+    int32_t cmd;
+    struct gf_flock flock;
+    dict_t *xdata_req;
+    unsigned char *locked_nodes;
+    struct list_head pos;
+    gf_lkowner_t lk_owner;
+    pid_t pid;
+    int32_t *child_up_event_gen;
+    int32_t *child_down_event_gen;
+} afr_lk_heal_info_t;
+
 typedef struct _afr_private {
     gf_lock_t lock;             /* to guard access to child_count, etc */
     unsigned int child_count;   /* total number of children   */
@@ -136,26 +179,27 @@ typedef struct _afr_private {
 
     inode_t *root_inode;
 
+    int favorite_child; /* subvolume to be preferred in resolving
+                                    split-brain cases */
     /* For thin-arbiter. */
-    unsigned int thin_arbiter_count; /* 0 or 1 at the moment.*/
     uuid_t ta_gfid;
-    unsigned char ta_child_up;
+    unsigned int thin_arbiter_count; /* 0 or 1 at the moment.*/
     int ta_bad_child_index;
-    off_t ta_notify_dom_lock_offset;
-    gf_boolean_t release_ta_notify_dom_lock;
+    int ta_event_gen;
     unsigned int ta_in_mem_txn_count;
     unsigned int ta_on_wire_txn_count;
     struct list_head ta_waitq;
     struct list_head ta_onwireq;
 
+    unsigned char *anon_inode;
     unsigned char *child_up;
+    unsigned char *halo_child_up;
     int64_t *child_latency;
     unsigned char *local;
 
     char **pending_key;
 
-    char *data_self_heal;                    /* on/off/open */
-    char *data_self_heal_algorithm;          /* name of algorithm */
+    afr_data_self_heal_type_t data_self_heal_algorithm;
     unsigned int data_self_heal_window_size; /* max number of pipelined
                                                 read/writes */
 
@@ -170,30 +214,31 @@ typedef struct _afr_private {
     int32_t healers; /* No. of elements currently undergoing background
                       heal*/
 
+    gf_boolean_t release_ta_notify_dom_lock;
+
     gf_boolean_t metadata_self_heal; /* on/off */
     gf_boolean_t entry_self_heal;    /* on/off */
 
     gf_boolean_t metadata_splitbrain_forced_heal; /* on/off */
     int read_child;                               /* read-subvolume */
-    unsigned int hash_mode;     /* for when read_child is not set */
     gf_atomic_t *pending_reads; /*No. of pending read cbks per child.*/
-    int favorite_child;         /* subvolume to be preferred in resolving
-                                            split-brain cases */
 
-    afr_favorite_child_policy fav_child_policy; /*Policy to use for automatic
-                                              resolution of split-brains.*/
+    gf_timer_t *timer; /* launched when parent up is received */
 
     unsigned int wait_count; /* # of servers to wait for success */
 
-    gf_timer_t *timer; /* launched when parent up is received */
-
+    unsigned char ta_child_up;
     gf_boolean_t optimistic_change_log;
     gf_boolean_t eager_lock;
     gf_boolean_t pre_op_compat; /* on/off */
     uint32_t post_op_delay_secs;
     unsigned int quorum_count;
 
-    char vol_uuid[UUID_SIZE + 1];
+    off_t ta_notify_dom_lock_offset;
+    afr_favorite_child_policy fav_child_policy; /*Policy to use for automatic
+                                              resolution of split-brains.*/
+    afr_read_hash_mode_t hash_mode; /* for when read_child is not set */
+
     int32_t *last_event;
 
     /* @event_generation: Keeps count of number of events received which can
@@ -206,33 +251,41 @@ typedef struct _afr_private {
        important as we might have had a network split brain.
     */
     uint32_t event_generation;
+    char vol_uuid[UUID_SIZE + 1];
 
     gf_boolean_t choose_local;
     gf_boolean_t did_discovery;
-    uint64_t sh_readdir_size;
     gf_boolean_t ensure_durability;
+    gf_boolean_t halo_enabled;
+    gf_boolean_t consistent_metadata;
+    gf_boolean_t need_heal;
+    gf_boolean_t granular_locks;
+    uint64_t sh_readdir_size;
     char *sh_domain;
     char *afr_dirty;
-    gf_boolean_t halo_enabled;
 
-    uint32_t halo_max_latency_msec;
-    uint32_t halo_max_replicas;
-    uint32_t halo_min_replicas;
+    uint64_t spb_choice_timeout;
 
     afr_self_heald_t shd;
     struct afr_nfsd nfsd;
 
-    gf_boolean_t consistent_metadata;
-    uint64_t spb_choice_timeout;
-    gf_boolean_t need_heal;
+    uint32_t halo_max_latency_msec;
+    uint32_t halo_max_replicas;
+    uint32_t halo_min_replicas;
 
-    /* pump dependencies */
-    void *pump_private;
-    gf_boolean_t use_afr_in_pump;
-    char *locking_scheme;
     gf_boolean_t full_lock;
     gf_boolean_t esh_granular;
     gf_boolean_t consistent_io;
+    gf_boolean_t data_self_heal; /* on/off */
+    gf_boolean_t use_anon_inode;
+
+    /*For lock healing.*/
+    struct list_head saved_locks;
+    struct list_head lk_healq;
+
+    /*For anon-inode handling */
+    char anon_inode_name[NAME_MAX + 1];
+    char anon_gfid_str[UUID_SIZE + 1];
 } afr_private_t;
 
 typedef enum {
@@ -281,12 +334,14 @@ afr_index_from_ia_type(ia_type_t type)
 }
 
 typedef struct {
+    struct gf_flock flock;
     loc_t loc;
+    fd_t *fd;
     char *basename;
     unsigned char *locked_nodes;
     int locked_count;
 
-} afr_entry_lockee_t;
+} afr_lockee_t;
 
 int
 afr_entry_lockee_cmp(const void *l1, const void *l2);
@@ -294,21 +349,17 @@ afr_entry_lockee_cmp(const void *l1, const void *l2);
 typedef struct {
     loc_t *lk_loc;
 
-    int lockee_count;
-    afr_entry_lockee_t lockee[AFR_LOCKEE_COUNT_MAX];
+    afr_lockee_t lockee[AFR_LOCKEE_COUNT_MAX];
 
-    struct gf_flock flock;
     const char *lk_basename;
     const char *lower_basename;
     const char *higher_basename;
-    char lower_locked;
-    char higher_locked;
 
-    unsigned char *locked_nodes;
     unsigned char *lower_locked_nodes;
 
-    int32_t lock_count;
-    int32_t entrylk_lock_count;
+    afr_lock_cbk_t lock_cbk;
+
+    int lockee_count;
 
     int32_t lk_call_count;
     int32_t lk_expected_count;
@@ -316,14 +367,15 @@ typedef struct {
 
     int32_t lock_op_ret;
     int32_t lock_op_errno;
-    afr_lock_cbk_t lock_cbk;
     char *domain; /* Domain on which inode/entry lock/unlock in progress.*/
+    int32_t lock_count;
+    char lower_locked;
+    char higher_locked;
 } afr_internal_lock_t;
 
 struct afr_reply {
     int valid;
     int32_t op_ret;
-    int32_t op_errno;
     dict_t *xattr; /*For xattrop*/
     dict_t *xdata;
     struct iatt poststat;
@@ -332,6 +384,7 @@ struct afr_reply {
     struct iatt preparent;
     struct iatt preparent2;
     struct iatt postparent2;
+    int32_t op_errno;
     /* For rchecksum */
     uint8_t checksum[SHA256_DIGEST_LENGTH];
     gf_boolean_t buf_has_zeroes;
@@ -355,6 +408,10 @@ typedef struct {
        arrives, we continue to read off this subvol.
     */
     int readdir_subvol;
+    /* lock-healing related members. */
+    gf_boolean_t is_fd_bad;
+    afr_lk_heal_info_t *lk_heal_info;
+
 } afr_fd_ctx_t;
 
 typedef enum {
@@ -371,8 +428,6 @@ typedef struct _afr_inode_lock_t {
     */
     int32_t num_inodelks;
     unsigned int event_generation;
-    gf_boolean_t release;
-    gf_boolean_t acquired;
     gf_timer_t *delay_timer;
     struct list_head owners;  /*Transactions that are performing fop*/
     struct list_head post_op; /*Transactions that are done with the fop
@@ -381,6 +436,8 @@ typedef struct _afr_inode_lock_t {
                                *conflicting transactions to complete*/
     struct list_head frozen;  /*Transactions that need to go as part of
                                * next batch of eager-lock*/
+    gf_boolean_t release;
+    gf_boolean_t acquired;
 } afr_lock_t;
 
 typedef struct _afr_inode_ctx {
@@ -389,15 +446,11 @@ typedef struct _afr_inode_ctx {
     int lock_count;
     int spb_choice;
     gf_timer_t *timer;
-    gf_boolean_t need_refresh;
     unsigned int *pre_op_done[AFR_NUM_CHANGE_LOGS];
     int inherited[AFR_NUM_CHANGE_LOGS];
     int on_disk[AFR_NUM_CHANGE_LOGS];
-
-    /* set if any write on this fd was a non stable write
-       (i.e, without O_SYNC or O_DSYNC)
-    */
-    gf_boolean_t witnessed_unstable_write;
+    /*Only 2 types of transactions support eager-locks now. DATA/METADATA*/
+    afr_lock_t lock[2];
 
     /* @open_fd_count:
        Number of open FDs queried from the server, as queried through
@@ -405,8 +458,12 @@ typedef struct _afr_inode_ctx {
        temporarily disabled.
     */
     uint32_t open_fd_count;
-    /*Only 2 types of transactions support eager-locks now. DATA/METADATA*/
-    afr_lock_t lock[2];
+    gf_boolean_t need_refresh;
+
+    /* set if any write on this fd was a non stable write
+       (i.e, without O_SYNC or O_DSYNC)
+    */
+    gf_boolean_t witnessed_unstable_write;
 } afr_inode_ctx_t;
 
 typedef struct _afr_local {
@@ -420,19 +477,15 @@ typedef struct _afr_local {
     unsigned int event_generation;
 
     uint32_t open_fd_count;
-    gf_boolean_t update_open_fd_count;
     int32_t num_inodelks;
-    gf_boolean_t update_num_inodelks;
-
-    gf_lkowner_t saved_lk_owner;
 
     int32_t op_ret;
     int32_t op_errno;
 
-    int32_t **pending;
-
     int dirty[AFR_NUM_CHANGE_LOGS];
 
+    int32_t **pending;
+
     loc_t loc;
     loc_t newloc;
 
@@ -463,14 +516,6 @@ typedef struct _afr_local {
 
     afr_read_txn_wind_t readfn;
 
-    /* @refreshed:
-
-       the inode was "refreshed" (i.e, pending xattrs from all subvols
-       freshly inspected and inode ctx updated accordingly) as part of
-       this transaction already.
-    */
-    gf_boolean_t refreshed;
-
     /* @inode:
 
        the inode on which the read txn is performed on. ref'ed and copied
@@ -495,8 +540,6 @@ typedef struct _afr_local {
     unsigned char *readable;
     unsigned char *readable2; /*For rename transaction*/
 
-    int read_subvol; /* Current read subvolume */
-
     afr_inode_refresh_cbk_t refreshfn;
 
     /* @refreshinode:
@@ -505,9 +548,30 @@ typedef struct _afr_local {
     */
     inode_t *refreshinode;
 
+    dict_t *xattr_req;
+
+    dict_t *dict;
+
+    int read_subvol; /* Current read subvolume */
+
+    int optimistic_change_log;
+
+    afr_internal_lock_t internal_lock;
+
     /*To handle setattr/setxattr on yet to be linked inode from dht*/
     uuid_t refreshgfid;
 
+    /* @refreshed:
+
+       the inode was "refreshed" (i.e, pending xattrs from all subvols
+       freshly inspected and inode ctx updated accordingly) as part of
+       this transaction already.
+    */
+    gf_boolean_t refreshed;
+
+    gf_boolean_t update_num_inodelks;
+    gf_boolean_t update_open_fd_count;
+
     /*
       @pre_op_compat:
 
@@ -517,14 +581,6 @@ typedef struct _afr_local {
 
     gf_boolean_t pre_op_compat;
 
-    dict_t *xattr_req;
-
-    afr_internal_lock_t internal_lock;
-
-    dict_t *dict;
-
-    int optimistic_change_log;
-
     /* Is the current writev() going to perform a stable write?
        i.e, is fd->flags or @flags writev param have O_SYNC or
        O_DSYNC?
@@ -543,25 +599,25 @@ typedef struct _afr_local {
 
     struct {
         struct {
-            gf_boolean_t needs_fresh_lookup;
-            uuid_t gfid_req;
-        } lookup;
-
-        struct {
-            unsigned char buf_set;
             struct statvfs buf;
+            unsigned char buf_set;
         } statfs;
 
         struct {
-            int32_t flags;
             fd_t *fd;
+            int32_t flags;
         } open;
 
         struct {
-            int32_t cmd;
             struct gf_flock user_flock;
             struct gf_flock ret_flock;
             unsigned char *locked_nodes;
+            int32_t cmd;
+            /*For lock healing only.*/
+            unsigned char *dom_locked_nodes;
+            int32_t *dom_lock_op_ret;
+            int32_t *dom_lock_op_errno;
+            struct gf_flock *getlk_rsp;
         } lk;
 
         /* inode read */
@@ -586,8 +642,8 @@ typedef struct _afr_local {
 
         struct {
             char *name;
-            int last_index;
             long xattr_len;
+            int last_index;
         } getxattr;
 
         struct {
@@ -600,11 +656,10 @@ typedef struct _afr_local {
         /* dir read */
 
         struct {
+            uint32_t *checksum;
             int success_count;
             int32_t op_ret;
             int32_t op_errno;
-
-            uint32_t *checksum;
         } opendir;
 
         struct {
@@ -613,8 +668,8 @@ typedef struct _afr_local {
             size_t size;
             off_t offset;
             dict_t *dict;
-            gf_boolean_t failed;
             int last_index;
+            gf_boolean_t failed;
         } readdir;
         /* inode write */
 
@@ -624,12 +679,11 @@ typedef struct _afr_local {
         } inode_wfop;  // common structure for all inode-write-fops
 
         struct {
-            int32_t op_ret;
-
             struct iovec *vector;
             struct iobref *iobref;
-            int32_t count;
             off_t offset;
+            int32_t op_ret;
+            int32_t count;
             uint32_t flags;
         } writev;
 
@@ -689,29 +743,25 @@ typedef struct _afr_local {
         } create;
 
         struct {
+            dict_t *params;
             dev_t dev;
             mode_t mode;
-            dict_t *params;
         } mknod;
 
         struct {
-            int32_t mode;
             dict_t *params;
+            int32_t mode;
         } mkdir;
 
         struct {
-            int flags;
-        } rmdir;
-
-        struct {
             dict_t *params;
             char *linkpath;
         } symlink;
 
         struct {
-            int32_t mode;
             off_t offset;
             size_t len;
+            int32_t mode;
         } fallocate;
 
         struct {
@@ -738,10 +788,10 @@ typedef struct _afr_local {
         struct {
             char *volume;
             char *basename;
+            void *xdata;
             entrylk_cmd in_cmd;
             entrylk_cmd cmd;
             entrylk_type type;
-            void *xdata;
         } entrylk;
 
         struct {
@@ -750,31 +800,33 @@ typedef struct _afr_local {
         } seek;
 
         struct {
-            int32_t datasync;
-        } fsync;
-
-        struct {
             struct gf_lease user_lease;
             struct gf_lease ret_lease;
             unsigned char *locked_nodes;
         } lease;
 
-    } cont;
+        struct {
+            int flags;
+        } rmdir;
 
-    struct {
-        off_t start, len;
+        struct {
+            int32_t datasync;
+        } fsync;
 
-        gf_boolean_t eager_lock_on;
-        gf_boolean_t do_eager_unlock;
+        struct {
+            uuid_t gfid_req;
+            gf_boolean_t needs_fresh_lookup;
+        } lookup;
+
+    } cont;
 
+    struct {
         char *basename;
         char *new_basename;
 
         loc_t parent_loc;
         loc_t new_parent_loc;
 
-        afr_transaction_type type;
-
         /* stub to resume on destruction
            of the transaction frame */
         call_stub_t *resume_stub;
@@ -792,6 +844,30 @@ typedef struct _afr_local {
             FOP failed. */
         unsigned char *failed_subvols;
 
+        call_frame_t *main_frame; /*Fop frame*/
+        call_frame_t *frame;      /*Transaction frame*/
+
+        int (*wind)(call_frame_t *frame, xlator_t *this, int subvol);
+
+        int (*unwind)(call_frame_t *frame, xlator_t *this);
+
+        off_t start, len;
+
+        afr_transaction_type type;
+
+        int32_t in_flight_sb_errno; /* This is where the cause of the
+                                       failure on the last good copy of
+                                       the file is stored.
+                                       */
+
+        /* @changelog_resume: function to be called after changlogging
+           (either pre-op or post-op) is done
+        */
+        afr_changelog_resume_t changelog_resume;
+
+        gf_boolean_t eager_lock_on;
+        gf_boolean_t do_eager_unlock;
+
         /* @dirtied: flag which indicates whether we set dirty flag
            in the OP. Typically true when we are performing operation
            on more than one subvol and optimistic changelog is disabled
@@ -816,6 +892,10 @@ typedef struct _afr_local {
         */
         gf_boolean_t no_uninherit;
 
+        gf_boolean_t in_flight_sb; /* Indicator for occurrence of
+                                      split-brain while in the middle of
+                                      a txn. */
+
         /* @uninherit_done:
            @uninherit_value:
 
@@ -828,27 +908,7 @@ typedef struct _afr_local {
         gf_boolean_t uninherit_done;
         gf_boolean_t uninherit_value;
 
-        gf_boolean_t in_flight_sb;  /* Indicator for occurrence of
-                                       split-brain while in the middle of
-                                       a txn. */
-        int32_t in_flight_sb_errno; /* This is where the cause of the
-                                       failure on the last good copy of
-                                       the file is stored.
-                                       */
-
-        /* @changelog_resume: function to be called after changlogging
-           (either pre-op or post-op) is done
-        */
-        afr_changelog_resume_t changelog_resume;
-
-        call_frame_t *main_frame; /*Fop frame*/
-        call_frame_t *frame;      /*Transaction frame*/
-
-        int (*wind)(call_frame_t *frame, xlator_t *this, int subvol);
-
-        int (*unwind)(call_frame_t *frame, xlator_t *this);
-
-        /* post-op hook */
+        gf_boolean_t disable_delayed_post_op;
     } transaction;
 
     syncbarrier_t barrier;
@@ -861,33 +921,36 @@ typedef struct _afr_local {
 
     mode_t umask;
     int xflag;
-    gf_boolean_t do_discovery;
     struct afr_reply *replies;
 
     /* For  client side background heals. */
     struct list_head healer;
     call_frame_t *heal_frame;
 
-    gf_boolean_t need_full_crawl;
-    afr_fop_lock_state_t fop_lock_state;
-
-    gf_boolean_t is_read_txn;
     afr_inode_ctx_t *inode_ctx;
 
     /*For thin-arbiter transactions.*/
-    unsigned char ta_child_up;
+    int ta_failed_subvol;
+    int ta_event_gen;
     struct list_head ta_waitq;
     struct list_head ta_onwireq;
     afr_ta_fop_state_t fop_state;
-    int ta_failed_subvol;
+    afr_fop_lock_state_t fop_lock_state;
+    gf_lkowner_t saved_lk_owner;
+    unsigned char read_txn_query_child;
+    unsigned char ta_child_up;
+    gf_boolean_t do_discovery;
+    gf_boolean_t need_full_crawl;
+    gf_boolean_t is_read_txn;
+    gf_boolean_t is_new_entry;
 } afr_local_t;
 
 typedef struct afr_spbc_timeout {
     call_frame_t *frame;
-    gf_boolean_t d_spb;
-    gf_boolean_t m_spb;
     loc_t *loc;
     int spb_child_index;
+    gf_boolean_t d_spb;
+    gf_boolean_t m_spb;
 } afr_spbc_timeout_t;
 
 typedef struct afr_spb_status {
@@ -897,9 +960,9 @@ typedef struct afr_spb_status {
 
 typedef struct afr_empty_brick_args {
     call_frame_t *frame;
+    char *op_type;
     loc_t loc;
     int empty_index;
-    char *op_type;
 } afr_empty_brick_args_t;
 
 typedef struct afr_read_subvol_args {
@@ -941,7 +1004,10 @@ afr_inode_read_subvol_set(inode_t *inode, xlator_t *this,
                           int event_generation);
 
 int
-afr_inode_event_gen_reset(inode_t *inode, xlator_t *this);
+__afr_inode_need_refresh_set(inode_t *inode, xlator_t *this);
+
+int
+afr_inode_need_refresh_set(inode_t *inode, xlator_t *this);
 
 int
 afr_read_subvol_select_by_policy(inode_t *inode, xlator_t *this,
@@ -973,11 +1039,14 @@ int
 xattr_is_equal(dict_t *this, char *key1, data_t *value1, void *data);
 
 int
-afr_init_entry_lockee(afr_entry_lockee_t *lockee, afr_local_t *local,
-                      loc_t *loc, char *basename, int child_count);
+afr_add_entry_lockee(afr_local_t *local, loc_t *loc, char *basename,
+                     int child_count);
+
+int
+afr_add_inode_lockee(afr_local_t *local, int child_count);
 
 void
-afr_entry_lockee_cleanup(afr_internal_lock_t *int_lock);
+afr_lockees_cleanup(afr_internal_lock_t *int_lock);
 
 int
 afr_attempt_lock_recovery(xlator_t *this, int32_t child_index);
@@ -995,10 +1064,7 @@ int32_t
 afr_unlock(call_frame_t *frame, xlator_t *this);
 
 int
-afr_nonblocking_entrylk(call_frame_t *frame, xlator_t *this);
-
-int
-afr_nonblocking_inodelk(call_frame_t *frame, xlator_t *this);
+afr_lock_nonblocking(call_frame_t *frame, xlator_t *this);
 
 int
 afr_blocking_lock(call_frame_t *frame, xlator_t *this);
@@ -1057,6 +1123,9 @@ afr_cleanup_fd_ctx(xlator_t *this, fd_t *fd);
             if (__local && __local->is_read_txn)                               \
                 afr_pending_read_decrement(__this->private,                    \
                                            __local->read_subvol);              \
+            if (__local && __local->xdata_req &&                               \
+                afr_is_lock_mode_mandatory(__local->xdata_req))                \
+                afr_dom_lock_release(frame);                                   \
             frame->local = NULL;                                               \
         }                                                                      \
                                                                                \
@@ -1084,8 +1153,8 @@ afr_cleanup_fd_ctx(xlator_t *this, fd_t *fd);
 #define AFR_FRAME_INIT(frame, op_errno)                                        \
     ({                                                                         \
         frame->local = mem_get0(THIS->local_pool);                             \
-        if (afr_local_init(frame->local, THIS->private, &op_errno)) {          \
-            afr_local_cleanup(frame->local, THIS);                             \
+        if (afr_local_init(frame->local, frame->this->private, &op_errno)) {   \
+            afr_local_cleanup(frame->local, frame->this);                      \
             mem_put(frame->local);                                             \
             frame->local = NULL;                                               \
         };                                                                     \
@@ -1209,8 +1278,8 @@ int
 afr_inode_split_brain_choice_set(inode_t *inode, xlator_t *this,
                                  int spb_choice);
 int
-afr_inode_split_brain_choice_get(inode_t *inode, xlator_t *this,
-                                 int *spb_choice);
+afr_split_brain_read_subvol_get(inode_t *inode, xlator_t *this,
+                                call_frame_t *frame, int *spb_subvol);
 int
 afr_get_child_index_from_name(xlator_t *this, char *name);
 
@@ -1265,9 +1334,6 @@ afr_writev_copy_outvars(call_frame_t *src_frame, call_frame_t *dst_frame);
 void
 afr_update_uninodelk(afr_local_t *local, afr_internal_lock_t *int_lock,
                      int32_t child_index);
-int
-afr_is_inodelk_transaction(afr_transaction_type type);
-
 afr_fd_ctx_t *
 __afr_fd_ctx_get(fd_t *fd, xlator_t *this);
 
@@ -1298,7 +1364,7 @@ int
 afr_set_inode_local(xlator_t *this, afr_local_t *local, inode_t *inode);
 
 int
-afr_fill_ta_loc(xlator_t *this, loc_t *loc);
+afr_fill_ta_loc(xlator_t *this, loc_t *loc, gf_boolean_t is_gfid_based_fop);
 
 int
 afr_ta_post_op_lock(xlator_t *this, loc_t *loc);
@@ -1320,4 +1386,38 @@ afr_ta_has_quorum(afr_private_t *priv, afr_local_t *local);
 
 void
 afr_ta_lock_release_synctask(xlator_t *this);
+
+void
+afr_ta_locked_priv_invalidate(afr_private_t *priv);
+
+gf_boolean_t
+afr_lookup_has_quorum(call_frame_t *frame,
+                      const unsigned int up_children_count);
+
+void
+afr_mark_new_entry_changelog(call_frame_t *frame, xlator_t *this);
+
+void
+afr_handle_replies_quorum(call_frame_t *frame, xlator_t *this);
+
+gf_boolean_t
+afr_ta_dict_contains_pending_xattr(dict_t *dict, afr_private_t *priv,
+                                   int child);
+
+void
+afr_selfheal_childup(xlator_t *this, afr_private_t *priv);
+
+gf_boolean_t
+afr_is_lock_mode_mandatory(dict_t *xdata);
+
+void
+afr_dom_lock_release(call_frame_t *frame);
+
+void
+afr_fill_success_replies(afr_local_t *local, afr_private_t *priv,
+                         unsigned char *replies);
+
+gf_boolean_t
+afr_is_private_directory(afr_private_t *priv, uuid_t pargfid, const char *name,
+                         pid_t pid);
 #endif /* __AFR_H__ */
diff --git a/xlators/cluster/dht/src/Makefile.am b/xlators/cluster/dht/src/Makefile.am
index caeb17b0f07..56f1f2ad7c8 100644
--- a/xlators/cluster/dht/src/Makefile.am
+++ b/xlators/cluster/dht/src/Makefile.am
@@ -14,19 +14,13 @@ dht_la_SOURCES = $(dht_common_source) dht.c
 nufa_la_SOURCES = $(dht_common_source) nufa.c
 switch_la_SOURCES = $(dht_common_source) switch.c
 
-dht_la_LDFLAGS = -module \
-	-export-symbols $(top_srcdir)/xlators/cluster/dht/src/dht.sym \
-	$(GF_XLATOR_LDFLAGS)
+dht_la_LDFLAGS = -module $(GF_XLATOR_DEFAULT_LDFLAGS)
 dht_la_LIBADD = $(top_builddir)/libglusterfs/src/libglusterfs.la
 
-nufa_la_LDFLAGS = -module \
-	-export-symbols $(top_srcdir)/xlators/cluster/dht/src/nufa.sym \
-	$(GF_XLATOR_LDFLAGS)
+nufa_la_LDFLAGS = -module $(GF_XLATOR_DEFAULT_LDFLAGS)
 nufa_la_LIBADD = $(top_builddir)/libglusterfs/src/libglusterfs.la
 
-switch_la_LDFLAGS = -module \
-	-export-symbols $(top_srcdir)/xlators/cluster/dht/src/switch.sym \
-	$(GF_XLATOR_LDFLAGS)
+switch_la_LDFLAGS = -module $(GF_XLATOR_DEFAULT_LDFLAGS)
 switch_la_LIBADD = $(top_builddir)/libglusterfs/src/libglusterfs.la
 
 noinst_HEADERS = dht-common.h dht-mem-types.h dht-messages.h \
@@ -41,8 +35,6 @@ AM_CPPFLAGS = $(GF_CPPFLAGS) -I$(top_srcdir)/libglusterfs/src \
 
 CLEANFILES =
 
-EXTRA_DIST = dht.sym nufa.sym switch.sym
-
 uninstall-local:
 	rm -f $(DESTDIR)$(xlatordir)/distribute.so
 
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index 24402399597..8ba0cc4c732 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -10,48 +10,52 @@
 
 /* TODO: add NS locking */
 
-#include "glusterfs.h"
-#include "xlator.h"
 #include "libxlator.h"
 #include "dht-common.h"
 #include "dht-lock.h"
-#include "defaults.h"
-#include "byte-order.h"
-#include "quota-common-utils.h"
-#include "upcall-utils.h"
+#include <glusterfs/byte-order.h>
+#include <glusterfs/quota-common-utils.h>
+#include <glusterfs/upcall-utils.h>
+#include "glusterfs/compat-errno.h"  // for ENODATA on BSD
+#include <glusterfs/common-utils.h>
 
 #include <sys/time.h>
 #include <libgen.h>
 #include <signal.h>
 
-int run_defrag = 0;
-
-int
-dht_link2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
+static int
+dht_rmdir_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                       int op_ret, int op_errno, gf_dirent_t *entries,
+                       dict_t *xdata);
 
-int
-dht_removexattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame,
-                 int ret);
+static int
+dht_link2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
 
-int
-dht_setxattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
+static int
+dht_set_dir_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req);
 
-int
-dht_rmdir_readdirp_do(call_frame_t *readdirp_frame, xlator_t *this);
+static int
+dht_lookup_everywhere_done(call_frame_t *frame, xlator_t *this);
 
-int
-dht_common_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, dict_t *dict,
-                       dict_t *xdata);
+static int
+dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                             int op_ret, int op_errno, dict_t *xdata);
 
-int
-dht_set_file_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req);
+static int
+dht_rmdir_unlock(call_frame_t *frame, xlator_t *this);
 
-int
-dht_set_dir_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req);
+static const char *dht_dbg_vxattrs[] = {DHT_DBG_HASHED_SUBVOL_PATTERN, NULL};
 
-int
-dht_do_fresh_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc);
+/* Check the xdata to make sure EBADF has been set by client xlator */
+int32_t
+dht_check_remote_fd_failed_error(dht_local_t *local, int op_ret, int op_errno)
+{
+    if (op_ret == -1 && (op_errno == EBADF || op_errno == EBADFD) &&
+        !(local->fd_checked)) {
+        return 1;
+    }
+    return 0;
+}
 
 /* Sets the blocks and size values to fixed values. This is to be called
  * only for dirs. The caller is responsible for checking the type
@@ -67,67 +71,17 @@ dht_set_fixed_dir_stat(struct iatt *stat)
     return -1;
 }
 
-/* Set both DHT_IATT_IN_XDATA_KEY and DHT_MODE_IN_XDATA_KEY
- * Use DHT_MODE_IN_XDATA_KEY if available. Else fall back to
- * DHT_IATT_IN_XDATA_KEY
- */
-int
-dht_request_iatt_in_xdata(xlator_t *this, dict_t *xattr_req)
-{
-    int ret = -1;
-
-    ret = dict_set_int8(xattr_req, DHT_MODE_IN_XDATA_KEY, 1);
-    ret = dict_set_int8(xattr_req, DHT_IATT_IN_XDATA_KEY, 1);
-
-    /* At least one call succeeded */
-    return ret;
-}
-
-/* Get both DHT_IATT_IN_XDATA_KEY and DHT_MODE_IN_XDATA_KEY
- * Use DHT_MODE_IN_XDATA_KEY if available, else fall back to
- * DHT_IATT_IN_XDATA_KEY
- * This will return a dummy iatt with only the mode and type set
- */
-int
-dht_read_iatt_from_xdata(xlator_t *this, dict_t *xdata, struct iatt *stbuf)
-{
-    int ret = -1;
-    int32_t mode = 0;
-
-    ret = dict_get_int32(xdata, DHT_MODE_IN_XDATA_KEY, &mode);
-
-    if (ret) {
-        ret = dict_get_bin(xdata, DHT_IATT_IN_XDATA_KEY, (void **)&stbuf);
-    } else {
-        stbuf->ia_prot = ia_prot_from_st_mode(mode);
-        stbuf->ia_type = ia_type_from_st_mode(mode);
-    }
-
-    return ret;
-}
-
-int
-dht_rmdir_unlock(call_frame_t *frame, xlator_t *this);
-
-char *xattrs_to_heal[] = {"user.",
-                          POSIX_ACL_ACCESS_XATTR,
-                          POSIX_ACL_DEFAULT_XATTR,
-                          QUOTA_LIMIT_KEY,
-                          QUOTA_LIMIT_OBJECTS_KEY,
-                          GF_SELINUX_XATTR_KEY,
-                          NULL};
-
-char *dht_dbg_vxattrs[] = {DHT_DBG_HASHED_SUBVOL_PATTERN, NULL};
-
 /* Return true if key exists in array
  */
 static gf_boolean_t
 dht_match_xattr(const char *key)
 {
+    char **xattrs_to_heal = get_xattrs_to_heal();
+
     return gf_get_index_by_elem(xattrs_to_heal, (char *)key) >= 0;
 }
 
-int
+static int
 dht_aggregate_quota_xattr(dict_t *dst, char *key, data_t *value)
 {
     int ret = -1;
@@ -190,7 +144,7 @@ out:
     return ret;
 }
 
-int
+static int
 add_opt(char **optsp, const char *opt)
 {
     char *newopts = NULL;
@@ -268,7 +222,7 @@ out:
 
 */
 
-int
+static int
 dht_aggregate_split_brain_xattr(dict_t *dst, char *key, data_t *value)
 {
     int ret = 0;
@@ -367,7 +321,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_aggregate(dict_t *this, char *key, data_t *value, void *data)
 {
     dict_t *dst = NULL;
@@ -414,7 +368,7 @@ out:
     return ret;
 }
 
-void
+static void
 dht_aggregate_xattr(dict_t *dst, dict_t *src)
 {
     if ((dst == NULL) || (src == NULL)) {
@@ -496,7 +450,7 @@ dht_inode_ctx_mdsvol_get(inode_t *inode, xlator_t *this, xlator_t **mdsvol)
    - complete linkfile selfheal
 */
 
-int
+static int
 dht_lookup_selfheal_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                         int op_ret, int op_errno, dict_t *xdata)
 {
@@ -538,7 +492,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_discover_complete(xlator_t *this, call_frame_t *discover_frame)
 {
     dht_local_t *local = NULL;
@@ -659,13 +613,14 @@ dht_discover_complete(xlator_t *this, call_frame_t *discover_frame)
 
         if (local->need_xattr_heal && !heal_path) {
             local->need_xattr_heal = 0;
-            ret = dht_dir_xattr_heal(this, local);
-            if (ret)
-                gf_msg(this->name, GF_LOG_ERROR, ret,
+            ret = dht_dir_xattr_heal(this, local, &op_errno);
+            if (ret) {
+                gf_msg(this->name, GF_LOG_ERROR, op_errno,
                        DHT_MSG_DIR_XATTR_HEAL_FAILED,
                        "xattr heal failed for "
                        "directory  gfid is %s ",
                        gfid_local);
+            }
         }
     }
 
@@ -726,7 +681,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                              int op_ret, int op_errno, dict_t *xdata)
 {
@@ -735,6 +690,7 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int ret = -1;
     dht_conf_t *conf = 0;
     dht_layout_t *layout = NULL;
+    int32_t mds_heal_fresh_lookup = 0;
 
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, frame->local, out);
@@ -742,6 +698,7 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     local = frame->local;
     conf = this->private;
     layout = local->selfheal.layout;
+    mds_heal_fresh_lookup = local->mds_heal_fresh_lookup;
 
     if (op_ret) {
         gf_msg_debug(this->name, op_ret,
@@ -762,11 +719,63 @@ dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                                  layout);
     }
 out:
-    if (local && local->mds_heal_fresh_lookup)
+    if (mds_heal_fresh_lookup)
         DHT_STACK_DESTROY(frame);
     return 0;
 }
 
+static xlator_t *
+dht_inode_get_hashed_subvol(inode_t *inode, xlator_t *this, loc_t *loc)
+{
+    char *path = NULL;
+    loc_t populate_loc = {
+        0,
+    };
+    char *name = NULL;
+    xlator_t *hash_subvol = NULL;
+
+    if (!inode)
+        return hash_subvol;
+
+    if (loc && loc->parent && loc->path) {
+        if (!loc->name) {
+            name = strrchr(loc->path, '/');
+            if (name) {
+                loc->name = name + 1;
+            } else {
+                goto out;
+            }
+        }
+        hash_subvol = dht_subvol_get_hashed(this, loc);
+        goto out;
+    }
+
+    if (!gf_uuid_is_null(inode->gfid)) {
+        populate_loc.inode = inode_ref(inode);
+        populate_loc.parent = inode_parent(populate_loc.inode, NULL, NULL);
+        inode_path(populate_loc.inode, NULL, &path);
+
+        if (!path)
+            goto out;
+
+        populate_loc.path = path;
+        if (!populate_loc.name && populate_loc.path) {
+            name = strrchr(populate_loc.path, '/');
+            if (name) {
+                populate_loc.name = name + 1;
+
+            } else {
+                goto out;
+            }
+        }
+        hash_subvol = dht_subvol_get_hashed(this, &populate_loc);
+    }
+out:
+    if (populate_loc.inode)
+        loc_wipe(&populate_loc);
+    return hash_subvol;
+}
+
 /* Common function call by revalidate/selfheal code path to populate
    internal xattr if it is not present, mark_during_fresh_lookup value
    determines either function is call by revalidate_cbk(discover_complete)
@@ -801,9 +810,8 @@ dht_common_mark_mdsxattr(call_frame_t *frame, int *errst,
     call_frame_t *xattr_frame = NULL;
     gf_boolean_t vol_down = _gf_false;
 
-    this = frame->this;
-
     GF_VALIDATE_OR_GOTO("dht", frame, out);
+    this = frame->this;
     GF_VALIDATE_OR_GOTO("dht", this, out);
     GF_VALIDATE_OR_GOTO(this->name, frame->local, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
@@ -812,6 +820,7 @@ dht_common_mark_mdsxattr(call_frame_t *frame, int *errst,
     conf = this->private;
     layout = local->selfheal.layout;
     local->mds_heal_fresh_lookup = mark_during_fresh_lookup;
+
     gf_uuid_unparse(local->gfid, gfid_local);
 
     /* Code to update hashed subvol consider as a mds subvol
@@ -852,7 +861,8 @@ dht_common_mark_mdsxattr(call_frame_t *frame, int *errst,
                    "Failed to get hashed subvol for path %s"
                    "gfid is %s ",
                    local->loc.path, gfid_local);
-            (*errst) = 1;
+            if (errst)
+                (*errst) = 1;
             ret = -1;
             goto out;
         }
@@ -923,7 +933,44 @@ out:
     return ret;
 }
 
-int
+/* Get the value of key from dict in the bytewise and save in array after
+   convert from network byte order to host byte order
+*/
+static int32_t
+dht_dict_get_array(dict_t *dict, char *key, int32_t value[], int32_t size,
+                   int *errst)
+{
+    void *ptr = NULL;
+    int32_t len = -1;
+    int32_t vindex = -1;
+    int32_t err = -1;
+    int ret = 0;
+
+    if (dict == NULL) {
+        (*errst) = -1;
+        return -EINVAL;
+    }
+    err = dict_get_ptr_and_len(dict, key, &ptr, &len);
+    if (err != 0) {
+        (*errst) = -1;
+        return err;
+    }
+
+    if (len != (size * sizeof(int32_t))) {
+        (*errst) = -1;
+        return -EINVAL;
+    }
+
+    for (vindex = 0; vindex < size; vindex++) {
+        value[vindex] = ntoh32(*((int32_t *)ptr + vindex));
+        if (value[vindex] < 0)
+            ret = -1;
+    }
+
+    return ret;
+}
+
+static int
 dht_discover_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                  int op_errno, inode_t *inode, struct iatt *stbuf,
                  dict_t *xattr, struct iatt *postparent)
@@ -1018,7 +1065,7 @@ dht_discover_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
         if (local->xattr == NULL) {
             local->xattr = dict_ref(xattr);
         } else {
-            /* Don't aggregate for files. See BZ#1484113 */
+            /* Don't aggregate for files. See BZ#1484709 */
             if (is_dir)
                 dht_aggregate_xattr(local->xattr, xattr);
         }
@@ -1084,7 +1131,53 @@ out:
     return 0;
 }
 
-int
+static int
+dht_set_file_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req)
+{
+    int ret = -EINVAL;
+    dht_conf_t *conf = NULL;
+
+    conf = this->private;
+    if (!conf) {
+        goto err;
+    }
+
+    if (!xattr_req) {
+        goto err;
+    }
+
+    /* Used to check whether this is a linkto file.
+     */
+    ret = dict_set_uint32(xattr_req, conf->link_xattr_name, 256);
+    if (ret < 0) {
+        gf_msg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
+               "Failed to set dictionary value:key = %s for "
+               "path %s",
+               conf->link_xattr_name, loc->path);
+        goto err;
+    }
+
+    /* This is used to make sure we don't unlink linkto files
+     * which are the target of an ongoing file migration.
+     */
+    ret = dict_set_uint32(xattr_req, GLUSTERFS_OPEN_FD_COUNT, 4);
+    if (ret) {
+        gf_msg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
+               "Failed to set dictionary value:key = %s for "
+               "path %s",
+               GLUSTERFS_OPEN_FD_COUNT, loc->path);
+        goto err;
+    }
+
+    ret = 0;
+err:
+    return ret;
+}
+
+/* This is a gfid based nameless lookup. Without a name, the hashed subvol
+ * cannot be calculated so a lookup is sent to all subvols.
+ */
+static int
 dht_do_discover(call_frame_t *frame, xlator_t *this, loc_t *loc)
 {
     int ret;
@@ -1098,6 +1191,9 @@ dht_do_discover(call_frame_t *frame, xlator_t *this, loc_t *loc)
     conf = this->private;
     local = frame->local;
 
+    /* As we do not know if this is a file or directory, request
+     * both file and directory xattrs
+     */
     ret = dht_set_file_xattr_req(this, loc, local->xattr_req);
     if (ret) {
         goto err;
@@ -1109,6 +1205,9 @@ dht_do_discover(call_frame_t *frame, xlator_t *this, loc_t *loc)
     }
 
     if (loc_is_root(loc)) {
+        /* Request the DHT commit hash xattr (trusted.glusterfs.dht.commithash)
+         * set on the brick root.
+         */
         ret = dict_set_uint32(local->xattr_req, conf->commithash_xattr_name,
                               sizeof(uint32_t));
     }
@@ -1150,48 +1249,11 @@ err:
     return 0;
 }
 
-/* Get the value of key from dict in the bytewise and save in array after
-   convert from network byte order to host byte order
-*/
-int32_t
-dht_dict_get_array(dict_t *dict, char *key, int32_t value[], int32_t size,
-                   int *errst)
-{
-    void *ptr = NULL;
-    int32_t len = -1;
-    int32_t vindex = -1;
-    int32_t err = -1;
-    int ret = 0;
-
-    if (dict == NULL) {
-        (*errst) = -1;
-        return -EINVAL;
-    }
-    err = dict_get_ptr_and_len(dict, key, &ptr, &len);
-    if (err != 0) {
-        (*errst) = -1;
-        return err;
-    }
-
-    if (len != (size * sizeof(int32_t))) {
-        (*errst) = -1;
-        return -EINVAL;
-    }
-
-    for (vindex = 0; vindex < size; vindex++) {
-        value[vindex] = ntoh32(*((int32_t *)ptr + vindex));
-        if (value[vindex] < 0)
-            ret = -1;
-    }
-
-    return ret;
-}
-
 /* Code to call syntask to heal custom xattr from hashed subvol
    to non hashed subvol
 */
 int
-dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
+dht_dir_xattr_heal(xlator_t *this, dht_local_t *local, int *op_errno)
 {
     dht_local_t *copy_local = NULL;
     call_frame_t *copy = NULL;
@@ -1203,6 +1265,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
                "No gfid exists for path %s "
                "so healing xattr is not possible",
                local->loc.path);
+        *op_errno = EIO;
         goto out;
     }
 
@@ -1216,6 +1279,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
                    "Memory allocation failed "
                    "for path %s gfid %s ",
                    local->loc.path, gfid_local);
+            *op_errno = ENOMEM;
             DHT_STACK_DESTROY(copy);
         } else {
             copy_local->stbuf = local->stbuf;
@@ -1230,6 +1294,7 @@ dht_dir_xattr_heal(xlator_t *this, dht_local_t *local)
                        "Synctask creation failed to heal xattr "
                        "for path %s gfid %s ",
                        local->loc.path, gfid_local);
+                *op_errno = ENOMEM;
                 DHT_STACK_DESTROY(copy);
             }
         }
@@ -1238,6 +1303,51 @@ out:
     return ret;
 }
 
+static int
+dht_needs_selfheal(call_frame_t *frame, xlator_t *this)
+{
+    dht_local_t *local = NULL;
+    dht_layout_t *layout = NULL;
+    int needs_selfheal = 0;
+    int ret = 0;
+
+    local = frame->local;
+    layout = local->layout;
+
+    if (local->need_attrheal || local->need_xattr_heal ||
+        local->need_selfheal) {
+        needs_selfheal = 1;
+    }
+
+    ret = dht_layout_normalize(this, &local->loc, layout);
+
+    if (ret != 0) {
+        gf_msg_debug(this->name, 0, "fixing assignment on %s", local->loc.path);
+        needs_selfheal = 1;
+    }
+    return needs_selfheal;
+}
+
+static int
+is_permission_different(ia_prot_t *prot1, ia_prot_t *prot2)
+{
+    if ((prot1->owner.read != prot2->owner.read) ||
+        (prot1->owner.write != prot2->owner.write) ||
+        (prot1->owner.exec != prot2->owner.exec) ||
+        (prot1->group.read != prot2->group.read) ||
+        (prot1->group.write != prot2->group.write) ||
+        (prot1->group.exec != prot2->group.exec) ||
+        (prot1->other.read != prot2->other.read) ||
+        (prot1->other.write != prot2->other.write) ||
+        (prot1->other.exec != prot2->other.exec) ||
+        (prot1->suid != prot2->suid) || (prot1->sgid != prot2->sgid) ||
+        (prot1->sticky != prot2->sticky)) {
+        return 1;
+    } else {
+        return 0;
+    }
+}
+
 int
 dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                    int op_ret, int op_errno, inode_t *inode, struct iatt *stbuf,
@@ -1255,8 +1365,6 @@ dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     char gfid_local[GF_UUID_BUF_SIZE] = {0};
     char gfid_node[GF_UUID_BUF_SIZE] = {0};
     int32_t mds_xattr_val[1] = {0};
-    call_frame_t *copy = NULL;
-    dht_local_t *copy_local = NULL;
 
     GF_VALIDATE_OR_GOTO("dht", frame, out);
     GF_VALIDATE_OR_GOTO("dht", this, out);
@@ -1269,7 +1377,11 @@ dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     conf = this->private;
 
     layout = local->layout;
+    gf_msg_debug(this->name, op_errno,
+                 "%s: lookup on %s returned with op_ret = %d, op_errno = %d",
+                 local->loc.path, prev->name, op_ret, op_errno);
 
+    /* The first successful lookup*/
     if (!op_ret && gf_uuid_is_null(local->gfid)) {
         memcpy(local->gfid, stbuf->ia_gfid, 16);
     }
@@ -1291,29 +1403,28 @@ dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     {
         /* TODO: assert equal mode on stbuf->st_mode and
            local->stbuf->st_mode
-
            else mkdir/chmod/chown and fix
         */
         ret = dht_layout_merge(this, layout, prev, op_ret, op_errno, xattr);
 
         if (op_ret == -1) {
             local->op_errno = op_errno;
-            gf_msg_debug(this->name, op_errno,
-                         "lookup of %s on %s returned error", local->loc.path,
-                         prev->name);
 
+            /* The GFID is missing on this subvol. Force a heal. */
+            if (op_errno == ENODATA) {
+                local->need_lookup_everywhere = 1;
+            }
             goto unlock;
         }
 
         is_dir = check_is_dir(inode, stbuf, xattr);
         if (!is_dir) {
             gf_msg_debug(this->name, 0,
-                         "lookup of %s on %s returned non"
-                         "dir 0%o"
+                         "%s: lookup on %s returned non dir 0%o"
                          "calling lookup_everywhere",
                          local->loc.path, prev->name, stbuf->ia_type);
 
-            local->need_selfheal = 1;
+            local->need_lookup_everywhere = 1;
             goto unlock;
         }
 
@@ -1324,19 +1435,31 @@ dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
             dht_aggregate_xattr(local->xattr, xattr);
         }
 
-        if (dict_get(xattr, conf->mds_xattr_key)) {
-            local->mds_subvol = prev;
-            local->mds_stbuf.ia_gid = stbuf->ia_gid;
-            local->mds_stbuf.ia_uid = stbuf->ia_uid;
-            local->mds_stbuf.ia_prot = stbuf->ia_prot;
+        if (__is_root_gfid(stbuf->ia_gfid)) {
+            ret = dht_dir_has_layout(xattr, conf->xattr_name);
+            if (ret >= 0) {
+                if (is_greater_time(local->prebuf.ia_ctime,
+                                    local->prebuf.ia_ctime_nsec,
+                                    stbuf->ia_ctime, stbuf->ia_ctime_nsec)) {
+                    /* Choose source */
+                    local->prebuf.ia_gid = stbuf->ia_gid;
+                    local->prebuf.ia_uid = stbuf->ia_uid;
+
+                    local->prebuf.ia_ctime = stbuf->ia_ctime;
+                    local->prebuf.ia_ctime_nsec = stbuf->ia_ctime_nsec;
+                    local->prebuf.ia_prot = stbuf->ia_prot;
+                }
+            }
         }
 
         if (local->stbuf.ia_type != IA_INVAL) {
-            if (!__is_root_gfid(stbuf->ia_gfid) &&
-                ((local->stbuf.ia_gid != stbuf->ia_gid) ||
-                 (local->stbuf.ia_uid != stbuf->ia_uid) ||
-                 (is_permission_different(&local->stbuf.ia_prot,
-                                          &stbuf->ia_prot)))) {
+            /* This is not the first subvol to respond
+             * Compare values to see if attrs need to be healed
+             */
+            if ((local->stbuf.ia_gid != stbuf->ia_gid) ||
+                (local->stbuf.ia_uid != stbuf->ia_uid) ||
+                (is_permission_different(&local->stbuf.ia_prot,
+                                         &stbuf->ia_prot))) {
                 local->need_attrheal = 1;
             }
         }
@@ -1349,125 +1472,99 @@ dht_lookup_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
         if (!dict_get(xattr, conf->mds_xattr_key)) {
             gf_msg_debug(this->name, 0,
-                         "Internal xattr %s is not present "
-                         " on path %s gfid is %s ",
-                         conf->mds_xattr_key, local->loc.path, gfid_local);
+                         "%s: mds xattr %s is not present "
+                         "on %s(gfid = %s)",
+                         local->loc.path, conf->mds_xattr_key, prev->name,
+                         gfid_local);
             goto unlock;
-        } else {
-            /* Save mds subvol on inode ctx */
-            ret = dht_inode_ctx_mdsvol_set(local->inode, this, prev);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, 0,
-                       DHT_MSG_SET_INODE_CTX_FAILED,
-                       "Failed to set hashed subvol for %s vol is %s",
-                       local->loc.path, prev->name);
-            }
+        }
+
+        /* Save the mds subvol info and stbuf. This is the value that will
+         * be used for healing
+         */
+        local->mds_subvol = prev;
+        local->mds_stbuf = *stbuf;
+
+        /* Save mds subvol on inode ctx */
+
+        ret = dht_inode_ctx_mdsvol_set(local->inode, this, prev);
+        if (ret) {
+            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SET_INODE_CTX_FAILED,
+                   "%s: Failed to set mds (%s)", local->loc.path, prev->name);
         }
         check_mds = dht_dict_get_array(xattr, conf->mds_xattr_key,
                                        mds_xattr_val, 1, &errst);
         if ((check_mds < 0) && !errst) {
+            /* Check if xattrs need to be healed on the directories */
             local->mds_xattr = dict_ref(xattr);
             gf_msg_debug(this->name, 0,
-                         "Value of %s is not zero on hashed subvol "
-                         "so xattr needs to be heal on non hashed"
-                         " path is %s and vol name is %s "
-                         " gfid is %s",
-                         conf->mds_xattr_key, local->loc.path, prev->name,
+                         "%s: %s is not zero on %s. Xattrs need to be healed."
+                         "(gfid = %s)",
+                         local->loc.path, conf->mds_xattr_key, prev->name,
                          gfid_local);
             local->need_xattr_heal = 1;
-            local->mds_subvol = prev;
         }
     }
+
 unlock:
     UNLOCK(&frame->lock);
 
     this_call_cnt = dht_frame_return(frame);
 
     if (is_last_call(this_call_cnt)) {
+        /* If the mds subvol is not set correctly*/
+        if (!__is_root_gfid(local->gfid) &&
+            (!dict_get(local->xattr, conf->mds_xattr_key))) {
+            local->need_selfheal = 1;
+        }
+
         /* No need to call xattr heal code if volume count is 1
          */
-        if (conf->subvolume_cnt == 1)
+        if (conf->subvolume_cnt == 1) {
             local->need_xattr_heal = 0;
+        }
 
-        /* Code to update all extended attributed from hashed subvol
-           to local->xattr
-        */
-        if (local->need_xattr_heal && (local->mds_xattr)) {
-            dht_dir_set_heal_xattr(this, local, local->xattr, local->mds_xattr,
-                                   NULL, NULL);
-            dict_unref(local->mds_xattr);
-            local->mds_xattr = NULL;
+        if (local->need_selfheal || local->need_lookup_everywhere) {
+            /* Set the gfid-req so posix will set the GFID*/
+            if (!gf_uuid_is_null(local->gfid)) {
+                /* Ok, this should _never_ happen */
+                ret = dict_set_static_bin(local->xattr_req, "gfid-req",
+                                          local->gfid, 16);
+            } else {
+                if (!gf_uuid_is_null(local->gfid_req))
+                    ret = dict_set_static_bin(local->xattr_req, "gfid-req",
+                                              local->gfid_req, 16);
+            }
         }
 
-        if (local->need_selfheal) {
-            local->need_selfheal = 0;
+        if (local->need_lookup_everywhere) {
+            local->need_lookup_everywhere = 0;
             dht_lookup_everywhere(frame, this, &local->loc);
             return 0;
         }
 
         if (local->op_ret == 0) {
-            ret = dht_layout_normalize(this, &local->loc, layout);
-
-            if (ret != 0) {
-                gf_msg_debug(this->name, 0, "fixing assignment on %s",
-                             local->loc.path);
+            if (dht_needs_selfheal(frame, this)) {
                 goto selfheal;
             }
 
             dht_layout_set(this, local->inode, layout);
-            if (!dict_get(local->xattr, conf->mds_xattr_key) ||
-                local->need_xattr_heal)
-                goto selfheal;
-        }
-
-        if (local->inode) {
-            dht_inode_ctx_time_update(local->inode, this, &local->stbuf, 1);
-        }
-
-        if (local->loc.parent) {
-            dht_inode_ctx_time_update(local->loc.parent, this,
-                                      &local->postparent, 1);
-        }
-
-        if (local->need_attrheal) {
-            local->need_attrheal = 0;
-            if (!__is_root_gfid(inode->gfid)) {
-                local->stbuf.ia_gid = local->mds_stbuf.ia_gid;
-                local->stbuf.ia_uid = local->mds_stbuf.ia_uid;
-                local->stbuf.ia_prot = local->mds_stbuf.ia_prot;
+            if (local->inode) {
+                dht_inode_ctx_time_update(local->inode, this, &local->stbuf, 1);
             }
-            copy = create_frame(this, this->ctx->pool);
-            if (copy) {
-                copy_local = dht_local_init(copy, &local->loc, NULL, 0);
-                if (!copy_local) {
-                    DHT_STACK_DESTROY(copy);
-                    goto skip_attr_heal;
-                }
-                copy_local->stbuf = local->stbuf;
-                gf_uuid_copy(copy_local->loc.gfid, local->stbuf.ia_gfid);
-                copy_local->mds_stbuf = local->mds_stbuf;
-                copy_local->mds_subvol = local->mds_subvol;
-                copy->local = copy_local;
-                FRAME_SU_DO(copy, dht_local_t);
-                ret = synctask_new(this->ctx->env, dht_dir_attr_heal,
-                                   dht_dir_attr_heal_done, copy, copy);
-                if (ret) {
-                    gf_msg(this->name, GF_LOG_ERROR, ENOMEM,
-                           DHT_MSG_DIR_ATTR_HEAL_FAILED,
-                           "Synctask creation failed to heal attr "
-                           "for path %s gfid %s ",
-                           local->loc.path, local->gfid);
-                    DHT_STACK_DESTROY(copy);
-                }
+
+            if (local->loc.parent) {
+                dht_inode_ctx_time_update(local->loc.parent, this,
+                                          &local->postparent, 1);
             }
         }
 
-    skip_attr_heal:
         DHT_STRIP_PHASE1_FLAGS(&local->stbuf);
         dht_set_fixed_dir_stat(&local->postparent);
         /* Delete mds xattr at the time of STACK UNWIND */
         if (local->xattr)
             GF_REMOVE_INTERNAL_XATTR(conf->mds_xattr_key, local->xattr);
+
         DHT_STACK_UNWIND(lookup, frame, local->op_ret, local->op_errno,
                          local->inode, &local->stbuf, local->xattr,
                          &local->postparent);
@@ -1483,24 +1580,57 @@ out:
     return ret;
 }
 
-int
-is_permission_different(ia_prot_t *prot1, ia_prot_t *prot2)
+static int
+dht_lookup_directory(call_frame_t *frame, xlator_t *this, loc_t *loc)
 {
-    if ((prot1->owner.read != prot2->owner.read) ||
-        (prot1->owner.write != prot2->owner.write) ||
-        (prot1->owner.exec != prot2->owner.exec) ||
-        (prot1->group.read != prot2->group.read) ||
-        (prot1->group.write != prot2->group.write) ||
-        (prot1->group.exec != prot2->group.exec) ||
-        (prot1->other.read != prot2->other.read) ||
-        (prot1->other.write != prot2->other.write) ||
-        (prot1->other.exec != prot2->other.exec) ||
-        (prot1->suid != prot2->suid) || (prot1->sgid != prot2->sgid) ||
-        (prot1->sticky != prot2->sticky)) {
-        return 1;
-    } else {
-        return 0;
+    int call_cnt = 0;
+    int i = 0;
+    dht_conf_t *conf = NULL;
+    dht_local_t *local = NULL;
+    int ret = 0;
+
+    GF_VALIDATE_OR_GOTO("dht", frame, out);
+    GF_VALIDATE_OR_GOTO("dht", this, unwind);
+    GF_VALIDATE_OR_GOTO("dht", frame->local, unwind);
+    GF_VALIDATE_OR_GOTO("dht", this->private, unwind);
+    GF_VALIDATE_OR_GOTO("dht", loc, unwind);
+
+    conf = this->private;
+    local = frame->local;
+
+    call_cnt = conf->subvolume_cnt;
+    local->call_cnt = call_cnt;
+
+    local->layout = dht_layout_new(this, conf->subvolume_cnt);
+    if (!local->layout) {
+        goto unwind;
     }
+
+    if (local->xattr != NULL) {
+        dict_unref(local->xattr);
+        local->xattr = NULL;
+    }
+
+    if (!gf_uuid_is_null(local->gfid)) {
+        /* use this gfid in order to heal any missing ones */
+        ret = dict_set_gfuuid(local->xattr_req, "gfid-req", local->gfid, true);
+        if (ret)
+            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                   "%s: Failed to set dictionary value:"
+                   " key = gfid-req",
+                   local->loc.path);
+    }
+
+    for (i = 0; i < call_cnt; i++) {
+        STACK_WIND_COOKIE(
+            frame, dht_lookup_dir_cbk, conf->subvolumes[i], conf->subvolumes[i],
+            conf->subvolumes[i]->fops->lookup, &local->loc, local->xattr_req);
+    }
+    return 0;
+unwind:
+    DHT_STACK_UNWIND(lookup, frame, -1, ENOMEM, NULL, NULL, NULL, NULL);
+out:
+    return 0;
 }
 
 int
@@ -1517,13 +1647,11 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int is_dir = 0;
     int is_linkfile = 0;
     int follow_link = 0;
-    call_frame_t *copy = NULL;
-    dht_local_t *copy_local = NULL;
     char gfid[GF_UUID_BUF_SIZE] = {0};
     uint32_t vol_commit_hash = 0;
     xlator_t *subvol = NULL;
     int32_t check_mds = 0;
-    int errst = 0;
+    int errst = 0, i = 0;
     int32_t mds_xattr_val[1] = {0};
 
     GF_VALIDATE_OR_GOTO("dht", frame, err);
@@ -1537,6 +1665,8 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     conf = this->private;
 
     if (!conf->vch_forced) {
+        /* Update the commithash value if available
+         */
         ret = dict_get_uint32(xattr, conf->commithash_xattr_name,
                               &vol_commit_hash);
         if (ret == 0) {
@@ -1546,17 +1676,16 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     gf_uuid_unparse(local->loc.gfid, gfid);
 
+    gf_msg_debug(this->name, op_errno,
+                 "%s: revalidate lookup on %s returned op_ret %d",
+                 local->loc.path, prev->name, op_ret);
+
     LOCK(&frame->lock);
     {
         if (gf_uuid_is_null(local->gfid)) {
             memcpy(local->gfid, local->loc.gfid, 16);
         }
 
-        gf_msg_debug(this->name, op_errno,
-                     "revalidate lookup of %s "
-                     "returned with op_ret %d",
-                     local->loc.path, op_ret);
-
         if (op_ret == -1) {
             local->op_errno = op_errno;
 
@@ -1588,8 +1717,27 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                                  local->loc.path);
 
                     local->need_lookup_everywhere = 1;
+                } else if (IA_ISDIR(local->loc.inode->ia_type)) {
+                    layout = local->layout;
+                    for (i = 0; i < layout->cnt; i++) {
+                        if (layout->list[i].xlator == prev) {
+                            layout->list[i].err = op_errno;
+                            break;
+                        }
+                    }
+
+                    local->need_selfheal = 1;
                 }
             }
+
+            /* The GFID is missing on this subvol. Lookup everywhere to force a
+             * gfid heal
+             */
+            if ((op_errno == ENODATA) &&
+                (IA_ISDIR(local->loc.inode->ia_type))) {
+                local->need_lookup_everywhere = 1;
+            }
+
             goto unlock;
         }
 
@@ -1639,15 +1787,16 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                     (local->stbuf.ia_uid != stbuf->ia_uid) ||
                     is_permission_different(&local->stbuf.ia_prot,
                                             &stbuf->ia_prot)) {
-                    local->need_selfheal = 1;
+                    local->need_attrheal = 1;
                 }
             }
 
             if (!dict_get(xattr, conf->mds_xattr_key)) {
                 gf_msg_debug(this->name, 0,
-                             "internal xattr %s is not present"
-                             " on path %s gfid is %s ",
-                             conf->mds_xattr_key, local->loc.path, gfid);
+                             "%s: internal xattr %s is not present"
+                             " on subvol %s(gfid is %s)",
+                             local->loc.path, conf->mds_xattr_key, prev->name,
+                             gfid);
             } else {
                 check_mds = dht_dict_get_array(xattr, conf->mds_xattr_key,
                                                mds_xattr_val, 1, &errst);
@@ -1665,6 +1814,8 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                            local->loc.path, prev->name);
                 }
                 if ((check_mds < 0) && !errst) {
+                    /* Check if xattrs need to be healed on the directory
+                     */
                     local->mds_xattr = dict_ref(xattr);
                     gf_msg_debug(this->name, 0,
                                  "Value of %s is not zero on "
@@ -1680,6 +1831,8 @@ dht_revalidate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
             ret = dht_layout_dir_mismatch(this, layout, prev, &local->loc,
                                           xattr);
             if (ret != 0) {
+                /* In memory layout does not match on-disk layout.
+                 */
                 gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LAYOUT_MISMATCH,
                        "Mismatching layouts for %s, gfid = %s", local->loc.path,
                        gfid);
@@ -1706,6 +1859,8 @@ unlock:
     UNLOCK(&frame->lock);
 
     if (follow_link) {
+        /* Found a linkto file. Follow it to see if the target file exists
+         */
         gf_uuid_copy(local->gfid, stbuf->ia_gfid);
 
         subvol = dht_linkfile_subvol(this, inode, stbuf, xattr);
@@ -1735,71 +1890,31 @@ unlock:
             local->need_xattr_heal = 0;
 
         if (IA_ISDIR(local->stbuf.ia_type)) {
-            /* Code to update all extended attributed from hashed
-               subvol to local->xattr and call heal code to heal
-               custom xattr from hashed subvol to non-hashed subvol
-            */
-            if (local->need_xattr_heal && (local->mds_xattr)) {
-                dht_dir_set_heal_xattr(this, local, local->xattr,
-                                       local->mds_xattr, NULL, NULL);
-                dict_unref(local->mds_xattr);
-                local->mds_xattr = NULL;
-                local->need_xattr_heal = 0;
-                ret = dht_dir_xattr_heal(this, local);
-                if (ret)
-                    gf_msg(this->name, GF_LOG_ERROR, ret,
-                           DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                           "xattr heal failed for directory %s "
-                           " gfid %s ",
-                           local->loc.path, gfid);
-            } else {
-                /* Call function to save hashed subvol on inode
-                   ctx if internal mds xattr is not present and
-                   all subvols are up
-                */
-                if (inode && !__is_root_gfid(inode->gfid) && (!local->op_ret))
-                    (void)dht_common_mark_mdsxattr(frame, NULL, 1);
-            }
-        }
-        if (local->need_selfheal) {
-            local->need_selfheal = 0;
-            if (!__is_root_gfid(inode->gfid)) {
-                gf_uuid_copy(local->gfid, local->mds_stbuf.ia_gfid);
-                local->stbuf.ia_gid = local->mds_stbuf.ia_gid;
-                local->stbuf.ia_uid = local->mds_stbuf.ia_uid;
-                local->stbuf.ia_prot = local->mds_stbuf.ia_prot;
-            } else {
-                gf_uuid_copy(local->gfid, local->stbuf.ia_gfid);
-                local->stbuf.ia_gid = local->prebuf.ia_gid;
-                local->stbuf.ia_uid = local->prebuf.ia_uid;
-                local->stbuf.ia_prot = local->prebuf.ia_prot;
-            }
-
-            copy = create_frame(this, this->ctx->pool);
-            if (copy) {
-                copy_local = dht_local_init(copy, &local->loc, NULL, 0);
-                if (!copy_local) {
-                    DHT_STACK_DESTROY(copy);
-                    goto cont;
-                }
-                copy_local->stbuf = local->stbuf;
-                copy_local->mds_stbuf = local->mds_stbuf;
-                copy_local->mds_subvol = local->mds_subvol;
-                copy->local = copy_local;
-                FRAME_SU_DO(copy, dht_local_t);
-                ret = synctask_new(this->ctx->env, dht_dir_attr_heal,
-                                   dht_dir_attr_heal_done, copy, copy);
-                if (ret) {
-                    gf_msg(this->name, GF_LOG_ERROR, ENOMEM,
-                           DHT_MSG_DIR_ATTR_HEAL_FAILED,
-                           "Synctask creation failed to heal attr "
-                           "for path %s gfid %s ",
-                           local->loc.path, local->gfid);
-                    DHT_STACK_DESTROY(copy);
+            /* No mds xattr found. Trigger a heal to set it */
+            if (!__is_root_gfid(local->loc.inode->gfid) &&
+                (!dict_get(local->xattr, conf->mds_xattr_key)))
+                local->need_selfheal = 1;
+
+            if (dht_needs_selfheal(frame, this)) {
+                if (!__is_root_gfid(local->loc.inode->gfid)) {
+                    if (local->mds_subvol) {
+                        local->stbuf.ia_gid = local->mds_stbuf.ia_gid;
+                        local->stbuf.ia_uid = local->mds_stbuf.ia_uid;
+                        local->stbuf.ia_prot = local->mds_stbuf.ia_prot;
+                    }
+                } else {
+                    local->stbuf.ia_gid = local->prebuf.ia_gid;
+                    local->stbuf.ia_uid = local->prebuf.ia_uid;
+                    local->stbuf.ia_prot = local->prebuf.ia_prot;
                 }
+
+                layout = local->layout;
+                dht_selfheal_directory(frame, dht_lookup_selfheal_cbk,
+                                       &local->loc, layout);
+                return 0;
             }
         }
-    cont:
+
         if (local->layout_mismatch) {
             /* Found layout mismatch in the directory, need to
                fix this in the inode context */
@@ -1815,9 +1930,16 @@ unlock:
             dht_layout_unref(this, local->layout);
             local->layout = NULL;
 
-            /* We know that current cached subvol is no more
+            /* We know that current cached subvol is no longer
                valid, get the new one */
             local->cached_subvol = NULL;
+            if (local->xattr_req) {
+                if (!gf_uuid_is_null(local->gfid)) {
+                    ret = dict_set_static_bin(local->xattr_req, "gfid-req",
+                                              local->gfid, 16);
+                }
+            }
+
             dht_lookup_everywhere(frame, this, &local->loc);
             return 0;
         }
@@ -1858,12 +1980,11 @@ err:
     return ret;
 }
 
-int
-dht_lookup_linkfile_create_cbk(call_frame_t *frame, void *cookie,
-                               xlator_t *this, int32_t op_ret, int32_t op_errno,
-                               inode_t *inode, struct iatt *stbuf,
-                               struct iatt *preparent, struct iatt *postparent,
-                               dict_t *xdata)
+static int
+dht_lookup_linkfile_create_cbk(call_frame_t *frame, void *cooie, xlator_t *this,
+                               int32_t op_ret, int32_t op_errno, inode_t *inode,
+                               struct iatt *stbuf, struct iatt *preparent,
+                               struct iatt *postparent, dict_t *xdata)
 {
     dht_local_t *local = NULL;
     xlator_t *cached_subvol = NULL;
@@ -1925,7 +2046,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_lookup_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int op_ret, int op_errno, struct iatt *preparent,
                       struct iatt *postparent, dict_t *xdata)
@@ -1951,7 +2072,7 @@ dht_lookup_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_lookup_unlink_of_false_linkto_cbk(call_frame_t *frame, void *cookie,
                                       xlator_t *this, int op_ret, int op_errno,
                                       struct iatt *preparent,
@@ -2003,7 +2124,7 @@ dht_lookup_unlink_of_false_linkto_cbk(call_frame_t *frame, void *cookie,
     return 0;
 }
 
-int
+static int
 dht_lookup_unlink_stale_linkto_cbk(call_frame_t *frame, void *cookie,
                                    xlator_t *this, int op_ret, int op_errno,
                                    struct iatt *preparent,
@@ -2020,52 +2141,41 @@ dht_lookup_unlink_stale_linkto_cbk(call_frame_t *frame, void *cookie,
 
     local = frame->local;
 
-    if (local && local->loc.path)
-        path = local->loc.path;
+    if (local) {
+        FRAME_SU_UNDO(frame, dht_local_t);
+        if (local->loc.path)
+            path = local->loc.path;
+    }
 
     gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_UNLINK_LOOKUP_INFO,
            "Returned with op_ret %d and "
            "op_errno %d for %s",
            op_ret, op_errno, ((path == NULL) ? "null" : path));
 
-    FRAME_SU_UNDO(frame, dht_local_t);
     DHT_STACK_UNWIND(lookup, frame, -1, ENOENT, NULL, NULL, NULL, NULL);
 
     return 0;
 }
 
-int
+static int
 dht_fill_dict_to_avoid_unlink_of_migrating_file(dict_t *dict)
 {
     int ret = 0;
-    xlator_t *this = NULL;
-    char *linktoskip_key = NULL;
 
-    this = THIS;
-    GF_VALIDATE_OR_GOTO("dht", this, err);
-
-    if (dht_is_tier_xlator(this))
-        linktoskip_key = TIER_SKIP_NON_LINKTO_UNLINK;
-    else
-        linktoskip_key = DHT_SKIP_NON_LINKTO_UNLINK;
-
-    ret = dict_set_int32(dict, linktoskip_key, 1);
+    ret = dict_set_int32_sizen(dict, DHT_SKIP_NON_LINKTO_UNLINK, 1);
 
     if (ret)
-        goto err;
+        return -1;
 
-    ret = dict_set_int32(dict, DHT_SKIP_OPEN_FD_UNLINK, 1);
+    ret = dict_set_int32_sizen(dict, DHT_SKIP_OPEN_FD_UNLINK, 1);
 
     if (ret)
-        goto err;
+        return -1;
 
     return 0;
-
-err:
-    return -1;
 }
 
-int32_t
+static int32_t
 dht_linkfile_create_lookup_cbk(call_frame_t *frame, void *cookie,
                                xlator_t *this, int32_t op_ret, int32_t op_errno,
                                inode_t *inode, struct iatt *buf, dict_t *xdata,
@@ -2113,7 +2223,7 @@ dht_linkfile_create_lookup_cbk(call_frame_t *frame, void *cookie,
                          "Creating linkto file on %s(hash) to "
                          "%s on %s (gfid = %s)",
                          local->hashed_subvol->name, local->loc.path,
-                         local->cached_subvol->name, gfid);
+                         local->cached_subvol->name, gfid_str);
 
             ret = dht_linkfile_create(frame, dht_lookup_linkfile_create_cbk,
                                       this, local->cached_subvol,
@@ -2139,7 +2249,7 @@ no_linkto:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_call_lookup_linkfile_create(call_frame_t *frame, void *cookie,
                                 xlator_t *this, int32_t op_ret,
                                 int32_t op_errno, dict_t *xdata)
@@ -2225,7 +2335,7 @@ err:
  * dht_lookup_everywhere_done takes decision based on any of the above case
  */
 
-int
+static int
 dht_lookup_everywhere_done(call_frame_t *frame, xlator_t *this)
 {
     int ret = 0;
@@ -2251,6 +2361,16 @@ dht_lookup_everywhere_done(call_frame_t *frame, xlator_t *this)
         DHT_STACK_UNWIND(lookup, frame, -1, EIO, NULL, NULL, NULL, NULL);
         return 0;
     }
+    if (local->op_ret && local->gfid_missing) {
+        if (gf_uuid_is_null(local->gfid_req)) {
+            DHT_STACK_UNWIND(lookup, frame, -1, ENODATA, NULL, NULL, NULL,
+                             NULL);
+            return 0;
+        }
+        /* A hack */
+        dht_lookup_directory(frame, this, &local->loc);
+        return 0;
+    }
 
     if (local->dir_count) {
         dht_lookup_directory(frame, this, &local->loc);
@@ -2529,7 +2649,7 @@ unwind_hashed_and_cached:
     return 0;
 }
 
-int
+static int
 dht_lookup_everywhere_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                           int32_t op_ret, int32_t op_errno, inode_t *inode,
                           struct iatt *buf, dict_t *xattr,
@@ -2570,6 +2690,8 @@ dht_lookup_everywhere_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (op_ret == -1) {
             if (op_errno != ENOENT)
                 local->op_errno = op_errno;
+            if (op_errno == ENODATA)
+                local->gfid_missing = _gf_true;
             goto unlock;
         }
 
@@ -2863,113 +2985,54 @@ out:
     return 0;
 }
 
-int
-dht_lookup_directory(call_frame_t *frame, xlator_t *this, loc_t *loc)
-{
-    int call_cnt = 0;
-    int i = 0;
-    dht_conf_t *conf = NULL;
-    dht_local_t *local = NULL;
-    int ret = 0;
-
-    GF_VALIDATE_OR_GOTO("dht", frame, out);
-    GF_VALIDATE_OR_GOTO("dht", this, unwind);
-    GF_VALIDATE_OR_GOTO("dht", frame->local, unwind);
-    GF_VALIDATE_OR_GOTO("dht", this->private, unwind);
-    GF_VALIDATE_OR_GOTO("dht", loc, unwind);
-
-    conf = this->private;
-    local = frame->local;
-
-    call_cnt = conf->subvolume_cnt;
-    local->call_cnt = call_cnt;
-
-    local->layout = dht_layout_new(this, conf->subvolume_cnt);
-    if (!local->layout) {
-        goto unwind;
-    }
-
-    if (local->xattr != NULL) {
-        dict_unref(local->xattr);
-        local->xattr = NULL;
-    }
-
-    if (!gf_uuid_is_null(local->gfid)) {
-        ret = dict_set_gfuuid(local->xattr_req, "gfid-req", local->gfid, true);
-        if (ret)
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-                   "%s: Failed to set dictionary value:"
-                   " key = gfid-req",
-                   local->loc.path);
-    }
-
-    for (i = 0; i < call_cnt; i++) {
-        STACK_WIND_COOKIE(
-            frame, dht_lookup_dir_cbk, conf->subvolumes[i], conf->subvolumes[i],
-            conf->subvolumes[i]->fops->lookup, &local->loc, local->xattr_req);
-    }
-    return 0;
-unwind:
-    DHT_STACK_UNWIND(lookup, frame, -1, ENOMEM, NULL, NULL, NULL, NULL);
-out:
-    return 0;
-}
-
 /* Code to get hashed subvol based on inode and loc
    First it check if loc->parent and loc->path exist then it get
    hashed subvol based on loc.
 */
 
-xlator_t *
-dht_inode_get_hashed_subvol(inode_t *inode, xlator_t *this, loc_t *loc)
+static gf_boolean_t
+dht_should_lookup_everywhere(xlator_t *this, dht_conf_t *conf, loc_t *loc)
 {
-    char *path = NULL;
-    loc_t populate_loc = {
-        0,
-    };
-    char *name = NULL;
-    xlator_t *hash_subvol = NULL;
-
-    if (!inode)
-        return hash_subvol;
+    dht_layout_t *parent_layout = NULL;
+    int ret = 0;
+    gf_boolean_t lookup_everywhere = _gf_true;
+
+    /* lookup-optimize supersedes lookup-unhashed settings.
+     * If it is set, do not process search_unhashed
+     * If lookup-optimize if enabled, lookup everywhere if:
+     *  - this is the rebalance daemon.
+     *  - loc->parent is unavailable.
+     *  - parent_layout is unavailable
+     *  - parent_layout->commit_hash != conf->vol_commit_hash
+     */
 
-    if (loc && loc->parent && loc->path) {
-        if (!loc->name) {
-            name = strrchr(loc->path, '/');
-            if (name) {
-                loc->name = name + 1;
-            } else {
-                goto out;
+    if (conf->lookup_optimize) {
+        if (!conf->defrag && loc->parent) {
+            ret = dht_inode_ctx_layout_get(loc->parent, this, &parent_layout);
+            if (!ret && parent_layout &&
+                (parent_layout->commit_hash == conf->vol_commit_hash)) {
+                lookup_everywhere = _gf_false;
             }
         }
-        hash_subvol = dht_subvol_get_hashed(this, loc);
         goto out;
-    }
-
-    if (!gf_uuid_is_null(inode->gfid)) {
-        populate_loc.inode = inode_ref(inode);
-        populate_loc.parent = inode_parent(populate_loc.inode, NULL, NULL);
-        inode_path(populate_loc.inode, NULL, &path);
-
-        if (!path)
-            goto out;
-
-        populate_loc.path = path;
-        if (!populate_loc.name && populate_loc.path) {
-            name = strrchr(populate_loc.path, '/');
-            if (name) {
-                populate_loc.name = name + 1;
-
+    } else {
+        if (conf->search_unhashed == GF_DHT_LOOKUP_UNHASHED_AUTO) {
+            if (loc->parent) {
+                ret = dht_inode_ctx_layout_get(loc->parent, this,
+                                               &parent_layout);
+                if (ret || !parent_layout ||
+                    (!parent_layout->search_unhashed)) {
+                    lookup_everywhere = _gf_false;
+                }
             } else {
-                goto out;
+                lookup_everywhere = _gf_false;
             }
+
+            goto out;
         }
-        hash_subvol = dht_subvol_get_hashed(this, &populate_loc);
     }
 out:
-    if (populate_loc.inode)
-        loc_wipe(&populate_loc);
-    return hash_subvol;
+    return lookup_everywhere;
 }
 
 int
@@ -2985,7 +3048,6 @@ dht_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     loc_t *loc = NULL;
     xlator_t *prev = NULL;
     int ret = 0;
-    dht_layout_t *parent_layout = NULL;
     uint32_t vol_commit_hash = 0;
 
     GF_VALIDATE_OR_GOTO("dht", frame, err);
@@ -3000,104 +3062,82 @@ dht_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     local = frame->local;
     loc = &local->loc;
 
-    /* This is required for handling stale linkfile deletion,
-     * or any more call which happens from this 'loc'.
-     */
-    if (!op_ret && gf_uuid_is_null(local->gfid))
-        memcpy(local->gfid, stbuf->ia_gfid, 16);
-
     gf_msg_debug(this->name, op_errno,
-                 "fresh_lookup returned for %s with op_ret %d", loc->path,
-                 op_ret);
+                 "%s: fresh_lookup on %s returned with op_ret %d", loc->path,
+                 prev->name, op_ret);
 
-    if (!conf->vch_forced) {
-        ret = dict_get_uint32(xattr, conf->commithash_xattr_name,
-                              &vol_commit_hash);
-        if (ret == 0) {
-            conf->vol_commit_hash = vol_commit_hash;
-        }
-    }
-
-    if (ENTRY_MISSING(op_ret, op_errno)) {
-        if (1 == conf->subvolume_cnt) {
-            /* No need to lookup again */
-            goto out;
-        }
+    if (op_ret == -1) {
+        if (ENTRY_MISSING(op_ret, op_errno)) {
+            if (1 == conf->subvolume_cnt) {
+                /* No need to lookup again */
+                goto out;
+            }
 
-        gf_msg_debug(this->name, 0, "Entry %s missing on subvol %s", loc->path,
-                     prev->name);
+            gf_msg_debug(this->name, 0, "Entry %s missing on subvol %s",
+                         loc->path, prev->name);
 
-        /* lookup-optimize supersedes lookup-unhashed settings,
-         *   - so if it is set, do not process search_unhashed
-         *   - except, in the case of rebalance daemon, we want to
-         *     force the lookup_everywhere behavior */
-        if (!conf->defrag && conf->lookup_optimize && loc->parent) {
-            ret = dht_inode_ctx_layout_get(loc->parent, this, &parent_layout);
-            if (ret || !parent_layout ||
-                (parent_layout->commit_hash != conf->vol_commit_hash)) {
-                gf_msg_debug(this->name, 0,
-                             "hashes don't match (ret - %d,"
-                             " parent_layout - %p, parent_hash - %x,"
-                             " vol_hash - %x), do global lookup",
-                             ret, parent_layout,
-                             (parent_layout ? parent_layout->commit_hash : -1),
-                             conf->vol_commit_hash);
+            if (dht_should_lookup_everywhere(this, conf, loc)) {
                 local->op_errno = ENOENT;
                 dht_lookup_everywhere(frame, this, loc);
                 return 0;
             }
+
         } else {
-            if (conf->search_unhashed == GF_DHT_LOOKUP_UNHASHED_ON) {
-                local->op_errno = ENOENT;
-                dht_lookup_everywhere(frame, this, loc);
+            /* posix returns ENODATA if the gfid is not set but the client and
+             * server protocol layers do not send the stbuf. We need to
+             * heal this so check if this is a directory on the other subvols.
+             */
+            if ((op_errno == ENOTCONN) || (op_errno == ENODATA)) {
+                dht_lookup_directory(frame, this, &local->loc);
                 return 0;
             }
-
-            if ((conf->search_unhashed == GF_DHT_LOOKUP_UNHASHED_AUTO) &&
-                (loc->parent)) {
-                ret = dht_inode_ctx_layout_get(loc->parent, this,
-                                               &parent_layout);
-                if (ret || !parent_layout)
-                    goto out;
-                if (parent_layout->search_unhashed) {
-                    local->op_errno = ENOENT;
-                    dht_lookup_everywhere(frame, this, loc);
-                    return 0;
-                }
-            }
         }
+        gf_msg_debug(this->name, op_errno, "%s: Lookup on subvolume %s failed",
+                     loc->path, prev->name);
+        goto out;
     }
 
-    if (op_ret == 0) {
-        is_dir = check_is_dir(inode, stbuf, xattr);
-        if (is_dir) {
-            local->inode = inode_ref(inode);
-            local->xattr = dict_ref(xattr);
+    /* Lookup succeeded - op_ret = 0 */
+
+    /* This is required for handling stale linkfile deletion,
+     * or any more call which happens from this 'loc'.
+     */
+    if (gf_uuid_is_null(local->gfid)) {
+        /*This is set from the first successful response*/
+        memcpy(local->gfid, stbuf->ia_gfid, 16);
+    }
+
+    if (!conf->vch_forced) {
+        /* Update the commit hash in conf if it is found */
+        ret = dict_get_uint32(xattr, conf->commithash_xattr_name,
+                              &vol_commit_hash);
+        if (ret == 0) {
+            conf->vol_commit_hash = vol_commit_hash;
         }
     }
 
-    if (is_dir || (op_ret == -1 && op_errno == ENOTCONN)) {
+    is_dir = check_is_dir(inode, stbuf, xattr);
+    if (is_dir) {
+        /* A directory is present on all subvols, send the lookup to
+         * all subvols now */
+        local->inode = inode_ref(inode);
+        local->xattr = dict_ref(xattr);
         dht_lookup_directory(frame, this, &local->loc);
         return 0;
     }
 
-    if (op_ret == -1) {
-        gf_msg_debug(this->name, op_errno,
-                     "Lookup of %s for subvolume"
-                     " %s failed",
-                     loc->path, prev->name);
-        goto out;
-    }
-
     is_linkfile = check_is_linkfile(inode, stbuf, xattr, conf->link_xattr_name);
 
     if (!is_linkfile) {
-        /* non-directory and not a linkfile */
+        /* non-directory and not a linkto file. This is a data file
+         * Update the layout to point to the cached subvol
+         */
 
         ret = dht_layout_preset(this, prev, inode);
         if (ret < 0) {
             gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LAYOUT_PRESET_FAILED,
-                   "could not set pre-set layout for subvolume %s", prev->name);
+                   "%s: could not set pre-set layout for subvolume %s",
+                   loc->path, prev->name);
             op_ret = -1;
             op_errno = EINVAL;
             goto out;
@@ -3105,22 +3145,19 @@ dht_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
         goto out;
     }
 
+    /* This is a linkto file. Get the value of the target subvol from the
+     * linkto xattr and lookup there to see if the file exists
+     */
     subvol = dht_linkfile_subvol(this, inode, stbuf, xattr);
     if (!subvol) {
         gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_SUBVOL_INFO,
-               "linkfile not having link "
-               "subvol for %s",
-               loc->path);
-
-        gf_msg_debug(this->name, 0,
-                     "linkfile not having link subvolume. path=%s", loc->path);
+               "%s: No link subvol for linkto", loc->path);
         dht_lookup_everywhere(frame, this, loc);
         return 0;
     }
 
-    gf_msg_debug(this->name, 0,
-                 "Calling lookup on linkto target %s for path %s", subvol->name,
-                 loc->path);
+    gf_msg_debug(this->name, 0, "%s: Calling lookup on linkto target %s",
+                 loc->path, subvol->name);
 
     STACK_WIND_COOKIE(frame, dht_lookup_linkfile_cbk, subvol, subvol,
                       subvol->fops->lookup, &local->loc, local->xattr_req);
@@ -3147,11 +3184,11 @@ err:
     return 0;
 }
 
-/* For directories, check if acl xattrs have been requested (by the acl xlator),
- * if not, request for them. These xattrs are needed for dht dir self-heal to
- * perform proper self-healing of dirs
+/* For directories, check if acl xattrs have been requested (by the acl
+ * xlator), if not, request for them. These xattrs are needed for dht dir
+ * self-heal to perform proper self-healing of dirs
  */
-void
+static void
 dht_check_and_set_acl_xattr_req(xlator_t *this, dict_t *xattr_req)
 {
     int ret = 0;
@@ -3182,7 +3219,7 @@ dht_check_and_set_acl_xattr_req(xlator_t *this, dict_t *xattr_req)
  * the mds information : trusted.glusterfs.dht.mds
  * the acl info: See above
  */
-int
+static int
 dht_set_dir_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req)
 {
     int ret = -EINVAL;
@@ -3223,50 +3260,109 @@ err:
     return ret;
 }
 
-int
-dht_set_file_xattr_req(xlator_t *this, loc_t *loc, dict_t *xattr_req)
+/* If the hashed subvol is present, send the lookup to only that subvol first.
+ * If no hashed subvol, send a lookup to all subvols and proceed based on the
+ * responses.
+ */
+static int
+dht_do_fresh_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc)
 {
-    int ret = -EINVAL;
+    int ret = -1;
     dht_conf_t *conf = NULL;
+    xlator_t *hashed_subvol = NULL;
+    dht_local_t *local = NULL;
+    int op_errno = -1;
+    int call_cnt = 0;
+    int i = 0;
 
     conf = this->private;
     if (!conf) {
+        op_errno = EINVAL;
         goto err;
     }
 
-    if (!xattr_req) {
+    local = frame->local;
+    if (!local) {
+        op_errno = EINVAL;
         goto err;
     }
 
-    /* Used to check whether this is a linkto file.
-     */
-    ret = dict_set_uint32(xattr_req, conf->link_xattr_name, 256);
-    if (ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value:key = %s for "
-               "path %s",
-               conf->link_xattr_name, loc->path);
+    /* Since we don't know whether this is a file or a directory,
+     * request all xattrs*/
+    ret = dht_set_file_xattr_req(this, loc, local->xattr_req);
+    if (ret) {
+        op_errno = -ret;
         goto err;
     }
 
-    /* This is used to make sure we don't unlink linkto files
-     * which are the target of an ongoing file migration.
-     */
-    ret = dict_set_uint32(xattr_req, GLUSTERFS_OPEN_FD_COUNT, 4);
+    ret = dht_set_dir_xattr_req(this, loc, local->xattr_req);
     if (ret) {
-        gf_msg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value:key = %s for "
-               "path %s",
-               GLUSTERFS_OPEN_FD_COUNT, loc->path);
+        op_errno = -ret;
         goto err;
     }
 
-    ret = 0;
+    /* Fuse sets a random value in gfid-req. If the gfid is missing
+     * on one or more subvols, posix will set the gfid to this value,
+     * causing GFID mismatches for directories. Remove the value fuse
+     * has sent before sending the lookup.
+     */
+    ret = dict_get_gfuuid(local->xattr_req, "gfid-req", &local->gfid_req);
+    if (ret) {
+        gf_msg_debug(this->name, 0, "%s: No gfid-req available", loc->path);
+    } else {
+        dict_del(local->xattr_req, "gfid-req");
+    }
+    /* This should have been set in dht_lookup */
+    hashed_subvol = local->hashed_subvol;
+
+    if (!hashed_subvol) {
+        gf_msg_debug(this->name, 0,
+                     "%s: no subvolume in layout for path, "
+                     "checking on all the subvols to see if "
+                     "it is a directory",
+                     loc->path);
+
+        call_cnt = conf->subvolume_cnt;
+        local->call_cnt = call_cnt;
+
+        /* Allocate a layout. This will be populated and saved in
+         * the dht inode_ctx on successful lookup
+         */
+        local->layout = dht_layout_new(this, conf->subvolume_cnt);
+        if (!local->layout) {
+            op_errno = ENOMEM;
+            goto err;
+        }
+
+        gf_msg_debug(this->name, 0,
+                     "%s: Found null hashed subvol. Calling lookup"
+                     " on all nodes.",
+                     loc->path);
+
+        for (i = 0; i < call_cnt; i++) {
+            STACK_WIND_COOKIE(frame, dht_lookup_dir_cbk, conf->subvolumes[i],
+                              conf->subvolumes[i],
+                              conf->subvolumes[i]->fops->lookup, &local->loc,
+                              local->xattr_req);
+        }
+        return 0;
+    }
+
+    /* if the hashed_subvol is non-null, send the lookup there first so
+     * as to see whether we have a file or a directory */
+    gf_msg_debug(this->name, 0, "%s: Calling fresh lookup on %s", loc->path,
+                 hashed_subvol->name);
+
+    STACK_WIND_COOKIE(frame, dht_lookup_cbk, hashed_subvol, hashed_subvol,
+                      hashed_subvol->fops->lookup, loc, local->xattr_req);
+    return 0;
 err:
-    return ret;
+    op_errno = (op_errno == -1) ? errno : op_errno;
+    DHT_STACK_UNWIND(lookup, frame, -1, op_errno, NULL, NULL, NULL, NULL);
+    return 0;
 }
 
-int
+static int
 dht_do_revalidate(call_frame_t *frame, xlator_t *this, loc_t *loc)
 {
     xlator_t *subvol = NULL;
@@ -3341,6 +3437,11 @@ dht_do_revalidate(call_frame_t *frame, xlator_t *this, loc_t *loc)
         }
         local->mds_subvol = mds_subvol;
         local->call_cnt = conf->subvolume_cnt;
+
+        /* local->call_cnt will change as responses are processed. Always use a
+         * local copy to loop through the STACK_WIND calls
+         */
+
         call_cnt = local->call_cnt;
 
         for (i = 0; i < call_cnt; i++) {
@@ -3374,91 +3475,11 @@ err:
     return 0;
 }
 
-int
-dht_do_fresh_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc)
-{
-    int ret = -1;
-    dht_conf_t *conf = NULL;
-    xlator_t *hashed_subvol = NULL;
-    dht_local_t *local = NULL;
-    int op_errno = -1;
-    int call_cnt = 0;
-    int i = 0;
-
-    conf = this->private;
-    if (!conf) {
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    local = frame->local;
-    if (!local) {
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    /* Since we don't know whether this is a file or a directory,
-     * request all xattrs*/
-    ret = dht_set_file_xattr_req(this, loc, local->xattr_req);
-    if (ret) {
-        op_errno = -ret;
-        goto err;
-    }
-
-    ret = dht_set_dir_xattr_req(this, loc, local->xattr_req);
-    if (ret) {
-        op_errno = -ret;
-        goto err;
-    }
-
-    /* This should have been set in dht_lookup */
-    hashed_subvol = local->hashed_subvol;
-
-    if (!hashed_subvol) {
-        gf_msg_debug(this->name, 0,
-                     "%s: no subvolume in layout for path, "
-                     "checking on all the subvols to see if "
-                     "it is a directory",
-                     loc->path);
-
-        call_cnt = conf->subvolume_cnt;
-        local->call_cnt = call_cnt;
-
-        local->layout = dht_layout_new(this, conf->subvolume_cnt);
-        if (!local->layout) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-
-        gf_msg_debug(this->name, 0,
-                     "%s: Found null hashed subvol. Calling lookup"
-                     " on all nodes.",
-                     loc->path);
-
-        for (i = 0; i < call_cnt; i++) {
-            STACK_WIND_COOKIE(frame, dht_lookup_dir_cbk, conf->subvolumes[i],
-                              conf->subvolumes[i],
-                              conf->subvolumes[i]->fops->lookup, &local->loc,
-                              local->xattr_req);
-        }
-        return 0;
-    }
-
-    /* if we have the hashed_subvol, send the lookup there first so
-     * as to see whether we have a file or a directory */
-    gf_msg_debug(this->name, 0,
-                 "Calling fresh lookup for %s on"
-                 " %s",
-                 loc->path, hashed_subvol->name);
-
-    STACK_WIND_COOKIE(frame, dht_lookup_cbk, hashed_subvol, hashed_subvol,
-                      hashed_subvol->fops->lookup, loc, local->xattr_req);
-    return 0;
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(lookup, frame, -1, op_errno, NULL, NULL, NULL, NULL);
-    return 0;
-}
+/* Depending on the input, decide if this is a:
+ * fresh-lookup: loc->name is provided but no dht inode ctx
+ * revalidation: loc->name is provided, dht inode ctx is present
+ * discover: gfid based nameless lookup.
+ */
 
 int
 dht_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
@@ -3512,6 +3533,10 @@ dht_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
 
     /* Nameless lookup */
 
+    /* This is usually sent by NFS. Lookups are done based on the gfid and
+     * no name information is available. Without the name, dht cannot calculate
+     * the hash and has to send a lookup to all subvols.
+     */
     if (gf_uuid_is_null(loc->pargfid) && !gf_uuid_is_null(loc->gfid) &&
         !__is_root_gfid(loc->inode->gfid)) {
         local->cached_subvol = NULL;
@@ -3520,6 +3545,9 @@ dht_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
     }
 
     if (loc_is_root(loc)) {
+        /* Request the DHT commit hash xattr (trusted.glusterfs.dht.commithash)
+         * set on the brick root.
+         */
         ret = dict_set_uint32(local->xattr_req, conf->commithash_xattr_name,
                               sizeof(uint32_t));
     }
@@ -3528,12 +3556,14 @@ dht_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr_req)
         hashed_subvol = dht_subvol_get_hashed(this, loc);
     local->hashed_subvol = hashed_subvol;
 
-    /* The entry has been looked up before and has an inode_ctx set
-     */
     if (is_revalidate(loc)) {
+        /* The entry has been looked up before and has a dht inode_ctx
+         */
         dht_do_revalidate(frame, this, loc);
         return 0;
     } else {
+        /* Entry has not been looked up before
+         */
         dht_do_fresh_lookup(frame, this, loc);
         return 0;
     }
@@ -3545,7 +3575,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_unlink_linkfile_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                         int op_ret, int op_errno, struct iatt *preparent,
                         struct iatt *postparent, dict_t *xdata)
@@ -3561,18 +3591,16 @@ dht_unlink_linkfile_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if ((op_ret == -1) &&
             !((op_errno == ENOENT) || (op_errno == ENOTCONN))) {
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno,
-                         "Unlink link: subvolume %s"
-                         " returned -1",
-                         prev->name);
-            goto unlock;
+                         "Unlink link: subvolume %s returned -1", prev->name);
+            goto post_unlock;
         }
 
         local->op_ret = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     dht_set_fixed_dir_stat(&local->preparent);
     dht_set_fixed_dir_stat(&local->postparent);
     DHT_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
@@ -3581,7 +3609,7 @@ unlock:
     return 0;
 }
 
-int
+static int
 dht_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                int op_errno, struct iatt *preparent, struct iatt *postparent,
                dict_t *xdata)
@@ -3602,9 +3630,10 @@ dht_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
             } else {
                 local->op_ret = 0;
             }
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno,
                          "Unlink: subvolume %s returned -1", prev->name);
-            goto unlock;
+            goto post_unlock;
         }
 
         local->op_ret = 0;
@@ -3619,9 +3648,8 @@ dht_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                                       &local->postparent, 1);
         }
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     if (!local->op_ret) {
         hashed_subvol = dht_subvol_get_hashed(this, &local->loc);
         if (hashed_subvol && hashed_subvol != local->cached_subvol) {
@@ -3672,7 +3700,7 @@ dht_fix_layout_setxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_err_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
             int op_errno, dict_t *xdata)
 {
@@ -3687,22 +3715,24 @@ dht_err_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     {
         if (op_ret == -1) {
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
                          prev->name);
-            goto unlock;
+            goto post_unlock;
         }
 
         local->op_ret = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt)) {
         if ((local->fop == GF_FOP_SETXATTR) ||
             (local->fop == GF_FOP_FSETXATTR)) {
             DHT_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
                              NULL);
+            /* 'local' itself may not be valid after this */
+            goto out;
         }
         if ((local->fop == GF_FOP_REMOVEXATTR) ||
             (local->fop == GF_FOP_FREMOVEXATTR)) {
@@ -3711,6 +3741,7 @@ unlock:
         }
     }
 
+out:
     return 0;
 }
 
@@ -3741,7 +3772,7 @@ dht_dict_set_array(dict_t *dict, char *key, int32_t value[], int32_t size)
     return ret;
 }
 
-int
+static int
 dht_common_mds_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                            int32_t op_ret, int32_t op_errno, dict_t *dict,
                            dict_t *xdata)
@@ -3757,27 +3788,34 @@ dht_common_mds_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     if (local->fop == GF_FOP_SETXATTR) {
         DHT_STACK_UNWIND(setxattr, frame, 0, op_errno, local->xdata);
+        /* 'local' itself may not be valid after this */
+        goto out;
     }
 
     if (local->fop == GF_FOP_FSETXATTR) {
         DHT_STACK_UNWIND(fsetxattr, frame, 0, op_errno, local->xdata);
+        /* 'local' itself may not be valid after this */
+        goto out;
     }
 
     if (local->fop == GF_FOP_REMOVEXATTR) {
         DHT_STACK_UNWIND(removexattr, frame, 0, op_errno, NULL);
+        /* 'local' itself may not be valid after this */
+        goto out;
     }
 
     if (local->fop == GF_FOP_FREMOVEXATTR) {
         DHT_STACK_UNWIND(fremovexattr, frame, 0, op_errno, NULL);
     }
 
+out:
     return 0;
 }
 
 /* Code to wind a xattrop call to add 1 on current mds internal xattr
    value
 */
-int
+static int
 dht_setxattr_non_mds_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                          int op_ret, int op_errno, dict_t *xdata)
 {
@@ -3798,11 +3836,14 @@ dht_setxattr_non_mds_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (op_ret && !local->op_ret) {
             local->op_ret = op_ret;
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
                          prev->this->name);
+            goto post_unlock;
         }
     }
     UNLOCK(&frame->lock);
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
 
     if (is_last_call(this_call_cnt)) {
@@ -3834,45 +3875,60 @@ dht_setxattr_non_mds_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         } else {
             if (local->fop == GF_FOP_SETXATTR) {
                 DHT_STACK_UNWIND(setxattr, frame, 0, 0, local->xdata);
+                /* 'local' itself may not be valid after this */
+                goto just_return;
             }
 
             if (local->fop == GF_FOP_FSETXATTR) {
                 DHT_STACK_UNWIND(fsetxattr, frame, 0, 0, local->xdata);
+                /* 'local' itself may not be valid after this */
+                goto just_return;
             }
 
             if (local->fop == GF_FOP_REMOVEXATTR) {
                 DHT_STACK_UNWIND(removexattr, frame, 0, 0, NULL);
+                /* 'local' itself may not be valid after this */
+                goto just_return;
             }
 
             if (local->fop == GF_FOP_FREMOVEXATTR) {
                 DHT_STACK_UNWIND(fremovexattr, frame, 0, 0, NULL);
+                /* 'local' itself may not be valid after this */
+                goto just_return;
             }
         }
     }
 out:
-    if (xattrop)
-        dict_unref(xattrop);
     if (ret) {
         if (local->fop == GF_FOP_SETXATTR) {
             DHT_STACK_UNWIND(setxattr, frame, 0, 0, local->xdata);
+            /* 'local' itself may not be valid after this */
+            goto just_return;
         }
 
         if (local->fop == GF_FOP_FSETXATTR) {
             DHT_STACK_UNWIND(fsetxattr, frame, 0, 0, local->xdata);
+            /* 'local' itself may not be valid after this */
+            goto just_return;
         }
 
         if (local->fop == GF_FOP_REMOVEXATTR) {
             DHT_STACK_UNWIND(removexattr, frame, 0, 0, NULL);
+            /* 'local' itself may not be valid after this */
+            goto just_return;
         }
 
         if (local->fop == GF_FOP_FREMOVEXATTR) {
             DHT_STACK_UNWIND(fremovexattr, frame, 0, 0, NULL);
         }
     }
+just_return:
+    if (xattrop)
+        dict_unref(xattrop);
     return 0;
 }
 
-int
+static int
 dht_setxattr_mds_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int op_ret, int op_errno, dict_t *xdata)
 {
@@ -3932,16 +3988,22 @@ out:
     if (local->fop == GF_FOP_SETXATTR) {
         DHT_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
                          xdata);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_FSETXATTR) {
         DHT_STACK_UNWIND(fsetxattr, frame, local->op_ret, local->op_errno,
                          xdata);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_REMOVEXATTR) {
         DHT_STACK_UNWIND(removexattr, frame, local->op_ret, local->op_errno,
                          NULL);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_FREMOVEXATTR) {
@@ -3949,10 +4011,11 @@ out:
                          NULL);
     }
 
+just_return:
     return 0;
 }
 
-int
+static int
 dht_xattrop_mds_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                     int op_ret, int op_errno, dict_t *dict, dict_t *xdata)
 {
@@ -3999,16 +4062,22 @@ out:
     if (local->fop == GF_FOP_SETXATTR) {
         DHT_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
                          xdata);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_FSETXATTR) {
         DHT_STACK_UNWIND(fsetxattr, frame, local->op_ret, local->op_errno,
                          xdata);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_REMOVEXATTR) {
         DHT_STACK_UNWIND(removexattr, frame, local->op_ret, local->op_errno,
                          NULL);
+        /* 'local' itself may not be valid after this */
+        goto just_return;
     }
 
     if (local->fop == GF_FOP_FREMOVEXATTR) {
@@ -4016,6 +4085,7 @@ out:
                          NULL);
     }
 
+just_return:
     return 0;
 }
 
@@ -4058,7 +4128,7 @@ dht_fill_pathinfo_xattr(xlator_t *this, dht_local_t *local, char *xattr_buf,
     }
 }
 
-int
+static int
 dht_vgetxattr_alloc_and_fill(dht_local_t *local, dict_t *xattr, xlator_t *this,
                              int op_errno)
 {
@@ -4107,7 +4177,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_vgetxattr_fill_and_set(dht_local_t *local, dict_t **dict, xlator_t *this,
                            gf_boolean_t flag)
 {
@@ -4163,7 +4233,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_find_local_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                           int op_ret, int op_errno, dict_t *xattr,
                           dict_t *xdata)
@@ -4202,11 +4272,15 @@ dht_find_local_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     {
         this_call_cnt = --local->call_cnt;
         if (op_ret < 0) {
-            gf_msg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_GET_XATTR_FAILED,
-                   "getxattr err for dir");
             local->op_ret = -1;
             local->op_errno = op_errno;
-            goto unlock;
+            UNLOCK(&frame->lock);
+            if (op_errno == ENODATA)
+                gf_msg_debug(this->name, 0, "failed to get node-uuid");
+            else
+                gf_msg(this->name, GF_LOG_ERROR, op_errno,
+                       DHT_MSG_GET_XATTR_FAILED, "failed to get node-uuid");
+            goto post_unlock;
         }
 
         ret = dict_get_str(xattr, local->xsel, &uuid_list);
@@ -4227,18 +4301,19 @@ dht_find_local_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         index = conf->local_subvols_cnt;
 
         uuid_list_copy = gf_strdup(uuid_list);
+        if (!uuid_list_copy)
+            goto unlock;
 
         for (uuid_str = strtok_r(uuid_list, " ", &saveptr); uuid_str;
              uuid_str = next_uuid_str) {
             next_uuid_str = strtok_r(NULL, " ", &saveptr);
             if (gf_uuid_parse(uuid_str, node_uuid)) {
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UUID_PARSE_ERROR,
-                       "Failed to parse uuid"
-                       " for %s",
-                       prev->name);
                 local->op_ret = -1;
                 local->op_errno = EINVAL;
-                goto unlock;
+                UNLOCK(&frame->lock);
+                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UUID_PARSE_ERROR,
+                       "Failed to parse uuid for %s", prev->name);
+                goto post_unlock;
             }
 
             count++;
@@ -4295,7 +4370,7 @@ dht_find_local_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     local->op_ret = 0;
 unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     if (!is_last_call(this_call_cnt))
         goto out;
 
@@ -4317,7 +4392,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_vgetxattr_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int op_ret, int op_errno, dict_t *xattr, dict_t *xdata)
 {
@@ -4336,23 +4411,28 @@ dht_vgetxattr_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         this_call_cnt = --local->call_cnt;
         if (op_ret < 0) {
             if (op_errno != ENOTCONN) {
-                gf_msg(this->name, GF_LOG_ERROR, op_errno,
-                       DHT_MSG_GET_XATTR_FAILED, "getxattr err for dir");
                 local->op_ret = -1;
                 local->op_errno = op_errno;
+                UNLOCK(&frame->lock);
+                gf_msg(this->name, GF_LOG_ERROR, op_errno,
+                       DHT_MSG_GET_XATTR_FAILED, "getxattr err for dir");
+                goto post_unlock;
             }
 
             goto unlock;
         }
 
         ret = dht_vgetxattr_alloc_and_fill(local, xattr, this, op_errno);
-        if (ret)
+        if (ret) {
+            UNLOCK(&frame->lock);
             gf_msg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_DICT_SET_FAILED,
                    "alloc or fill failure");
+            goto post_unlock;
+        }
     }
 unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     if (!is_last_call(this_call_cnt))
         goto out;
 
@@ -4378,7 +4458,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_vgetxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                   int op_errno, dict_t *xattr, dict_t *xdata)
 {
@@ -4424,7 +4504,7 @@ cleanup:
     return 0;
 }
 
-int
+static int
 dht_linkinfo_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                           int op_ret, int op_errno, dict_t *xattr,
                           dict_t *xdata)
@@ -4446,16 +4526,16 @@ dht_linkinfo_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_mds_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int op_ret, int op_errno, dict_t *xattr, dict_t *xdata)
 {
     dht_local_t *local = NULL;
     dht_conf_t *conf = NULL;
 
-    VALIDATE_OR_GOTO(frame, out);
-    VALIDATE_OR_GOTO(frame->local, out);
-    VALIDATE_OR_GOTO(this->private, out);
+    VALIDATE_OR_GOTO(frame, err);
+    VALIDATE_OR_GOTO(frame->local, err);
+    VALIDATE_OR_GOTO(this->private, err);
 
     conf = this->private;
     local = frame->local;
@@ -4464,9 +4544,7 @@ dht_mds_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         local->op_ret = op_ret;
         goto out;
     }
-    if (dict_get(xattr, conf->xattr_name)) {
-        dict_del(xattr, conf->xattr_name);
-    }
+    dict_del(xattr, conf->xattr_name);
     local->op_ret = 0;
 
     if (!local->xattr) {
@@ -4477,6 +4555,9 @@ out:
     DHT_STACK_UNWIND(getxattr, frame, local->op_ret, op_errno, local->xattr,
                      xdata);
     return 0;
+err:
+    DHT_STACK_UNWIND(getxattr, frame, -1, EINVAL, NULL, NULL);
+    return 0;
 }
 
 int
@@ -4486,14 +4567,22 @@ dht_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     int this_call_cnt = 0;
     dht_local_t *local = NULL;
     dht_conf_t *conf = NULL;
+    int ret = 0;
 
-    VALIDATE_OR_GOTO(frame, out);
-    VALIDATE_OR_GOTO(frame->local, out);
-    VALIDATE_OR_GOTO(this->private, out);
+    VALIDATE_OR_GOTO(frame, err);
+    VALIDATE_OR_GOTO(frame->local, err);
+    VALIDATE_OR_GOTO(this->private, err);
 
     conf = this->private;
     local = frame->local;
 
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
+        ret = dht_check_and_open_fd_on_subvol(this, frame);
+        if (ret)
+            goto err;
+        return 0;
+    }
+
     LOCK(&frame->lock);
     {
         if (!xattr || (op_ret == -1)) {
@@ -4501,27 +4590,10 @@ dht_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
             goto unlock;
         }
 
-        if (dict_get(xattr, conf->xattr_name)) {
-            dict_del(xattr, conf->xattr_name);
-        }
-
-        if (dict_get(xattr, conf->mds_xattr_key)) {
-            dict_del(xattr, conf->mds_xattr_key);
-        }
-
-        /* filter out following two xattrs that need not
-         * be visible on the mount point for geo-rep -
-         * trusted.tier.fix.layout.complete and
-         * trusted.tier.tier-dht.commithash
-         */
-
-        if (dict_get(xattr, conf->commithash_xattr_name)) {
-            dict_del(xattr, conf->commithash_xattr_name);
-        }
+        dict_del(xattr, conf->xattr_name);
+        dict_del(xattr, conf->mds_xattr_key);
 
-        if (frame->root->pid >= 0 && dht_is_tier_xlator(this)) {
-            dict_del(xattr, GF_XATTR_TIER_LAYOUT_FIXED_KEY);
-        }
+        dict_del(xattr, conf->commithash_xattr_name);
 
         if (frame->root->pid >= 0) {
             GF_REMOVE_INTERNAL_XATTR("trusted.glusterfs.quota*", xattr);
@@ -4547,7 +4619,6 @@ unlock:
     UNLOCK(&frame->lock);
 
     this_call_cnt = dht_frame_return(frame);
-out:
     if (is_last_call(this_call_cnt)) {
         /* If we have a valid xattr received from any one of the
          * subvolume, let's return it */
@@ -4559,9 +4630,12 @@ out:
                          local->xdata);
     }
     return 0;
+err:
+    DHT_STACK_UNWIND(getxattr, frame, -1, EINVAL, NULL, NULL);
+    return 0;
 }
 
-int32_t
+static int32_t
 dht_getxattr_unwind(call_frame_t *frame, int op_ret, int op_errno, dict_t *dict,
                     dict_t *xdata)
 {
@@ -4569,7 +4643,7 @@ dht_getxattr_unwind(call_frame_t *frame, int op_ret, int op_errno, dict_t *dict,
     return 0;
 }
 
-int
+static int
 dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
                                    xlator_t *this, int op_ret, int op_errno,
                                    dict_t *xattr, dict_t *xdata)
@@ -4581,7 +4655,7 @@ dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
 
     LOCK(&frame->lock);
     {
-        if (local->op_errno == ENODATA || local->op_errno == EOPNOTSUPP) {
+        if (local->op_errno == EOPNOTSUPP) {
             /* Nothing to do here, we have already found
              * a subvol which does not have the get_real_filename
              * optimization. If condition is for simple logic.
@@ -4590,7 +4664,7 @@ dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
         }
 
         if (op_ret == -1) {
-            if (op_errno == ENODATA || op_errno == EOPNOTSUPP) {
+            if (op_errno == EOPNOTSUPP) {
                 /* This subvol does not have the optimization.
                  * Better let the user know we don't support it.
                  * Remove previous results if any.
@@ -4608,16 +4682,17 @@ dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
 
                 local->op_ret = op_ret;
                 local->op_errno = op_errno;
+                UNLOCK(&frame->lock);
                 gf_msg(this->name, GF_LOG_WARNING, op_errno,
                        DHT_MSG_UPGRADE_BRICKS,
                        "At least "
                        "one of the bricks does not support "
                        "this operation. Please upgrade all "
                        "bricks.");
-                goto unlock;
+                goto post_unlock;
             }
 
-            if (op_errno == ENOENT) {
+            if (op_errno == ENOATTR) {
                 /* Do nothing, our defaults are set to this.
                  */
                 goto unlock;
@@ -4629,9 +4704,10 @@ dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
              * down subvol and return a good result(if any)
              * from other subvol.
              */
+            UNLOCK(&frame->lock);
             gf_msg(this->name, GF_LOG_WARNING, op_errno,
                    DHT_MSG_GET_XATTR_FAILED, "Failed to get real filename.");
-            goto unlock;
+            goto post_unlock;
         }
 
         /* This subvol has the required file.
@@ -4652,13 +4728,13 @@ dht_getxattr_get_real_filename_cbk(call_frame_t *frame, void *cookie,
 
         local->op_ret = op_ret;
         local->op_errno = 0;
-        gf_msg_debug(this->name, 0,
-                     "Found a matching "
-                     "file.");
+        UNLOCK(&frame->lock);
+        gf_msg_debug(this->name, 0, "Found a matching file.");
+        goto post_unlock;
     }
 unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt)) {
         DHT_STACK_UNWIND(getxattr, frame, local->op_ret, local->op_errno,
@@ -4668,7 +4744,7 @@ unlock:
     return 0;
 }
 
-int
+static int
 dht_getxattr_get_real_filename(call_frame_t *frame, xlator_t *this, loc_t *loc,
                                const char *key, dict_t *xdata)
 {
@@ -4684,7 +4760,7 @@ dht_getxattr_get_real_filename(call_frame_t *frame, xlator_t *this, loc_t *loc,
     cnt = local->call_cnt = layout->cnt;
 
     local->op_ret = -1;
-    local->op_errno = ENOENT;
+    local->op_errno = ENOATTR;
 
     for (i = 0; i < cnt; i++) {
         subvol = layout->list[i].xlator;
@@ -4695,7 +4771,7 @@ dht_getxattr_get_real_filename(call_frame_t *frame, xlator_t *this, loc_t *loc,
     return 0;
 }
 
-int
+static int
 dht_marker_populate_args(call_frame_t *frame, int type, int *gauge,
                          xlator_t **subvols)
 {
@@ -4712,8 +4788,8 @@ dht_marker_populate_args(call_frame_t *frame, int type, int *gauge,
     return layout->cnt;
 }
 
-int
-dht_is_debug_xattr_key(char **array, char *key)
+static int
+dht_is_debug_xattr_key(const char **array, char *key)
 {
     int i = 0;
 
@@ -4727,7 +4803,7 @@ dht_is_debug_xattr_key(char **array, char *key)
 
 /* Note we already have frame->local initialised here*/
 
-int
+static int
 dht_handle_debug_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
                           const char *key)
 {
@@ -4739,10 +4815,6 @@ dht_handle_debug_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
     const char *name = NULL;
 
     local = frame->local;
-    if (!key) {
-        op_errno = EINVAL;
-        goto out;
-    }
 
     if (dht_is_debug_xattr_key(dht_dbg_vxattrs, (char *)key) == -1) {
         goto out;
@@ -4796,6 +4868,60 @@ out:
     return 0;
 }
 
+/* Virtual Xattr which returns 1 if all subvols are up,
+   else returns 0. Geo-rep then uses this virtual xattr
+   after a fresh mount and starts the I/O.
+*/
+
+enum dht_vxattr_subvol {
+    DHT_VXATTR_SUBVOLS_UP = 1,
+    DHT_VXATTR_SUBVOLS_DOWN = 0,
+};
+
+int
+dht_vgetxattr_subvol_status(call_frame_t *frame, xlator_t *this,
+                            const char *key)
+{
+    dht_local_t *local = NULL;
+    int ret = -1;
+    int op_errno = ENODATA;
+    int value = DHT_VXATTR_SUBVOLS_UP;
+    int i = 0;
+    dht_conf_t *conf = NULL;
+
+    conf = this->private;
+    local = frame->local;
+
+    if (!key) {
+        op_errno = EINVAL;
+        goto out;
+    }
+    local->xattr = dict_new();
+    if (!local->xattr) {
+        op_errno = ENOMEM;
+        goto out;
+    }
+    for (i = 0; i < conf->subvolume_cnt; i++) {
+        if (!conf->subvolume_status[i]) {
+            value = DHT_VXATTR_SUBVOLS_DOWN;
+            gf_msg_debug(this->name, 0, "subvol %s is down ",
+                         conf->subvolumes[i]->name);
+            break;
+        }
+    }
+    ret = dict_set_int8(local->xattr, (char *)key, value);
+    if (ret < 0) {
+        op_errno = -ret;
+        ret = -1;
+        goto out;
+    }
+    ret = 0;
+
+out:
+    DHT_STACK_UNWIND(getxattr, frame, ret, op_errno, local->xattr, NULL);
+    return 0;
+}
+
 int
 dht_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
              dict_t *xdata)
@@ -4853,6 +4979,11 @@ dht_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, const char *key,
         goto err;
     }
 
+    if (strncmp(key, DHT_SUBVOL_STATUS_KEY, SLEN(DHT_SUBVOL_STATUS_KEY)) == 0) {
+        dht_vgetxattr_subvol_status(frame, this, key);
+        return 0;
+    }
+
     /* skip over code which is irrelevant if !DHT_IS_DIR(layout) */
     if (!DHT_IS_DIR(layout))
         goto no_dht_is_dir;
@@ -5085,8 +5216,7 @@ dht_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *key,
         }
     }
 
-    if (fd->inode)
-        gf_uuid_unparse(fd->inode->gfid, gfid);
+    gf_uuid_unparse(fd->inode->gfid, gfid);
 
     if ((fd->inode->ia_type == IA_IFDIR) && key &&
         (strncmp(key, GF_XATTR_LOCKINFO_KEY, SLEN(GF_XATTR_LOCKINFO_KEY)) !=
@@ -5149,6 +5279,53 @@ err:
     return 0;
 }
 
+static int
+dht_setxattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
+{
+    dht_local_t *local = NULL;
+    int op_errno = EINVAL;
+
+    if (!frame || !frame->local)
+        goto err;
+
+    local = frame->local;
+    op_errno = local->op_errno;
+
+    if (we_are_not_migrating(ret)) {
+        /* This dht xlator is not migrating the file. Unwind and
+         * pass on the original mode bits so the higher DHT layer
+         * can handle this.
+         */
+        DHT_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
+                         local->rebalance.xdata);
+        return 0;
+    }
+
+    if (subvol == NULL)
+        goto err;
+
+    local->call_cnt = 2; /* This is the second attempt */
+
+    if (local->fop == GF_FOP_SETXATTR) {
+        STACK_WIND_COOKIE(frame, dht_file_setxattr_cbk, subvol, subvol,
+                          subvol->fops->setxattr, &local->loc,
+                          local->rebalance.xattr, local->rebalance.flags,
+                          local->xattr_req);
+    } else {
+        STACK_WIND_COOKIE(frame, dht_file_setxattr_cbk, subvol, subvol,
+                          subvol->fops->fsetxattr, local->fd,
+                          local->rebalance.xattr, local->rebalance.flags,
+                          local->xattr_req);
+    }
+
+    return 0;
+
+err:
+    DHT_STACK_UNWIND(setxattr, frame, (local ? local->op_ret : -1), op_errno,
+                     NULL);
+    return 0;
+}
+
 int
 dht_file_setxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int op_ret, int op_errno, dict_t *xdata)
@@ -5165,8 +5342,8 @@ dht_file_setxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     local->op_errno = op_errno;
 
-    if ((local->fop == GF_FOP_FSETXATTR) && op_ret == -1 &&
-        (op_errno == EBADF) && !(local->fd_checked)) {
+    if ((local->fop == GF_FOP_FSETXATTR) &&
+        dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -5239,7 +5416,7 @@ dht_is_user_xattr(dict_t *this, char *key, data_t *value, void *data)
 
 /* Common code to wind a (f)(set|remove)xattr call to set xattr on directory
  */
-int
+static int
 dht_dir_common_set_remove_xattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
                                 fd_t *fd, dict_t *xattr, int flags,
                                 dict_t *xdata, int *op_errno)
@@ -5256,11 +5433,13 @@ dht_dir_common_set_remove_xattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
     int call_cnt = 0;
     dht_local_t *local = NULL;
     char gfid_local[GF_UUID_BUF_SIZE] = {0};
+    char **xattrs_to_heal;
 
     conf = this->private;
     local = frame->local;
     call_cnt = conf->subvolume_cnt;
     local->flags = flags;
+    xattrs_to_heal = get_xattrs_to_heal();
 
     if (!gf_uuid_is_null(local->gfid)) {
         gf_uuid_unparse(local->gfid, gfid_local);
@@ -5343,9 +5522,8 @@ dht_dir_common_set_remove_xattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
         } else {
             gf_msg(this->name, GF_LOG_ERROR, 0,
                    DHT_MSG_HASHED_SUBVOL_GET_FAILED,
-                   "Failed to get mds subvol for path %s"
-                   "gfid is %s ",
-                   loc->path, gfid_local);
+                   "%s: Failed to get mds subvol. (gfid is %s)", loc->path,
+                   gfid_local);
         }
         (*op_errno) = ENOENT;
         goto err;
@@ -5493,7 +5671,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_checking_pathinfo_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                           int op_ret, int op_errno, dict_t *xattr,
                           dict_t *xdata)
@@ -5532,54 +5710,7 @@ out:
     return 0;
 }
 
-int
-dht_setxattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
-{
-    dht_local_t *local = NULL;
-    int op_errno = EINVAL;
-
-    if (!frame || !frame->local)
-        goto err;
-
-    local = frame->local;
-    op_errno = local->op_errno;
-
-    if (we_are_not_migrating(ret)) {
-        /* This dht xlator is not migrating the file. Unwind and
-         * pass on the original mode bits so the higher DHT layer
-         * can handle this.
-         */
-        DHT_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
-                         local->rebalance.xdata);
-        return 0;
-    }
-
-    if (subvol == NULL)
-        goto err;
-
-    local->call_cnt = 2; /* This is the second attempt */
-
-    if (local->fop == GF_FOP_SETXATTR) {
-        STACK_WIND_COOKIE(frame, dht_file_setxattr_cbk, subvol, subvol,
-                          subvol->fops->setxattr, &local->loc,
-                          local->rebalance.xattr, local->rebalance.flags,
-                          local->xattr_req);
-    } else {
-        STACK_WIND_COOKIE(frame, dht_file_setxattr_cbk, subvol, subvol,
-                          subvol->fops->fsetxattr, local->fd,
-                          local->rebalance.xattr, local->rebalance.flags,
-                          local->xattr_req);
-    }
-
-    return 0;
-
-err:
-    DHT_STACK_UNWIND(setxattr, frame, (local ? local->op_ret : -1), op_errno,
-                     NULL);
-    return 0;
-}
-
-int
+static int
 dht_nuke_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                  int32_t op_ret, int32_t op_errno, struct iatt *preparent,
                  struct iatt *postparent, dict_t *xdata)
@@ -5588,7 +5719,7 @@ dht_nuke_dir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_nuke_dir(call_frame_t *frame, xlator_t *this, loc_t *loc, data_t *tmp)
 {
     if (!IA_ISDIR(loc->inode->ia_type)) {
@@ -5741,22 +5872,7 @@ dht_setxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xattr,
         if (local->rebalance.target_node) {
             local->flags = forced_rebalance;
 
-            /* Flag to suggest its a tiering migration
-             * The reason for this dic key-value is that
-             * promotions and demotions are multithreaded
-             * so the original frame from gf_defrag_start()
-             * is not carried. A new frame will be created when
-             * we do syncop_setxattr(). This does not have the
-             * frame->root->pid of the original frame. So we pass
-             * this dic key-value when we do syncop_setxattr() to do
-             * data migration and set the frame->root->pid to
-             * GF_CLIENT_PID_TIER_DEFRAG in dht_setxattr() just before
-             * calling dht_start_rebalance_task() */
-            tmp = dict_get(xattr, TIERING_MIGRATION_KEY);
-            if (tmp)
-                frame->root->pid = GF_CLIENT_PID_TIER_DEFRAG;
-            else
-                frame->root->pid = GF_CLIENT_PID_DEFRAG;
+            frame->root->pid = GF_CLIENT_PID_DEFRAG;
 
             ret = dht_start_rebalance_task(this, frame);
             if (!ret)
@@ -5875,6 +5991,50 @@ err:
     return 0;
 }
 
+static int
+dht_removexattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
+{
+    dht_local_t *local = NULL;
+    int op_errno = EINVAL;
+
+    if (!frame || !frame->local)
+        goto err;
+
+    local = frame->local;
+    op_errno = local->op_errno;
+
+    local->call_cnt = 2; /* This is the second attempt */
+
+    if (we_are_not_migrating(ret)) {
+        /* This dht xlator is not migrating the file. Unwind and
+         * pass on the original mode bits so the higher DHT layer
+         * can handle this.
+         */
+        DHT_STACK_UNWIND(removexattr, frame, local->op_ret, local->op_errno,
+                         local->rebalance.xdata);
+        return 0;
+    }
+
+    if (subvol == NULL)
+        goto err;
+
+    if (local->fop == GF_FOP_REMOVEXATTR) {
+        STACK_WIND_COOKIE(frame, dht_file_removexattr_cbk, subvol, subvol,
+                          subvol->fops->removexattr, &local->loc, local->key,
+                          local->xattr_req);
+    } else {
+        STACK_WIND_COOKIE(frame, dht_file_removexattr_cbk, subvol, subvol,
+                          subvol->fops->fremovexattr, local->fd, local->key,
+                          local->xattr_req);
+    }
+
+    return 0;
+
+err:
+    DHT_STACK_UNWIND(removexattr, frame, -1, op_errno, NULL);
+    return 0;
+}
+
 int
 dht_file_removexattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                          int op_ret, int op_errno, dict_t *xdata)
@@ -5891,8 +6051,8 @@ dht_file_removexattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     local->op_errno = op_errno;
 
-    if ((local->fop == GF_FOP_FREMOVEXATTR) && (op_ret == -1) &&
-        (op_errno == EBADF) && !(local->fd_checked)) {
+    if ((local->fop == GF_FOP_FREMOVEXATTR) &&
+        dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -5952,84 +6112,6 @@ out:
 }
 
 int
-dht_removexattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
-{
-    dht_local_t *local = NULL;
-    int op_errno = EINVAL;
-
-    if (!frame || !frame->local)
-        goto err;
-
-    local = frame->local;
-    op_errno = local->op_errno;
-
-    local->call_cnt = 2; /* This is the second attempt */
-
-    if (we_are_not_migrating(ret)) {
-        /* This dht xlator is not migrating the file. Unwind and
-         * pass on the original mode bits so the higher DHT layer
-         * can handle this.
-         */
-        DHT_STACK_UNWIND(removexattr, frame, local->op_ret, local->op_errno,
-                         local->rebalance.xdata);
-        return 0;
-    }
-
-    if (subvol == NULL)
-        goto err;
-
-    if (local->fop == GF_FOP_REMOVEXATTR) {
-        STACK_WIND_COOKIE(frame, dht_file_removexattr_cbk, subvol, subvol,
-                          subvol->fops->removexattr, &local->loc, local->key,
-                          local->xattr_req);
-    } else {
-        STACK_WIND_COOKIE(frame, dht_file_removexattr_cbk, subvol, subvol,
-                          subvol->fops->fremovexattr, local->fd, local->key,
-                          local->xattr_req);
-    }
-
-    return 0;
-
-err:
-    DHT_STACK_UNWIND(removexattr, frame, -1, op_errno, NULL);
-    return 0;
-}
-
-int
-dht_removexattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int op_ret, int op_errno, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    int this_call_cnt = 0;
-    xlator_t *prev = NULL;
-
-    local = frame->local;
-    prev = cookie;
-
-    LOCK(&frame->lock);
-    {
-        if (op_ret == -1) {
-            local->op_errno = op_errno;
-            gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
-                         prev->name);
-            goto unlock;
-        }
-
-        local->op_ret = 0;
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    this_call_cnt = dht_frame_return(frame);
-    if (is_last_call(this_call_cnt)) {
-        DHT_STACK_UNWIND(removexattr, frame, local->op_ret, local->op_errno,
-                         NULL);
-    }
-
-    return 0;
-}
-
-int
 dht_removexattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
                 const char *key, dict_t *xdata)
 {
@@ -6206,16 +6288,16 @@ dht_fd_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     {
         if (op_ret == -1) {
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
                          prev->name);
-            goto unlock;
+            goto post_unlock;
         }
 
         local->op_ret = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt))
         DHT_STACK_UNWIND(open, frame, local->op_ret, local->op_errno, local->fd,
@@ -6227,7 +6309,7 @@ unlock:
 /*
  * dht_normalize_stats -
  */
-void
+static void
 dht_normalize_stats(struct statvfs *buf, unsigned long bsize,
                     unsigned long frsize)
 {
@@ -6246,7 +6328,7 @@ dht_normalize_stats(struct statvfs *buf, unsigned long bsize,
     }
 }
 
-int
+static int
 dht_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                int op_errno, struct statvfs *statvfs, dict_t *xdata)
 {
@@ -6356,9 +6438,7 @@ dht_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
     int i = -1;
     inode_t *inode = NULL;
     inode_table_t *itable = NULL;
-    uuid_t root_gfid = {
-        0,
-    };
+    static uuid_t root_gfid = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1};
     loc_t newloc = {
         0,
     };
@@ -6384,7 +6464,6 @@ dht_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
         }
 
         loc = &local->loc2;
-        root_gfid[15] = 1;
 
         inode = inode_find(itable, root_gfid);
         if (!inode) {
@@ -6493,15 +6572,16 @@ err:
 }
 
 /* dht_readdirp_cbk creates a new dentry and dentry->inode is not assigned.
-   This functions assigns an inode if all of the following conditions are true:
+   This functions assigns an inode if all of the following conditions are
+   true:
 
-   * DHT has only one child. In this case the entire layout is present on this
-     single child and hence we can set complete layout in inode.
-   * backend has complete layout and there are no anomalies in it and from this
-    information layout can be constructed and set in inode.
+   * DHT has only one child. In this case the entire layout is present on
+   this single child and hence we can set complete layout in inode.
+   * backend has complete layout and there are no anomalies in it and from
+   this information layout can be constructed and set in inode.
 */
 
-void
+static void
 dht_populate_inode_for_dentry(xlator_t *this, xlator_t *subvol,
                               gf_dirent_t *entry, gf_dirent_t *orig_entry)
 {
@@ -6545,9 +6625,10 @@ out:
     return;
 }
 
-/* Posix returns op_errno = ENOENT to indicate that there are no more entries
+/* Posix returns op_errno = ENOENT to indicate that there are no more
+ * entries
  */
-int
+static int
 dht_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                  int op_errno, gf_dirent_t *orig_entries, dict_t *xdata)
 {
@@ -6603,10 +6684,9 @@ dht_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
 
     layout = local->layout;
 
-    /* We have seen crashes in while running "rm -rf" on tier volumes
-       when the layout was NULL on the hot tier. This will skip the
-       entries on the subvol without a layout, hence preventing the crash
-       but rmdir might fail with "directory not empty" errors*/
+    /* This will skip the entries on the subvol without a layout,
+     * hence preventing the crash but rmdir might fail with
+     * "directory not empty" errors*/
 
     if (layout == NULL)
         goto done;
@@ -6624,13 +6704,12 @@ dht_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                      orig_entry->d_name, orig_entry->d_type);
 
         if (IA_ISINVAL(orig_entry->d_stat.ia_type)) {
-            /*stat failed somewhere- ignore this entry*/
-            gf_msg_debug(this->name, EINVAL,
-                         "Invalid stat, ignoring entry "
-                         "%s gfid %s",
+            /*stat failed somewhere- display this entry but the data may
+             * be inaccurate.
+             */
+            gf_msg_debug(this->name, EINVAL, "Invalid stat for %s (gfid %s)",
                          orig_entry->d_name,
                          uuid_utoa(orig_entry->d_stat.ia_gfid));
-            continue;
         }
 
         if (check_is_linkfile(NULL, (&orig_entry->d_stat), orig_entry->dict,
@@ -6712,6 +6791,12 @@ dht_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                 }
             }
         } else {
+            if (orig_entry->dict &&
+                dict_get(orig_entry->dict, conf->link_xattr_name)) {
+                /* Strip out the S and T flags set by rebalance*/
+                DHT_STRIP_PHASE1_FLAGS(&entry->d_stat);
+            }
+
             if (orig_entry->inode) {
                 ret = dht_layout_preset(this, prev, orig_entry->inode);
                 if (ret)
@@ -6833,7 +6918,7 @@ unwind:
     return 0;
 }
 
-int
+static int
 dht_readdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                 int op_errno, gf_dirent_t *orig_entries, dict_t *xdata)
 {
@@ -6956,7 +7041,7 @@ unwind:
     return 0;
 }
 
-int
+static int
 dht_do_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
                off_t yoff, int whichop, dict_t *dict)
 {
@@ -7080,7 +7165,7 @@ dht_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
     return 0;
 }
 
-int
+static int
 dht_fsyncdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                  int op_errno, dict_t *xdata)
 {
@@ -7093,12 +7178,10 @@ dht_fsyncdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     {
         if (op_ret == -1)
             local->op_errno = op_errno;
-
-        if (op_ret == 0)
+        else if (op_ret == 0)
             local->op_ret = 0;
     }
     UNLOCK(&frame->lock);
-
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt))
         DHT_STACK_UNWIND(fsyncdir, frame, local->op_ret, local->op_errno,
@@ -7212,7 +7295,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_mknod_linkfile_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                               int32_t op_ret, int32_t op_errno, inode_t *inode,
                               struct iatt *stbuf, struct iatt *preparent,
@@ -7263,7 +7346,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_mknod_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
                                xlator_t *subvol, loc_t *loc, dev_t rdev,
                                mode_t mode, mode_t umask, dict_t *params)
@@ -7309,7 +7392,7 @@ out:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_mknod_do(call_frame_t *frame)
 {
     dht_local_t *local = NULL;
@@ -7359,7 +7442,7 @@ err:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_mknod_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -7367,7 +7450,7 @@ dht_mknod_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int32_t
+static int32_t
 dht_mknod_finish(call_frame_t *frame, xlator_t *this, int op_ret,
                  int invoke_cbk)
 {
@@ -7419,7 +7502,7 @@ done:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_mknod_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                    int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -7448,11 +7531,15 @@ dht_mknod_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     return 0;
 err:
-    dht_mknod_finish(frame, this, -1, 0);
+    if (local)
+        dht_mknod_finish(frame, this, -1, 0);
+    else
+        DHT_STACK_UNWIND(mknod, frame, -1, EINVAL, NULL, NULL, NULL, NULL,
+                         NULL);
     return 0;
 }
 
-int32_t
+static int32_t
 dht_mknod_lock(call_frame_t *frame, xlator_t *subvol)
 {
     dht_local_t *local = NULL;
@@ -7497,7 +7584,7 @@ err:
     return -1;
 }
 
-int
+static int
 dht_refresh_parent_layout_resume(call_frame_t *frame, xlator_t *this, int ret,
                                  int invoke_cbk)
 {
@@ -7527,7 +7614,7 @@ dht_refresh_parent_layout_resume(call_frame_t *frame, xlator_t *this, int ret,
     return 0;
 }
 
-int
+static int
 dht_refresh_parent_layout_done(call_frame_t *frame)
 {
     dht_local_t *local = NULL;
@@ -7548,7 +7635,7 @@ resume:
     return 0;
 }
 
-int
+static int
 dht_handle_parent_layout_change(xlator_t *this, call_stub_t *stub)
 {
     call_frame_t *refresh_frame = NULL, *frame = NULL;
@@ -7583,7 +7670,7 @@ dht_handle_parent_layout_change(xlator_t *this, call_stub_t *stub)
     return 0;
 }
 
-int32_t
+static int32_t
 dht_call_mkdir_stub(call_frame_t *frame, void *cookie, xlator_t *this,
                     int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -7606,7 +7693,7 @@ dht_call_mkdir_stub(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int32_t
+static int32_t
 dht_guard_parent_layout_and_namespace(xlator_t *subvol, call_stub_t *stub)
 {
     dht_local_t *local = NULL;
@@ -7917,7 +8004,58 @@ err:
     return 0;
 }
 
-int
+static int
+dht_remove_stale_linkto_cbk(int ret, call_frame_t *sync_frame, void *data)
+{
+    DHT_STACK_DESTROY(sync_frame);
+    return 0;
+}
+
+static int
+dht_remove_stale_linkto(void *data)
+{
+    call_frame_t *frame = NULL;
+    dht_local_t *local = NULL;
+    xlator_t *this = NULL;
+    dict_t *xdata_in = NULL;
+    int ret = 0;
+
+    GF_VALIDATE_OR_GOTO("dht", data, out);
+
+    frame = data;
+    local = frame->local;
+    this = frame->this;
+    GF_VALIDATE_OR_GOTO("dht", this, out);
+    GF_VALIDATE_OR_GOTO("dht", local, out);
+    GF_VALIDATE_OR_GOTO("dht", local->link_subvol, out);
+
+    xdata_in = dict_new();
+    if (!xdata_in)
+        goto out;
+
+    ret = dht_fill_dict_to_avoid_unlink_of_migrating_file(xdata_in);
+    if (ret) {
+        gf_msg(this->name, GF_LOG_WARNING, -ret, 0,
+               "Failed to set keys for stale linkto"
+               "deletion on path %s",
+               local->loc.path);
+        goto out;
+    }
+
+    ret = syncop_unlink(local->link_subvol, &local->loc, xdata_in, NULL);
+    if (ret) {
+        gf_msg(this->name, GF_LOG_WARNING, -ret, 0,
+               "Removal of linkto failed"
+               " on path %s at subvol %s",
+               local->loc.path, local->link_subvol->name);
+    }
+out:
+    if (xdata_in)
+        dict_unref(xdata_in);
+    return ret;
+}
+
+static int
 dht_link_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
              int op_errno, inode_t *inode, struct iatt *stbuf,
              struct iatt *preparent, struct iatt *postparent, dict_t *xdata)
@@ -8033,7 +8171,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_link2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -8089,7 +8227,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_link_linkfile_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int op_ret, int op_errno, inode_t *inode,
                       struct iatt *stbuf, struct iatt *preparent,
@@ -8192,6 +8330,11 @@ dht_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     xlator_t *prev = NULL;
     int ret = -1;
     dht_local_t *local = NULL;
+    gf_boolean_t parent_layout_changed = _gf_false;
+    char pgfid[GF_UUID_BUF_SIZE] = {0};
+    xlator_t *subvol = NULL;
+
+    local = frame->local;
 
     local = frame->local;
     if (!local) {
@@ -8200,8 +8343,69 @@ dht_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
         goto out;
     }
 
-    if (op_ret == -1)
+    if (op_ret == -1) {
+        local->op_errno = op_errno;
+        parent_layout_changed = (xdata &&
+                                 dict_get(xdata, GF_PREOP_CHECK_FAILED))
+                                    ? _gf_true
+                                    : _gf_false;
+
+        if (parent_layout_changed) {
+            if (local && local->lock[0].layout.parent_layout.locks) {
+                /* Returning failure as the layout could not be fixed even under
+                 * the lock */
+                goto out;
+            }
+
+            gf_uuid_unparse(local->loc.parent->gfid, pgfid);
+            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_PARENT_LAYOUT_CHANGED,
+                   "create (%s/%s) (path: %s): parent layout "
+                   "changed. Attempting a layout refresh and then a "
+                   "retry",
+                   pgfid, local->loc.name, local->loc.path);
+
+            /*
+              dht_refresh_layout needs directory info in local->loc.Hence,
+              storing the parent_loc in local->loc and storing the create
+              context in local->loc2. We will restore this information in
+              dht_creation_do.
+             */
+
+            loc_wipe(&local->loc2);
+
+            ret = loc_copy(&local->loc2, &local->loc);
+            if (ret) {
+                gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
+                       "loc_copy failed %s", local->loc.path);
+
+                goto out;
+            }
+
+            loc_wipe(&local->loc);
+
+            ret = dht_build_parent_loc(this, &local->loc, &local->loc2,
+                                       &op_errno);
+
+            if (ret) {
+                gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_LOC_FAILED,
+                       "parent loc build failed");
+                goto out;
+            }
+
+            subvol = dht_subvol_get_hashed(this, &local->loc2);
+
+            ret = dht_create_lock(frame, subvol);
+            if (ret < 0) {
+                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_INODE_LK_ERROR,
+                       "locking parent failed");
+                goto out;
+            }
+
+            return 0;
+        }
+
         goto out;
+    }
 
     prev = cookie;
 
@@ -8256,7 +8460,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_create_linkfile_create_cbk(call_frame_t *frame, void *cookie,
                                xlator_t *this, int32_t op_ret, int32_t op_errno,
                                inode_t *inode, struct iatt *stbuf,
@@ -8307,7 +8511,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
                                 xlator_t *subvol, loc_t *loc, int32_t flags,
                                 mode_t mode, mode_t umask, fd_t *fd,
@@ -8322,6 +8526,8 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
         gf_msg_debug(this->name, 0, "creating %s on %s", loc->path,
                      subvol->name);
 
+        dht_set_parent_layout_in_dict(loc, this, local);
+
         STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
                           subvol->fops->create, loc, flags, mode, umask, fd,
                           params);
@@ -8330,10 +8536,6 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
         avail_subvol = dht_free_disk_available_subvol(this, subvol, local);
 
         if (avail_subvol != subvol) {
-            local->params = dict_ref(params);
-            local->flags = flags;
-            local->mode = mode;
-            local->umask = umask;
             local->cached_subvol = avail_subvol;
             local->hashed_subvol = subvol;
 
@@ -8349,6 +8551,8 @@ dht_create_wind_to_avail_subvol(call_frame_t *frame, xlator_t *this,
         gf_msg_debug(this->name, 0, "creating %s on %s", loc->path,
                      subvol->name);
 
+        dht_set_parent_layout_in_dict(loc, this, local);
+
         STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
                           subvol->fops->create, loc, flags, mode, umask, fd,
                           params);
@@ -8417,7 +8621,7 @@ out:
     return ret;
 }
 
-int32_t
+static int32_t
 dht_create_do(call_frame_t *frame)
 {
     dht_local_t *local = NULL;
@@ -8467,7 +8671,7 @@ err:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_create_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -8475,7 +8679,7 @@ dht_create_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int32_t
+static int32_t
 dht_create_finish(call_frame_t *frame, xlator_t *this, int op_ret,
                   int invoke_cbk)
 {
@@ -8527,7 +8731,7 @@ done:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_create_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                     int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -8556,7 +8760,11 @@ dht_create_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     return 0;
 err:
-    dht_create_finish(frame, this, -1, 0);
+    if (local)
+        dht_create_finish(frame, this, -1, 0);
+    else
+        DHT_STACK_UNWIND(create, frame, -1, EINVAL, NULL, NULL, NULL, NULL,
+                         NULL, NULL);
     return 0;
 }
 
@@ -8606,6 +8814,60 @@ err:
 }
 
 int
+dht_set_parent_layout_in_dict(loc_t *loc, xlator_t *this, dht_local_t *local)
+{
+    dht_conf_t *conf = this->private;
+    dht_layout_t *parent_layout = NULL;
+    int *parent_disk_layout = NULL;
+    xlator_t *hashed_subvol = NULL;
+    char pgfid[GF_UUID_BUF_SIZE] = {0};
+    int ret = 0;
+
+    gf_uuid_unparse(loc->parent->gfid, pgfid);
+
+    parent_layout = dht_layout_get(this, loc->parent);
+    hashed_subvol = dht_subvol_get_hashed(this, loc);
+
+    ret = dht_disk_layout_extract_for_subvol(this, parent_layout, hashed_subvol,
+                                             &parent_disk_layout);
+    if (ret == -1) {
+        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+               DHT_MSG_PARENT_LAYOUT_CHANGED,
+               "%s (%s/%s) (path: %s): "
+               "extracting in-memory layout of parent failed. ",
+               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        goto err;
+    }
+
+    ret = dict_set_str_sizen(local->params, GF_PREOP_PARENT_KEY,
+                             conf->xattr_name);
+    if (ret < 0) {
+        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+               DHT_MSG_PARENT_LAYOUT_CHANGED,
+               "%s (%s/%s) (path: %s): "
+               "setting %s key in params dictionary failed. ",
+               gf_fop_list[local->fop], pgfid, loc->name, loc->path,
+               GF_PREOP_PARENT_KEY);
+        goto err;
+    }
+
+    ret = dict_set_bin(local->params, conf->xattr_name, parent_disk_layout,
+                       4 * 4);
+    if (ret < 0) {
+        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
+               DHT_MSG_PARENT_LAYOUT_CHANGED,
+               "%s (%s/%s) (path: %s): "
+               "setting parent-layout in params dictionary failed. ",
+               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        goto err;
+    }
+
+err:
+    dht_layout_unref(this, parent_layout);
+    return ret;
+}
+
+int
 dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
            mode_t mode, mode_t umask, fd_t *fd, dict_t *params)
 {
@@ -8631,6 +8893,11 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
         goto err;
     }
 
+    local->params = dict_ref(params);
+    local->flags = flags;
+    local->mode = mode;
+    local->umask = umask;
+
     if (dht_filter_loc_subvol_key(this, loc, &local->loc, &subvol)) {
         gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_SUBVOL_INFO,
                "creating %s on %s (got create on %s)", local->loc.path,
@@ -8646,10 +8913,6 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
 
         if (hashed_subvol && (hashed_subvol != subvol)) {
             /* Create the linkto file and then the data file */
-            local->params = dict_ref(params);
-            local->flags = flags;
-            local->mode = mode;
-            local->umask = umask;
             local->cached_subvol = subvol;
             local->hashed_subvol = hashed_subvol;
 
@@ -8662,6 +8925,9 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
          * file as we expect a lookup everywhere if there are problems
          * with the parent layout
          */
+
+        dht_set_parent_layout_in_dict(loc, this, local);
+
         STACK_WIND_COOKIE(frame, dht_create_cbk, subvol, subvol,
                           subvol->fops->create, &local->loc, flags, mode, umask,
                           fd, params);
@@ -8713,11 +8979,6 @@ dht_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
                     goto err;
                 }
 
-                local->params = dict_ref(params);
-                local->flags = flags;
-                local->mode = mode;
-                local->umask = umask;
-
                 loc_wipe(&local->loc);
 
                 ret = dht_build_parent_loc(this, &local->loc, loc, &op_errno);
@@ -8755,7 +9016,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_mkdir_selfheal_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                        int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -8789,7 +9050,7 @@ dht_mkdir_selfheal_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
               int op_errno, inode_t *inode, struct iatt *stbuf,
               struct iatt *preparent, struct iatt *postparent, dict_t *xdata)
@@ -8857,13 +9118,13 @@ unlock:
     return 0;
 }
 
-int
+static int
 dht_mkdir_hashed_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int op_ret, int op_errno, inode_t *inode,
                      struct iatt *stbuf, struct iatt *preparent,
                      struct iatt *postparent, dict_t *xdata);
 
-int
+static int
 dht_mkdir_helper(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
                  mode_t umask, dict_t *params)
 {
@@ -8986,7 +9247,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_mkdir_hashed_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int op_ret, int op_errno, inode_t *inode,
                      struct iatt *stbuf, struct iatt *preparent,
@@ -9003,8 +9264,6 @@ dht_mkdir_hashed_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     gf_boolean_t parent_layout_changed = _gf_false;
     call_stub_t *stub = NULL;
 
-    VALIDATE_OR_GOTO(this->private, err);
-
     local = frame->local;
     prev = cookie;
     layout = local->layout;
@@ -9119,7 +9378,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_mkdir_guard_parent_layout_cbk(call_frame_t *frame, xlator_t *this,
                                   loc_t *loc, mode_t mode, mode_t umask,
                                   dict_t *params)
@@ -9270,7 +9529,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_rmdir_selfheal_cbk(call_frame_t *heal_frame, void *cookie, xlator_t *this,
                        int op_ret, int op_errno, dict_t *xdata)
 {
@@ -9292,7 +9551,7 @@ dht_rmdir_selfheal_cbk(call_frame_t *heal_frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_rmdir_hashed_subvol_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                             int op_ret, int op_errno, struct iatt *preparent,
                             struct iatt *postparent, dict_t *xdata)
@@ -9395,7 +9654,63 @@ err:
     return 0;
 }
 
-int
+static int
+dht_rmdir_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                     int32_t op_ret, int32_t op_errno, dict_t *xdata)
+{
+    DHT_STACK_DESTROY(frame);
+    return 0;
+}
+
+static int
+dht_rmdir_unlock(call_frame_t *frame, xlator_t *this)
+{
+    dht_local_t *local = NULL, *lock_local = NULL;
+    call_frame_t *lock_frame = NULL;
+    int lock_count = 0;
+
+    local = frame->local;
+
+    /* Unlock entrylk */
+    dht_unlock_entrylk_wrapper(frame, &local->lock[0].ns.directory_ns);
+
+    /* Unlock inodelk */
+    lock_count = dht_lock_count(local->lock[0].ns.parent_layout.locks,
+                                local->lock[0].ns.parent_layout.lk_count);
+
+    if (lock_count == 0)
+        goto done;
+
+    lock_frame = copy_frame(frame);
+    if (lock_frame == NULL)
+        goto done;
+
+    lock_local = dht_local_init(lock_frame, &local->loc, NULL,
+                                lock_frame->root->op);
+    if (lock_local == NULL)
+        goto done;
+
+    lock_local->lock[0].ns.parent_layout.locks = local->lock[0]
+                                                     .ns.parent_layout.locks;
+    lock_local->lock[0]
+        .ns.parent_layout.lk_count = local->lock[0].ns.parent_layout.lk_count;
+
+    local->lock[0].ns.parent_layout.locks = NULL;
+    local->lock[0].ns.parent_layout.lk_count = 0;
+    dht_unlock_inodelk(lock_frame, lock_local->lock[0].ns.parent_layout.locks,
+                       lock_local->lock[0].ns.parent_layout.lk_count,
+                       dht_rmdir_unlock_cbk);
+    lock_frame = NULL;
+
+done:
+    if (lock_frame != NULL) {
+        DHT_STACK_DESTROY(lock_frame);
+    }
+
+    return 0;
+}
+
+static int
 dht_rmdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
               int op_errno, struct iatt *preparent, struct iatt *postparent,
               dict_t *xdata)
@@ -9530,63 +9845,7 @@ err:
     return 0;
 }
 
-int
-dht_rmdir_unlock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                     int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    DHT_STACK_DESTROY(frame);
-    return 0;
-}
-
-int
-dht_rmdir_unlock(call_frame_t *frame, xlator_t *this)
-{
-    dht_local_t *local = NULL, *lock_local = NULL;
-    call_frame_t *lock_frame = NULL;
-    int lock_count = 0;
-
-    local = frame->local;
-
-    /* Unlock entrylk */
-    dht_unlock_entrylk_wrapper(frame, &local->lock[0].ns.directory_ns);
-
-    /* Unlock inodelk */
-    lock_count = dht_lock_count(local->lock[0].ns.parent_layout.locks,
-                                local->lock[0].ns.parent_layout.lk_count);
-
-    if (lock_count == 0)
-        goto done;
-
-    lock_frame = copy_frame(frame);
-    if (lock_frame == NULL)
-        goto done;
-
-    lock_local = dht_local_init(lock_frame, &local->loc, NULL,
-                                lock_frame->root->op);
-    if (lock_local == NULL)
-        goto done;
-
-    lock_local->lock[0].ns.parent_layout.locks = local->lock[0]
-                                                     .ns.parent_layout.locks;
-    lock_local->lock[0]
-        .ns.parent_layout.lk_count = local->lock[0].ns.parent_layout.lk_count;
-
-    local->lock[0].ns.parent_layout.locks = NULL;
-    local->lock[0].ns.parent_layout.lk_count = 0;
-    dht_unlock_inodelk(lock_frame, lock_local->lock[0].ns.parent_layout.locks,
-                       lock_local->lock[0].ns.parent_layout.lk_count,
-                       dht_rmdir_unlock_cbk);
-    lock_frame = NULL;
-
-done:
-    if (lock_frame != NULL) {
-        DHT_STACK_DESTROY(lock_frame);
-    }
-
-    return 0;
-}
-
-int
+static int
 dht_rmdir_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                    int32_t op_ret, int32_t op_errno, dict_t *xdata)
 {
@@ -9595,8 +9854,6 @@ dht_rmdir_lock_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     int i = 0;
     xlator_t *hashed_subvol;
 
-    VALIDATE_OR_GOTO(this->private, err);
-
     conf = this->private;
     local = frame->local;
 
@@ -9629,7 +9886,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_rmdir_do(call_frame_t *frame, xlator_t *this)
 {
     dht_local_t *local = NULL;
@@ -9638,13 +9895,13 @@ dht_rmdir_do(call_frame_t *frame, xlator_t *this)
     xlator_t *hashed_subvol = NULL;
     char gfid[GF_UUID_BUF_SIZE] = {0};
 
-    VALIDATE_OR_GOTO(this->private, err);
-
-    conf = this->private;
+    VALIDATE_OR_GOTO(frame->local, err);
     local = frame->local;
+    VALIDATE_OR_GOTO(this->private, out);
+    conf = this->private;
 
     if (local->op_ret == -1)
-        goto err;
+        goto out;
 
     local->call_cnt = conf->subvolume_cnt;
 
@@ -9676,21 +9933,80 @@ dht_rmdir_do(call_frame_t *frame, xlator_t *this)
     if (ret < 0) {
         local->op_ret = -1;
         local->op_errno = errno ? errno : EINVAL;
-        goto err;
+        goto out;
     }
 
     return 0;
 
-err:
+out:
     dht_set_fixed_dir_stat(&local->preparent);
     dht_set_fixed_dir_stat(&local->postparent);
 
     DHT_STACK_UNWIND(rmdir, frame, local->op_ret, local->op_errno,
                      &local->preparent, &local->postparent, NULL);
     return 0;
+err:
+    DHT_STACK_UNWIND(rmdir, frame, -1, EINVAL, NULL, NULL, NULL);
+    return 0;
 }
 
-int
+static void
+dht_rmdir_readdirp_done(call_frame_t *readdirp_frame, xlator_t *this)
+{
+    call_frame_t *main_frame = NULL;
+    dht_local_t *main_local = NULL;
+    dht_local_t *local = NULL;
+    int this_call_cnt = 0;
+
+    local = readdirp_frame->local;
+    main_frame = local->main_frame;
+    main_local = main_frame->local;
+
+    /* At least one readdirp failed.
+     * This is a bit hit or miss - if readdirp failed on more than
+     * one subvol, we don't know which error is returned.
+     */
+    if (local->op_ret == -1) {
+        main_local->op_ret = local->op_ret;
+        main_local->op_errno = local->op_errno;
+    }
+
+    this_call_cnt = dht_frame_return(main_frame);
+
+    if (is_last_call(this_call_cnt))
+        dht_rmdir_do(main_frame, this);
+
+    DHT_STACK_DESTROY(readdirp_frame);
+}
+
+/* Keep sending readdirp on the subvol until it returns no more entries
+ * It is possible that not all entries will fit in a single readdirp in
+ * which case the rmdir will keep failing with ENOTEMPTY
+ */
+
+static int
+dht_rmdir_readdirp_do(call_frame_t *readdirp_frame, xlator_t *this)
+{
+    dht_local_t *local = NULL;
+
+    local = readdirp_frame->local;
+
+    if (local->op_ret == -1) {
+        /* there is no point doing another readdirp on this
+         * subvol . */
+        dht_rmdir_readdirp_done(readdirp_frame, this);
+        return 0;
+    }
+
+    STACK_WIND_COOKIE(readdirp_frame, dht_rmdir_readdirp_cbk,
+                      local->hashed_subvol, local->hashed_subvol,
+                      local->hashed_subvol->fops->readdirp, local->fd, 4096, 0,
+                      local->xattr);
+
+    return 0;
+}
+
+static int
 dht_rmdir_linkfile_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                               int op_ret, int op_errno, struct iatt *preparent,
                               struct iatt *postparent, dict_t *xdata)
@@ -9734,7 +10050,7 @@ dht_rmdir_linkfile_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-int
+static int
 dht_rmdir_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                      int op_ret, int op_errno, inode_t *inode,
                      struct iatt *stbuf, dict_t *xattr, struct iatt *parent)
@@ -9759,8 +10075,7 @@ dht_rmdir_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     if (op_ret != 0) {
         gf_msg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_FILE_LOOKUP_FAILED,
-               "lookup failed for %s on %s  (type=0%o)", local->loc.path,
-               src->name, stbuf->ia_type);
+               "lookup failed for %s on %s", local->loc.path, src->name);
         goto err;
     }
 
@@ -9790,7 +10105,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_rmdir_cached_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                             int op_ret, int op_errno, inode_t *inode,
                             struct iatt *stbuf, dict_t *xattr,
@@ -9870,7 +10185,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_rmdir_is_subvol_empty(call_frame_t *frame, xlator_t *this,
                           gf_dirent_t *entries, xlator_t *src)
 {
@@ -9966,7 +10281,10 @@ dht_rmdir_is_subvol_empty(call_frame_t *frame, xlator_t *this,
                      lookup_local->loc.path, src->name, gfid);
 
         subvol = dht_linkfile_subvol(this, NULL, &trav->d_stat, trav->dict);
-        if (!subvol) {
+        if (!subvol || (subvol == src)) {
+            /* we need to delete the linkto file if it does not have a
+             * valid subvol or it points to itself.
+             */
             gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_INVALID_LINKFILE,
                    "Linkfile does not have link subvolume. "
                    "path = %s, gfid = %s",
@@ -10028,36 +10346,7 @@ err:
  * No more entries on this subvol. Proceed to the actual rmdir operation.
  */
 
-void
-dht_rmdir_readdirp_done(call_frame_t *readdirp_frame, xlator_t *this)
-{
-    call_frame_t *main_frame = NULL;
-    dht_local_t *main_local = NULL;
-    dht_local_t *local = NULL;
-    int this_call_cnt = 0;
-
-    local = readdirp_frame->local;
-    main_frame = local->main_frame;
-    main_local = main_frame->local;
-
-    /* At least one readdirp failed.
-     * This is a bit hit or miss - if readdirp failed on more than
-     * one subvol, we don't know which error is returned.
-     */
-    if (local->op_ret == -1) {
-        main_local->op_ret = local->op_ret;
-        main_local->op_errno = local->op_errno;
-    }
-
-    this_call_cnt = dht_frame_return(main_frame);
-
-    if (is_last_call(this_call_cnt))
-        dht_rmdir_do(main_frame, this);
-
-    DHT_STACK_DESTROY(readdirp_frame);
-}
-
-int
+static int
 dht_rmdir_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                        int op_ret, int op_errno, gf_dirent_t *entries,
                        dict_t *xdata)
@@ -10107,34 +10396,7 @@ dht_rmdir_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
-/* Keep sending readdirp on the subvol until it returns no more entries
- * It is possible that not all entries will fit in a single readdirp in which
- * case the rmdir will keep failing with ENOTEMPTY
- */
-
-int
-dht_rmdir_readdirp_do(call_frame_t *readdirp_frame, xlator_t *this)
-{
-    dht_local_t *local = NULL;
-
-    local = readdirp_frame->local;
-
-    if (local->op_ret == -1) {
-        /* there is no point doing another readdirp on this
-         * subvol . */
-        dht_rmdir_readdirp_done(readdirp_frame, this);
-        return 0;
-    }
-
-    STACK_WIND_COOKIE(readdirp_frame, dht_rmdir_readdirp_cbk,
-                      local->hashed_subvol, local->hashed_subvol,
-                      local->hashed_subvol->fops->readdirp, local->fd, 4096, 0,
-                      local->xattr);
-
-    return 0;
-}
-
-int
+static int
 dht_rmdir_opendir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int op_ret, int op_errno, fd_t *fd, dict_t *xdata)
 {
@@ -10254,6 +10516,8 @@ dht_rmdir(call_frame_t *frame, xlator_t *this, loc_t *loc, int flags,
     dht_conf_t *conf = NULL;
     int op_errno = -1;
     int i = -1;
+    int ret = -1;
+    dict_t *xattr_req = NULL;
 
     VALIDATE_OR_GOTO(frame, err);
     VALIDATE_OR_GOTO(this, err);
@@ -10285,14 +10549,37 @@ dht_rmdir(call_frame_t *frame, xlator_t *this, loc_t *loc, int flags,
     if (flags) {
         return dht_rmdir_do(frame, this);
     }
+    if (xdata) {
+        xattr_req = dict_ref(xdata);
+    } else {
+        xattr_req = dict_new();
+    }
+    if (xattr_req) {
+        ret = dict_set_uint32(xattr_req, conf->link_xattr_name, 256);
+        /* If parallel-readdir is enabled, this is required
+         * to handle stale linkto files in the directory
+         * being deleted. If this fails, log an error but
+         * do not prevent the operation.
+         */
+        if (ret) {
+            gf_msg(this->name, GF_LOG_ERROR, 0, 0, "%s: failed to set key %s",
+                   loc->path, conf->link_xattr_name);
+        }
+    } else {
+        gf_msg(this->name, GF_LOG_ERROR, 0, 0, "%s: failed to set key %s",
+               loc->path, conf->link_xattr_name);
+    }
 
     for (i = 0; i < conf->subvolume_cnt; i++) {
         STACK_WIND_COOKIE(frame, dht_rmdir_opendir_cbk, conf->subvolumes[i],
                           conf->subvolumes[i],
                           conf->subvolumes[i]->fops->opendir, loc, local->fd,
-                          NULL);
+                          xattr_req);
     }
 
+    if (xattr_req) {
+        dict_unref(xattr_req);
+    }
     return 0;
 
 err:
@@ -10302,7 +10589,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_entrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                 int32_t op_ret, int32_t op_errno, dict_t *xdata)
 
@@ -10362,7 +10649,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_fentrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                  int32_t op_ret, int32_t op_errno, dict_t *xdata)
 
@@ -10409,7 +10696,7 @@ err:
     return 0;
 }
 
-int32_t
+static int32_t
 dht_ipc_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
             int32_t op_errno, dict_t *xdata)
 {
@@ -10526,23 +10813,17 @@ dht_notify(xlator_t *this, int event, void *data, ...)
 
     int had_heard_from_all = 0;
     int have_heard_from_all = 0;
-    struct timeval time = {
-        0,
-    };
     gf_defrag_info_t *defrag = NULL;
     dict_t *dict = NULL;
     gf_defrag_type cmd = 0;
     dict_t *output = NULL;
     va_list ap;
-    dht_methods_t *methods = NULL;
     struct gf_upcall *up_data = NULL;
     struct gf_upcall_cache_invalidation *up_ci = NULL;
 
     conf = this->private;
     GF_VALIDATE_OR_GOTO(this->name, conf, out);
 
-    methods = &(conf->methods);
-
     /* had all subvolumes reported status once till now? */
     had_heard_from_all = 1;
     for (i = 0; i < conf->subvolume_cnt; i++) {
@@ -10572,12 +10853,11 @@ dht_notify(xlator_t *this, int event, void *data, ...)
                 break;
             }
 
-            gettimeofday(&time, NULL);
             LOCK(&conf->subvolume_lock);
             {
                 conf->subvolume_status[cnt] = 1;
                 conf->last_event[cnt] = event;
-                conf->subvol_up_time[cnt] = time.tv_sec;
+                conf->subvol_up_time[cnt] = gf_time();
             }
             UNLOCK(&conf->subvolume_lock);
 
@@ -10685,21 +10965,13 @@ dht_notify(xlator_t *this, int event, void *data, ...)
                 if (defrag->is_exiting)
                     goto unlock;
                 if ((cmd == GF_DEFRAG_CMD_STATUS) ||
-                    (cmd == GF_DEFRAG_CMD_STATUS_TIER) ||
                     (cmd == GF_DEFRAG_CMD_DETACH_STATUS))
                     gf_defrag_status_get(conf, output);
-                else if (cmd == GF_DEFRAG_CMD_START_DETACH_TIER)
-                    gf_defrag_start_detach_tier(defrag);
                 else if (cmd == GF_DEFRAG_CMD_DETACH_START)
                     defrag->cmd = GF_DEFRAG_CMD_DETACH_START;
                 else if (cmd == GF_DEFRAG_CMD_STOP ||
-                         cmd == GF_DEFRAG_CMD_STOP_DETACH_TIER ||
                          cmd == GF_DEFRAG_CMD_DETACH_STOP)
                     gf_defrag_stop(conf, GF_DEFRAG_STATUS_STOPPED, output);
-                else if (cmd == GF_DEFRAG_CMD_PAUSE_TIER)
-                    ret = gf_defrag_pause_tier(this, defrag);
-                else if (cmd == GF_DEFRAG_CMD_RESUME_TIER)
-                    ret = gf_defrag_resume_tier(this, defrag);
             }
         unlock:
             UNLOCK(&defrag->lock);
@@ -10747,6 +11019,7 @@ dht_notify(xlator_t *this, int event, void *data, ...)
     }
 
     if (!had_heard_from_all && have_heard_from_all) {
+        static int run_defrag = 0;
         /* This is the first event which completes aggregation
            of events from all subvolumes. If at least one subvol
            had come up, propagate CHILD_UP, but only this time
@@ -10773,15 +11046,13 @@ dht_notify(xlator_t *this, int event, void *data, ...)
          * thread has already started.
          */
         if (conf->defrag && !run_defrag) {
-            if (methods->migration_needed(this)) {
-                run_defrag = 1;
-                ret = gf_thread_create(&conf->defrag->th, NULL, gf_defrag_start,
-                                       this, "dhtdg");
-                if (ret) {
-                    GF_FREE(conf->defrag);
-                    conf->defrag = NULL;
-                    kill(getpid(), SIGTERM);
-                }
+            run_defrag = 1;
+            ret = gf_thread_create(&conf->defrag->th, NULL, gf_defrag_start,
+                                   this, "dhtdg");
+            if (ret) {
+                GF_FREE(conf->defrag);
+                conf->defrag = NULL;
+                kill(getpid(), SIGTERM);
             }
         }
     }
@@ -10850,8 +11121,8 @@ dht_log_new_layout_for_dir_selfheal(xlator_t *this, loc_t *loc,
     len += ret;
 
     /* Calculation  of total length of the string required to calloc
-     * output_string. Log includes subvolume-name, start-range, end-range and
-     * err value.
+     * output_string. Log includes subvolume-name, start-range, end-range
+     * and err value.
      *
      * This log will help to debug cases where:
      * a) Different processes set different layout of a directory.
@@ -10862,8 +11133,7 @@ dht_log_new_layout_for_dir_selfheal(xlator_t *this, loc_t *loc,
     for (i = 0; i < layout->cnt; i++) {
         ret = snprintf(string, sizeof(string),
                        "[Subvol_name: %s, Err: %d , Start: "
-                       "%" PRIu32 " , Stop: %" PRIu32 " , Hash: %" PRIu32
-                       " ], ",
+                       "0x%x, Stop: 0x%x, Hash: 0x%x], ",
                        layout->list[i].xlator->name, layout->list[i].err,
                        layout->list[i].start, layout->list[i].stop,
                        layout->list[i].commit_hash);
@@ -10892,8 +11162,7 @@ dht_log_new_layout_for_dir_selfheal(xlator_t *this, loc_t *loc,
     for (i = 0; i < layout->cnt; i++) {
         ret = snprintf(output_string + off, len - off,
                        "[Subvol_name: %s, Err: %d , Start: "
-                       "%" PRIu32 " , Stop: %" PRIu32 " , Hash: %" PRIu32
-                       " ], ",
+                       "0x%x, Stop: 0x%x, Hash: 0x%x], ",
                        layout->list[i].xlator->name, layout->list[i].err,
                        layout->list[i].start, layout->list[i].stop,
                        layout->list[i].commit_hash);
@@ -10928,28 +11197,6 @@ out:
     return ret;
 }
 
-int32_t
-dht_migration_needed(xlator_t *this)
-{
-    gf_defrag_info_t *defrag = NULL;
-    dht_conf_t *conf = NULL;
-    int ret = 0;
-
-    conf = this->private;
-
-    GF_VALIDATE_OR_GOTO("dht", conf, out);
-    GF_VALIDATE_OR_GOTO("dht", conf->defrag, out);
-
-    defrag = conf->defrag;
-
-    if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
-        (defrag->cmd != GF_DEFRAG_CMD_START_DETACH_TIER))
-        ret = 1;
-
-out:
-    return ret;
-}
-
 /*
 This function should not be called more then once during a FOP
 handling path. It is valid only for for ops on files
@@ -10984,72 +11231,13 @@ dht_set_local_rebalance(xlator_t *this, dht_local_t *local, struct iatt *stbuf,
     return 0;
 }
 
-gf_boolean_t
-dht_is_tier_xlator(xlator_t *this)
-{
-    if (strcmp(this->type, "cluster/tier") == 0)
-        return _gf_true;
-    return _gf_false;
-}
-
 int32_t
 dht_release(xlator_t *this, fd_t *fd)
 {
     return dht_fd_ctx_destroy(this, fd);
 }
 
-int
-dht_remove_stale_linkto(void *data)
-{
-    call_frame_t *frame = NULL;
-    dht_local_t *local = NULL;
-    xlator_t *this = NULL;
-    dict_t *xdata_in = NULL;
-    int ret = 0;
-
-    GF_VALIDATE_OR_GOTO("dht", data, out);
-
-    frame = data;
-    local = frame->local;
-    this = frame->this;
-    GF_VALIDATE_OR_GOTO("dht", this, out);
-    GF_VALIDATE_OR_GOTO("dht", local, out);
-    GF_VALIDATE_OR_GOTO("dht", local->link_subvol, out);
-
-    xdata_in = dict_new();
-    if (!xdata_in)
-        goto out;
-
-    ret = dht_fill_dict_to_avoid_unlink_of_migrating_file(xdata_in);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, 0,
-               "Failed to set keys for stale linkto"
-               "deletion on path %s",
-               local->loc.path);
-        goto out;
-    }
-
-    ret = syncop_unlink(local->link_subvol, &local->loc, xdata_in, NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, 0,
-               "Removal of linkto failed"
-               " on path %s at subvol %s",
-               local->loc.path, local->link_subvol->name);
-    }
-out:
-    if (xdata_in)
-        dict_unref(xdata_in);
-    return ret;
-}
-
-int
-dht_remove_stale_linkto_cbk(int ret, call_frame_t *sync_frame, void *data)
-{
-    DHT_STACK_DESTROY(sync_frame);
-    return 0;
-}
-
-int
+static int
 dht_pt_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                  int op_errno, inode_t *inode, struct iatt *stbuf,
                  struct iatt *preparent, struct iatt *postparent, dict_t *xdata)
@@ -11135,6 +11323,8 @@ dht_pt_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     conf = this->private;
     dict_del(xattr, conf->xattr_name);
+    dict_del(xattr, conf->mds_xattr_key);
+    dict_del(xattr, conf->commithash_xattr_name);
 
     if (frame->root->pid >= 0) {
         GF_REMOVE_INTERNAL_XATTR("trusted.glusterfs.quota*", xattr);
@@ -11180,3 +11370,22 @@ dht_pt_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, const char *key,
                FIRST_CHILD(this)->fops->fgetxattr, fd, key, xdata);
     return 0;
 }
+
+/* The job of this function is to check if all the xlators have updated
+ * error in the layout. */
+int
+dht_dir_layout_error_check(xlator_t *this, inode_t *inode)
+{
+    dht_layout_t *layout = NULL;
+    int i = 0;
+
+    layout = dht_layout_get(this, inode);
+    for (i = 0; i < layout->cnt; i++) {
+        if (layout->list[i].err == 0) {
+            return 0;
+        }
+    }
+
+    /* Returning the first xlator error as all xlators have errors */
+    return layout->list[0].err;
+}
diff --git a/xlators/cluster/dht/src/dht-common.h b/xlators/cluster/dht/src/dht-common.h
index 580f57e6e25..fe0dc3db34a 100644
--- a/xlators/cluster/dht/src/dht-common.h
+++ b/xlators/cluster/dht/src/dht-common.h
@@ -9,24 +9,21 @@
 */
 
 #include <regex.h>
-#include <signal.h>
-#include <fnmatch.h>
 
 #include "dht-mem-types.h"
 #include "dht-messages.h"
-#include "call-stub.h"
+#include <glusterfs/call-stub.h>
 #include "libxlator.h"
-#include "syncop.h"
-#include "refcount.h"
-#include "timer.h"
+#include <glusterfs/syncop.h>
+#include <glusterfs/refcount.h>
+#include <glusterfs/timer.h>
 #include "protocol-common.h"
-#include "glusterfs-acl.h"
+#include <glusterfs/glusterfs-acl.h>
 
 #ifndef _DHT_H
 #define _DHT_H
 
 #define GF_XATTR_FIX_LAYOUT_KEY "distribute.fix.layout"
-#define GF_XATTR_TIER_LAYOUT_FIXED_KEY "trusted.tier.fix.layout.complete"
 #define GF_XATTR_FILE_MIGRATE_KEY "trusted.distribute.migrate-data"
 #define DHT_MDS_STR "mds"
 #define GF_DHT_LOOKUP_UNHASHED_OFF 0
@@ -38,22 +35,21 @@
 #define DHT_LAYOUT_HEAL_DOMAIN "dht.layout.heal"
 /* Namespace synchronization */
 #define DHT_ENTRY_SYNC_DOMAIN "dht.entry.sync"
-#define TIERING_MIGRATION_KEY "tiering.migration"
 #define DHT_LAYOUT_HASH_INVALID 1
 #define MAX_REBAL_THREADS sysconf(_SC_NPROCESSORS_ONLN)
 
 #define DHT_DIR_STAT_BLOCKS 8
 #define DHT_DIR_STAT_SIZE 4096
 
+/* Virtual xattr for subvols status */
+
+#define DHT_SUBVOL_STATUS_KEY "dht.subvol.status"
+
 /* Virtual xattrs for debugging */
 
 #define DHT_DBG_HASHED_SUBVOL_PATTERN "dht.file.hashed-subvol.*"
 #define DHT_DBG_HASHED_SUBVOL_KEY "dht.file.hashed-subvol."
 
-/* Array to hold custom xattr keys
- */
-extern char *xattrs_to_heal[];
-
 /* Rebalance nodeuuid flags */
 #define REBAL_NODEUUID_MINE 0x01
 
@@ -152,8 +148,8 @@ struct dht_rebalance_ {
     dict_t *xdata;
     dict_t *xattr;
     dict_t *dict;
-    int32_t set;
     struct gf_flock flock;
+    int32_t set;
     int lock_cmd;
 };
 
@@ -176,24 +172,24 @@ typedef enum {
 } dht_reaction_type_t;
 
 struct dht_skip_linkto_unlink {
-    gf_boolean_t handle_valid_link;
-    int opend_fd_count;
     xlator_t *hash_links_to;
     uuid_t cached_gfid;
     uuid_t hashed_gfid;
+    int opend_fd_count;
+    gf_boolean_t handle_valid_link;
 };
 
 typedef struct {
     xlator_t *xl;
     loc_t loc;      /* contains/points to inode to lock on. */
-    short type;     /* read/write lock.                     */
     char *domain;   /* Only locks within a single domain
                      * contend with each other
                      */
     char *basename; /* Required for entrylk */
-    gf_lkowner_t lk_owner;
     gf_boolean_t locked;
     dht_reaction_type_t do_on_failure;
+    short type; /* read/write lock.                     */
+    gf_lkowner_t lk_owner;
 } dht_lock_t;
 
 /* The lock structure represents inodelk. */
@@ -244,23 +240,10 @@ typedef gf_boolean_t (*dht_need_heal_t)(call_frame_t *frame,
                                         dht_layout_t **inmem,
                                         dht_layout_t **ondisk);
 
-typedef struct {
-    uint64_t blocks_used;
-    uint64_t pblocks_used;
-    uint64_t files_used;
-    uint64_t pfiles_used;
-    uint64_t unhashed_blocks_used;
-    uint64_t unhashed_pblocks_used;
-    uint64_t unhashed_files_used;
-    uint64_t unhashed_pfiles_used;
-    uint64_t unhashed_fsid;
-    uint64_t hashed_fsid;
-} tier_statvfs_t;
-
 struct dht_local {
-    int call_cnt;
     loc_t loc;
     loc_t loc2;
+    int call_cnt;
     int op_ret;
     int op_errno;
     int layout_mismatch;
@@ -274,7 +257,6 @@ struct dht_local {
     struct iatt preparent;
     struct iatt postparent;
     struct statvfs statvfs;
-    tier_statvfs_t tier_statvfs;
     fd_t *fd;
     inode_t *inode;
     dict_t *params;
@@ -290,9 +272,6 @@ struct dht_local {
     xlator_t *cached_subvol;
     xlator_t *hashed_subvol;
     xlator_t *mds_subvol; /* This is use for dir only */
-    char need_selfheal;
-    char need_xattr_heal;
-    char need_attrheal;
     int file_count;
     int dir_count;
     call_frame_t *main_frame;
@@ -310,12 +289,12 @@ struct dht_local {
         uint32_t overlaps_cnt;
         uint32_t down;
         uint32_t misc;
-        uint32_t missing_cnt;
         dht_selfheal_dir_cbk_t dir_cbk;
         dht_selfheal_layout_t healer;
         dht_need_heal_t should_heal;
-        gf_boolean_t force_mkdir;
         dht_layout_t *layout, *refreshed_layout;
+        uint32_t missing_cnt;
+        gf_boolean_t force_mkdir;
     } selfheal;
 
     dht_refresh_layout_unlock refresh_layout_unlock;
@@ -323,7 +302,13 @@ struct dht_local {
 
     uint32_t uid;
     uint32_t gid;
-    pid_t    pid;
+    pid_t pid;
+
+    glusterfs_fop_t fop;
+
+    /* need for file-info */
+    char *xattr_val;
+    char *key;
 
     /* needed by nufa */
     int32_t flags;
@@ -331,44 +316,25 @@ struct dht_local {
     dev_t rdev;
     mode_t umask;
 
-    /* need for file-info */
-    char *xattr_val;
-    char *key;
-
     /* which xattr request? */
     char xsel[256];
     int32_t alloc_len;
 
     /* gfid related */
     uuid_t gfid;
+    uuid_t gfid_req;
 
-    /* flag used to make sure we need to return estale in
-       {lookup,revalidate}_cbk */
-    char return_estale;
-    char need_lookup_everywhere;
-
-    glusterfs_fop_t fop;
-
-    gf_boolean_t linked;
     xlator_t *link_subvol;
 
     struct dht_rebalance_ rebalance;
     xlator_t *first_up_subvol;
 
-    gf_boolean_t quota_deem_statfs;
-
-    gf_boolean_t added_link;
-    gf_boolean_t is_linkfile;
-
     struct dht_skip_linkto_unlink skip_unlink;
 
     dht_dir_transaction_t lock[2], *current;
 
     /* inodelks during filerename for backward compatibility */
     dht_lock_t **rename_inodelk_backward_compatible;
-    int rename_inodelk_bc_count;
-
-    short lock_type;
 
     call_stub_t *stub;
     int32_t parent_disk_layout[4];
@@ -376,15 +342,30 @@ struct dht_local {
     /* rename rollback */
     int *ret_cache;
 
-    /* fd open check */
-    gf_boolean_t fd_checked;
+    loc_t loc2_copy;
+
+    int rename_inodelk_bc_count;
     /* This is use only for directory operation */
     int32_t valid;
-    gf_boolean_t heal_layout;
     int32_t mds_heal_fresh_lookup;
-    loc_t loc2_copy;
+    short lock_type;
+    char need_selfheal;
+    char need_xattr_heal;
+    char need_attrheal;
+    /* flag used to make sure we need to return estale in
+       {lookup,revalidate}_cbk */
+    char return_estale;
+    char need_lookup_everywhere;
+    /* fd open check */
+    gf_boolean_t fd_checked;
+    gf_boolean_t linked;
+    gf_boolean_t added_link;
+    gf_boolean_t is_linkfile;
+    gf_boolean_t quota_deem_statfs;
+    gf_boolean_t heal_layout;
     gf_boolean_t locked;
     gf_boolean_t dont_create_linkto;
+    gf_boolean_t gfid_missing;
 };
 typedef struct dht_local dht_local_t;
 
@@ -408,14 +389,7 @@ enum gf_defrag_type {
     GF_DEFRAG_CMD_STATUS = 1 + 2,
     GF_DEFRAG_CMD_START_LAYOUT_FIX = 1 + 3,
     GF_DEFRAG_CMD_START_FORCE = 1 + 4,
-    GF_DEFRAG_CMD_START_TIER = 1 + 5,
-    GF_DEFRAG_CMD_STATUS_TIER = 1 + 6,
-    GF_DEFRAG_CMD_START_DETACH_TIER = 1 + 7,
-    GF_DEFRAG_CMD_STOP_DETACH_TIER = 1 + 8,
-    GF_DEFRAG_CMD_PAUSE_TIER = 1 + 9,
-    GF_DEFRAG_CMD_RESUME_TIER = 1 + 10,
     GF_DEFRAG_CMD_DETACH_STATUS = 1 + 11,
-    GF_DEFRAG_CMD_STOP_TIER = 1 + 12,
     GF_DEFRAG_CMD_DETACH_START = 1 + 13,
     GF_DEFRAG_CMD_DETACH_COMMIT = 1 + 14,
     GF_DEFRAG_CMD_DETACH_COMMIT_FORCE = 1 + 15,
@@ -466,75 +440,6 @@ struct dht_container {
     int local_subvol_index;
 };
 
-typedef enum tier_mode_ {
-    TIER_MODE_NONE = 0,
-    TIER_MODE_TEST,
-    TIER_MODE_WM
-} tier_mode_t;
-
-typedef enum tier_pause_state_ {
-    TIER_RUNNING = 0,
-    TIER_REQUEST_PAUSE,
-    TIER_PAUSED
-} tier_pause_state_t;
-
-/* This Structure is only used in tiering fixlayout */
-typedef struct gf_tier_fix_layout_arg {
-    xlator_t *this;
-    dict_t *fix_layout;
-    pthread_t thread_id;
-} gf_tier_fix_layout_arg_t;
-
-typedef struct gf_tier_conf {
-    int is_tier;
-    int watermark_hi;
-    int watermark_low;
-    int watermark_last;
-    unsigned long block_size;
-    fsblkcnt_t blocks_total;
-    fsblkcnt_t blocks_used;
-    int percent_full;
-    uint64_t max_migrate_bytes;
-    int max_migrate_files;
-    int query_limit;
-    tier_mode_t mode;
-    /* These flags are only used for tier-compact */
-    gf_boolean_t compact_active;
-    /* These 3 flags are set to true when the client changes the */
-    /* compaction mode on the command line. */
-    /* When they are set, the daemon will trigger compaction as */
-    /* soon as possible to activate or deactivate compaction. */
-    /* If in the middle of a compaction, then the switches take */
-    /* effect on the next compaction, not the current one. */
-    /* If the user switches it off, we want to avoid needless */
-    /* compactions. */
-    /* If the user switches it on, they want to compact as soon */
-    /* as possible. */
-    gf_boolean_t compact_mode_switched;
-    gf_boolean_t compact_mode_switched_hot;
-    gf_boolean_t compact_mode_switched_cold;
-    int tier_max_promote_size;
-    int tier_promote_frequency;
-    int tier_demote_frequency;
-    int tier_compact_hot_frequency;
-    int tier_compact_cold_frequency;
-    uint64_t st_last_promoted_size;
-    uint64_t st_last_demoted_size;
-    tier_pause_state_t pause_state;
-    struct synctask *pause_synctask;
-    gf_timer_t *pause_timer;
-    pthread_mutex_t pause_mutex;
-    int promote_in_progress;
-    int demote_in_progress;
-    /* This Structure is only used in tiering fixlayout */
-    gf_tier_fix_layout_arg_t tier_fix_layout_arg;
-    /* Indicates the index of the first queryfile picked
-     * in the last cycle of promote or demote */
-    int32_t last_promote_qfile_index;
-    int32_t last_demote_qfile_index;
-    char volname[GD_VOLUME_NAME_MAX + 1];
-} gf_tier_conf_t;
-
 typedef struct nodeuuid_info {
     char info;   /* Set to 1 is this is my node's uuid*/
     uuid_t uuid; /* Store the nodeuuid as well for debugging*/
@@ -554,26 +459,18 @@ struct gf_defrag_info_ {
     uint64_t num_dirs_processed;
     uint64_t size_processed;
     gf_lock_t lock;
-    int cmd;
     pthread_t th;
-    gf_defrag_status_t defrag_status;
     struct rpc_clnt *rpc;
     uint32_t connected;
     uint32_t is_exiting;
     pid_t pid;
+    int cmd;
     inode_t *root_inode;
     uuid_t node_uuid;
-    struct timeval start_time;
-    gf_boolean_t stats;
+    time_t start_time;
     uint32_t new_commit_hash;
+    gf_defrag_status_t defrag_status;
     gf_defrag_pattern_list_t *defrag_pattern;
-    gf_tier_conf_t tier_conf;
-
-    /*Data Tiering params for scanner*/
-    uint64_t total_files_promoted;
-    uint64_t total_files_demoted;
-    int write_freq_threshold;
-    int read_freq_threshold;
 
     pthread_cond_t parallel_migration_cond;
     pthread_mutex_t dfq_mutex;
@@ -588,18 +485,20 @@ struct gf_defrag_info_ {
     /*Throttle params*/
     /*stands for reconfigured thread count*/
     int32_t recon_thread_count;
-    /*stands for current running thread count*/
-    int32_t current_thread_count;
     pthread_cond_t df_wakeup_thread;
 
-    /* lock migration flag */
-    gf_boolean_t lock_migration_enabled;
-
     /* backpointer to make it easier to write functions for rebalance */
     xlator_t *this;
 
     pthread_cond_t fc_wakeup_cond;
     pthread_mutex_t fc_mutex;
+
+    /*stands for current running thread count*/
+    int32_t current_thread_count;
+
+    gf_boolean_t stats;
+    /* lock migration flag */
+    gf_boolean_t lock_migration_enabled;
 };
 
 typedef struct gf_defrag_info_ gf_defrag_info_t;
@@ -607,7 +506,6 @@ typedef struct gf_defrag_info_ gf_defrag_info_t;
 struct dht_methods_s {
     int32_t (*migration_get_dst_subvol)(xlator_t *this, dht_local_t *local);
     int32_t (*migration_other)(xlator_t *this, gf_defrag_info_t *defrag);
-    int32_t (*migration_needed)(xlator_t *this);
     xlator_t *(*layout_search)(xlator_t *this, dht_layout_t *layout,
                                const char *name);
 };
@@ -615,36 +513,26 @@ struct dht_methods_s {
 typedef struct dht_methods_s dht_methods_t;
 
 struct dht_conf {
-    gf_lock_t subvolume_lock;
-    int subvolume_cnt;
     xlator_t **subvolumes;
     char *subvolume_status;
     int *last_event;
     dht_layout_t **file_layouts;
     dht_layout_t **dir_layouts;
     unsigned int search_unhashed;
-    gf_boolean_t lookup_optimize;
     int gen;
     dht_du_t *du_stats;
     double min_free_disk;
     double min_free_inodes;
-    char disk_unit;
+    int subvolume_cnt;
     int32_t refresh_interval;
-    gf_boolean_t unhashed_sticky_bit;
-    struct timeval last_stat_fetch;
+    gf_lock_t subvolume_lock;
+    time_t last_stat_fetch;
     gf_lock_t layout_lock;
     dict_t *leaf_to_subvol;
     void *private; /* Can be used by wrapper xlators over
                       dht */
-    gf_boolean_t use_readdirp;
-    char vol_uuid[UUID_SIZE + 1];
-    gf_boolean_t assert_no_child_down;
     time_t *subvol_up_time;
 
-    /* This is the count used as the distribute layout for a directory */
-    /* Will be a global flag to control the layout spread count */
-    uint32_t dir_spread_cnt;
-
     /* to keep track of nodes which are decommissioned */
     xlator_t **decommissioned_bricks;
     int decommission_in_progress;
@@ -653,15 +541,9 @@ struct dht_conf {
     /* defrag related */
     gf_defrag_info_t *defrag;
 
-    /* Request to filter directory entries in readdir request */
-
-    gf_boolean_t readdir_optimize;
-
     /* Support regex-based name reinterpretation. */
     regex_t rsync_regex;
-    gf_boolean_t rsync_regex_valid;
     regex_t extra_regex;
-    gf_boolean_t extra_regex_valid;
 
     /* Support variable xattr names. */
     char *xattr_name;
@@ -670,11 +552,6 @@ struct dht_conf {
     char *commithash_xattr_name;
     char *wild_xattr_name;
 
-    /* Support size-weighted rebalancing (heterogeneous bricks). */
-    gf_boolean_t do_weighting;
-    gf_boolean_t randomize_by_gfid;
-    int dthrottle;
-
     dht_methods_t methods;
 
     struct mem_pool *lock_pool;
@@ -684,24 +561,55 @@ struct dht_conf {
     subvol_nodeuuids_info_t *local_nodeuuids;
     int32_t local_subvols_cnt;
 
+    int dthrottle;
+
+    /* Hard link handle requirement for migration triggered from client*/
+    synclock_t link_lock;
+
+    /* lock migration */
+    gf_lock_t lock;
+
+    /* This is the count used as the distribute layout for a directory */
+    /* Will be a global flag to control the layout spread count */
+    uint32_t dir_spread_cnt;
+
     /*
      * "Commit hash" for this volume topology.  Changed whenever bricks
      * are added or removed.
      */
     uint32_t vol_commit_hash;
-    gf_boolean_t vch_forced;
 
-    /* lock migration */
+    char vol_uuid[UUID_SIZE + 1];
+
+    char disk_unit;
 
     gf_boolean_t lock_migration_enabled;
-    gf_lock_t lock;
 
-    /* Hard link handle requirement for migration triggered from client*/
-    synclock_t link_lock;
+    gf_boolean_t vch_forced;
 
     gf_boolean_t use_fallocate;
 
     gf_boolean_t force_migration;
+
+    gf_boolean_t lookup_optimize;
+
+    gf_boolean_t unhashed_sticky_bit;
+
+    gf_boolean_t assert_no_child_down;
+
+    gf_boolean_t use_readdirp;
+
+    /* Request to filter directory entries in readdir request */
+    gf_boolean_t readdir_optimize;
+
+    gf_boolean_t rsync_regex_valid;
+
+    gf_boolean_t extra_regex_valid;
+
+    /* Support size-weighted rebalancing (heterogeneous bricks). */
+    gf_boolean_t do_weighting;
+
+    gf_boolean_t randomize_by_gfid;
 };
 typedef struct dht_conf dht_conf_t;
 
@@ -740,6 +648,8 @@ struct dir_dfmeta {
     struct list_head **head;
     struct list_head **iterator;
     int *fetch_entries;
+    /* fds corresponding to local subvols only */
+    fd_t **lfd;
 };
 
 typedef struct dht_migrate_info {
@@ -815,22 +725,18 @@ typedef struct dht_fd_ctx {
         dht_local_wipe(__xl, __local);                                         \
     } while (0)
 
-#define DHT_UPDATE_TIME(ctx_sec, ctx_nsec, new_sec, new_nsec, inode, post)     \
+#define DHT_UPDATE_TIME(ctx_sec, ctx_nsec, new_sec, new_nsec, post)            \
     do {                                                                       \
-        LOCK(&inode->lock);                                                    \
-        {                                                                      \
-            if (ctx_sec == new_sec)                                            \
-                new_nsec = max(new_nsec, ctx_nsec);                            \
-            else if (ctx_sec > new_sec) {                                      \
-                new_sec = ctx_sec;                                             \
-                new_nsec = ctx_nsec;                                           \
-            }                                                                  \
-            if (post) {                                                        \
-                ctx_sec = new_sec;                                             \
-                ctx_nsec = new_nsec;                                           \
-            }                                                                  \
+        if (ctx_sec == new_sec)                                                \
+            new_nsec = max(new_nsec, ctx_nsec);                                \
+        else if (ctx_sec > new_sec) {                                          \
+            new_sec = ctx_sec;                                                 \
+            new_nsec = ctx_nsec;                                               \
+        }                                                                      \
+        if (post) {                                                            \
+            ctx_sec = new_sec;                                                 \
+            ctx_nsec = new_nsec;                                               \
         }                                                                      \
-        UNLOCK(&inode->lock);                                                  \
     } while (0)
 
 #define is_greater_time(a, an, b, bn)                                          \
@@ -875,7 +781,6 @@ dht_layout_anomalies(xlator_t *this, loc_t *loc, dht_layout_t *layout,
 int
 dht_layout_dir_mismatch(xlator_t *this, dht_layout_t *layout, xlator_t *subvol,
                         loc_t *loc, dict_t *xattr);
-
 xlator_t *
 dht_linkfile_subvol(xlator_t *this, inode_t *inode, struct iatt *buf,
                     dict_t *xattr);
@@ -893,9 +798,6 @@ int
 dht_disk_layout_extract(xlator_t *this, dht_layout_t *layout, int pos,
                         int32_t **disk_layout_p);
 int
-dht_disk_layout_merge(xlator_t *this, dht_layout_t *layout, int pos,
-                      void *disk_layout_raw, int disk_layout_len);
-int
 dht_disk_layout_extract_for_subvol(xlator_t *this, dht_layout_t *layout,
                                    xlator_t *subvol, int32_t **disk_layout_p);
 
@@ -931,25 +833,17 @@ dht_linkfile_create(call_frame_t *frame, fop_mknod_cbk_t linkfile_cbk,
                     xlator_t *this, xlator_t *tovol, xlator_t *fromvol,
                     loc_t *loc);
 int
-dht_lookup_directory(call_frame_t *frame, xlator_t *this, loc_t *loc);
-int
 dht_lookup_everywhere(call_frame_t *frame, xlator_t *this, loc_t *loc);
 int
 dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t cbk,
                        loc_t *loc, dht_layout_t *layout);
-
-int
-dht_selfheal_directory_for_nameless_lookup(call_frame_t *frame,
-                                           dht_selfheal_dir_cbk_t cbk,
-                                           loc_t *loc, dht_layout_t *layout);
-
 int
 dht_selfheal_new_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t cbk,
                            dht_layout_t *layout);
 int
 dht_selfheal_restore(call_frame_t *frame, dht_selfheal_dir_cbk_t cbk,
                      loc_t *loc, dht_layout_t *layout);
-int
+void
 dht_layout_sort_volname(dht_layout_t *layout);
 
 int
@@ -966,14 +860,14 @@ dht_get_du_info_for_subvol(xlator_t *this, int subvol_idx);
 int
 dht_layout_preset(xlator_t *this, xlator_t *subvol, inode_t *inode);
 int
-dht_layout_index_for_subvol(dht_layout_t *layout, xlator_t *subvol);
-int
 dht_layout_set(xlator_t *this, inode_t *inode, dht_layout_t *layout);
 ;
 void
 dht_layout_unref(xlator_t *this, dht_layout_t *layout);
 dht_layout_t *
 dht_layout_ref(xlator_t *this, dht_layout_t *layout);
+int
+dht_layout_index_for_subvol(dht_layout_t *layout, xlator_t *subvol);
 xlator_t *
 dht_first_up_subvol(xlator_t *this);
 xlator_t *
@@ -1228,25 +1122,19 @@ dht_newfile_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                 struct iatt *preparent, struct iatt *postparent, dict_t *xdata);
 
 int
-gf_defrag_status_get(dht_conf_t *conf, dict_t *dict);
-
-void
-gf_defrag_set_pause_state(gf_tier_conf_t *tier_conf, tier_pause_state_t state);
-
-tier_pause_state_t
-gf_defrag_get_pause_state(gf_tier_conf_t *tier_conf);
+dht_finodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                 int32_t op_ret, int32_t op_errno, dict_t *xdata);
 
 int
-gf_defrag_pause_tier(xlator_t *this, gf_defrag_info_t *defrag);
-
-tier_pause_state_t
-gf_defrag_check_pause_tier(gf_tier_conf_t *defrag);
+dht_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
+                 int op_errno, dict_t *xattr, dict_t *xdata);
 
 int
-gf_defrag_resume_tier(xlator_t *this, gf_defrag_info_t *defrag);
-
+dht_common_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                       int32_t op_ret, int32_t op_errno, dict_t *dict,
+                       dict_t *xdata);
 int
-gf_defrag_start_detach_tier(gf_defrag_info_t *defrag);
+gf_defrag_status_get(dht_conf_t *conf, dict_t *dict);
 
 int
 gf_defrag_stop(dht_conf_t *conf, gf_defrag_status_t status, dict_t *output);
@@ -1279,10 +1167,6 @@ int
 dht_dir_attr_heal(void *data);
 int
 dht_dir_attr_heal_done(int ret, call_frame_t *sync_frame, void *data);
-int
-dht_dir_has_layout(dict_t *xattr, char *name);
-gf_boolean_t
-dht_is_subvol_in_layout(dht_layout_t *layout, xlator_t *xlator);
 xlator_t *
 dht_subvol_with_free_space_inodes(xlator_t *this, xlator_t *subvol,
                                   xlator_t *ignore, dht_layout_t *layout,
@@ -1291,15 +1175,18 @@ xlator_t *
 dht_subvol_maxspace_nonzeroinode(xlator_t *this, xlator_t *subvol,
                                  dht_layout_t *layout);
 int
+dht_dir_has_layout(dict_t *xattr, char *name);
+int
 dht_linkfile_attr_heal(call_frame_t *frame, xlator_t *this);
 
-void
-dht_layout_dump(dht_layout_t *layout, const char *prefix);
 int32_t
 dht_priv_dump(xlator_t *this);
 int32_t
 dht_inodectx_dump(xlator_t *this, inode_t *inode);
 
+gf_boolean_t
+dht_is_subvol_in_layout(dht_layout_t *layout, xlator_t *xlator);
+
 int
 dht_inode_ctx_get_mig_info(xlator_t *this, inode_t *inode,
                            xlator_t **src_subvol, xlator_t **dst_subvol);
@@ -1313,11 +1200,6 @@ dht_subvol_status(dht_conf_t *conf, xlator_t *subvol);
 void
 dht_log_new_layout_for_dir_selfheal(xlator_t *this, loc_t *loc,
                                     dht_layout_t *layout);
-int
-dht_lookup_everywhere_done(call_frame_t *frame, xlator_t *this);
-
-int
-dht_fill_dict_to_avoid_unlink_of_migrating_file(dict_t *dict);
 
 int
 dht_layout_sort(dht_layout_t *layout);
@@ -1334,9 +1216,6 @@ dht_layout_missing_dirs(dht_layout_t *layout);
 int
 dht_refresh_layout(call_frame_t *frame);
 
-gf_boolean_t
-dht_is_tier_xlator(xlator_t *this);
-
 int
 dht_build_parent_loc(xlator_t *this, loc_t *parent, loc_t *child,
                      int32_t *op_errno);
@@ -1367,22 +1246,6 @@ dht_get_lock_subvolume(xlator_t *this, struct gf_flock *lock,
 int
 dht_lk_inode_unref(call_frame_t *frame, int32_t op_ret);
 
-void
-dht_normalize_stats(struct statvfs *buf, unsigned long bsize,
-                    unsigned long frsize);
-
-int
-add_opt(char **optsp, const char *opt);
-
-int
-dht_aggregate_split_brain_xattr(dict_t *dst, char *key, data_t *value);
-
-int
-dht_remove_stale_linkto(void *data);
-
-int
-dht_remove_stale_linkto_cbk(int ret, call_frame_t *sync_frame, void *data);
-
 int
 dht_fd_ctx_set(xlator_t *this, fd_t *fd, xlator_t *subvol);
 
@@ -1454,9 +1317,6 @@ dht_dir_heal_xattrs(void *data);
 int
 dht_dir_heal_xattrs_done(int ret, call_frame_t *sync_frame, void *data);
 
-void
-dht_aggregate_xattr(dict_t *dst, dict_t *src);
-
 int32_t
 dht_dict_set_array(dict_t *dict, char *key, int32_t value[], int32_t size);
 
@@ -1468,25 +1328,12 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
                        dict_t *src, int *uret, int *uflag);
 
 int
-dht_dir_xattr_heal(xlator_t *this, dht_local_t *local);
-
-int32_t
-dht_dict_get_array(dict_t *dict, char *key, int32_t value[], int32_t size,
-                   int *errst);
-
-xlator_t *
-dht_inode_get_hashed_subvol(inode_t *inode, xlator_t *this, loc_t *loc);
+dht_dir_xattr_heal(xlator_t *this, dht_local_t *local, int *op_errno);
 
 int
 dht_common_mark_mdsxattr(call_frame_t *frame, int *errst, int flag);
 
 int
-dht_common_mark_mdsxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                             int op_ret, int op_errno, dict_t *xdata);
-
-int
-dht_inode_ctx_mdsvol_set(inode_t *inode, xlator_t *this, xlator_t *mds_subvol);
-int
 dht_inode_ctx_mdsvol_get(inode_t *inode, xlator_t *this, xlator_t **mdsvol);
 
 int
@@ -1495,14 +1342,6 @@ dht_selfheal_dir_setattr(call_frame_t *frame, loc_t *loc, struct iatt *stbuf,
 
 /* Abstract out the DHT-IATT-IN-DICT */
 
-int
-dht_request_iatt_in_xdata(xlator_t *this, dict_t *xattr_req);
-
-int
-dht_read_iatt_from_xdata(xlator_t *this, dict_t *xdata, struct iatt *stbuf);
-
-int
-is_permission_different(ia_prot_t *prot1, ia_prot_t *prot2);
 void
 dht_selfheal_layout_new_directory(call_frame_t *frame, loc_t *loc,
                                   dht_layout_t *new_layout);
@@ -1523,4 +1362,23 @@ int
 dht_pt_rename(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
               dict_t *xdata);
 
+int32_t
+dht_check_remote_fd_failed_error(dht_local_t *local, int op_ret, int op_errno);
+
+int
+dht_common_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                       int32_t op_ret, int32_t op_errno, dict_t *dict,
+                       dict_t *xdata);
+
+int32_t
+dht_create_lock(call_frame_t *frame, xlator_t *subvol);
+
+int
+dht_set_parent_layout_in_dict(loc_t *loc, xlator_t *this, dht_local_t *local);
+
+int
+dht_dir_layout_error_check(xlator_t *this, inode_t *inode);
+
+int
+dht_inode_ctx_mdsvol_set(inode_t *inode, xlator_t *this, xlator_t *mds_subvol);
 #endif /* _DHT_H */
diff --git a/xlators/cluster/dht/src/dht-diskusage.c b/xlators/cluster/dht/src/dht-diskusage.c
index 148b4f01133..c0588828fdb 100644
--- a/xlators/cluster/dht/src/dht-diskusage.c
+++ b/xlators/cluster/dht/src/dht-diskusage.c
@@ -10,14 +10,10 @@
 
 /* TODO: add NS locking */
 
-#include "glusterfs.h"
-#include "xlator.h"
 #include "dht-common.h"
-#include "dht-messages.h"
-#include "defaults.h"
 
 #include <sys/time.h>
-#include "events.h"
+#include <glusterfs/events.h>
 
 int
 dht_du_info_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
@@ -155,22 +151,18 @@ dht_get_du_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
     dht_conf_t *conf = NULL;
     call_frame_t *statfs_frame = NULL;
     dht_local_t *statfs_local = NULL;
-    struct timeval tv = {
-        0,
-    };
     loc_t tmp_loc = {
         0,
     };
+    time_t now;
 
     conf = this->private;
-
-    gettimeofday(&tv, NULL);
-
+    now = gf_time();
     /* make it root gfid, should be enough to get the proper
        info back */
     tmp_loc.gfid[15] = 1;
 
-    if (tv.tv_sec > (conf->refresh_interval + conf->last_stat_fetch.tv_sec)) {
+    if (now > (conf->refresh_interval + conf->last_stat_fetch)) {
         statfs_frame = copy_frame(frame);
         if (!statfs_frame) {
             goto err;
@@ -202,7 +194,7 @@ dht_get_du_info(call_frame_t *frame, xlator_t *this, loc_t *loc)
                               statfs_local->params);
         }
 
-        conf->last_stat_fetch.tv_sec = tv.tv_sec;
+        conf->last_stat_fetch = now;
     }
     return 0;
 err:
diff --git a/xlators/cluster/dht/src/dht-hashfn.c b/xlators/cluster/dht/src/dht-hashfn.c
index 2f15c0370cc..acda67c312a 100644
--- a/xlators/cluster/dht/src/dht-hashfn.c
+++ b/xlators/cluster/dht/src/dht-hashfn.c
@@ -8,13 +8,12 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "glusterfs.h"
-#include "xlator.h"
 #include "dht-common.h"
-#include "hashfn.h"
+#include <glusterfs/hashfn.h>
 
-int
-dht_hash_compute_internal(int type, const char *name, uint32_t *hash_p)
+static int
+dht_hash_compute_internal(int type, const char *name, const int len,
+                          uint32_t *hash_p)
 {
     int ret = 0;
     uint32_t hash = 0;
@@ -22,7 +21,7 @@ dht_hash_compute_internal(int type, const char *name, uint32_t *hash_p)
     switch (type) {
         case DHT_HASH_TYPE_DM:
         case DHT_HASH_TYPE_DM_USER:
-            hash = gf_dm_hashfn(name, strlen(name));
+            hash = gf_dm_hashfn(name, len);
             break;
         default:
             ret = -1;
@@ -36,7 +35,12 @@ dht_hash_compute_internal(int type, const char *name, uint32_t *hash_p)
     return ret;
 }
 
-static gf_boolean_t
+/* The function returns:
+ * 0  : in case no munge took place
+ * >0 : the length (inc. terminating NULL!) of the newly modified string,
+ *      if it was munged.
+ */
+static int
 dht_munge_name(const char *original, char *modified, size_t len, regex_t *re)
 {
     regmatch_t matches[2] = {
@@ -54,14 +58,14 @@ dht_munge_name(const char *original, char *modified, size_t len, regex_t *re)
             if (new_len < len) {
                 memcpy(modified, original + matches[1].rm_so, new_len);
                 modified[new_len] = '\0';
-                return _gf_true;
+                return new_len + 1; /* +1 for the terminating NULL */
             }
         }
     }
 
     /* This is guaranteed safe because of how the dest was allocated. */
     strcpy(modified, original);
-    return _gf_false;
+    return 0;
 }
 
 int
@@ -70,36 +74,37 @@ dht_hash_compute(xlator_t *this, int type, const char *name, uint32_t *hash_p)
     char *rsync_friendly_name = NULL;
     dht_conf_t *priv = NULL;
     size_t len = 0;
-    gf_boolean_t munged = _gf_false;
+    int munged = 0;
 
     priv = this->private;
 
+    if (name == NULL)
+        return -1;
+
+    len = strlen(name) + 1;
+    rsync_friendly_name = alloca(len);
+
     LOCK(&priv->lock);
     {
         if (priv->extra_regex_valid) {
-            len = strlen(name) + 1;
-            rsync_friendly_name = alloca(len);
             munged = dht_munge_name(name, rsync_friendly_name, len,
                                     &priv->extra_regex);
         }
 
         if (!munged && priv->rsync_regex_valid) {
-            len = strlen(name) + 1;
-            rsync_friendly_name = alloca(len);
             gf_msg_trace(this->name, 0, "trying regex for %s", name);
             munged = dht_munge_name(name, rsync_friendly_name, len,
                                     &priv->rsync_regex);
-            if (munged) {
-                gf_msg_debug(this->name, 0, "munged down to %s",
-                             rsync_friendly_name);
-            }
         }
     }
     UNLOCK(&priv->lock);
-
-    if (!munged) {
+    if (munged) {
+        gf_msg_debug(this->name, 0, "munged down to %s", rsync_friendly_name);
+        len = munged;
+    } else {
         rsync_friendly_name = (char *)name;
     }
 
-    return dht_hash_compute_internal(type, rsync_friendly_name, hash_p);
+    return dht_hash_compute_internal(type, rsync_friendly_name, len - 1,
+                                     hash_p);
 }
diff --git a/xlators/cluster/dht/src/dht-helper.c b/xlators/cluster/dht/src/dht-helper.c
index 22ce8bc4d4c..3f2fe43d5f3 100644
--- a/xlators/cluster/dht/src/dht-helper.c
+++ b/xlators/cluster/dht/src/dht-helper.c
@@ -8,10 +8,9 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "glusterfs.h"
-#include "xlator.h"
 #include "dht-common.h"
 #include "dht-lock.h"
+#include "glusterfs/compat-errno.h"  // for ENODATA on BSD
 
 static void
 dht_free_fd_ctx(dht_fd_ctx_t *fd_ctx)
@@ -65,8 +64,8 @@ __dht_fd_ctx_set(xlator_t *this, fd_t *fd, xlator_t *dst)
 
     ret = __fd_ctx_set(fd, this, value);
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_FD_CTX_SET_FAILED,
-               "Failed to set fd ctx in fd=0x%p", fd);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_FD_CTX_SET_FAILED,
+                "fd=0x%p", fd, NULL);
         GF_REF_PUT(fd_ctx);
     }
 out:
@@ -94,12 +93,13 @@ dht_fd_ctx_set(xlator_t *this, fd_t *fd, xlator_t *dst)
                 goto unlock;
             } else {
                 /* This would be a big problem*/
-                gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_INVALID_VALUE,
-                       "Different dst found in the fd ctx");
-
                 /* Overwrite and hope for the best*/
                 fd_ctx->opened_on_dst = (uint64_t)(uintptr_t)dst;
-                goto unlock;
+                UNLOCK(&fd->lock);
+                gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_INVALID_VALUE,
+                        NULL);
+
+                goto out;
             }
         }
         ret = __dht_fd_ctx_set(this, fd, dst);
@@ -124,13 +124,13 @@ dht_fd_ctx_get(xlator_t *this, fd_t *fd)
     {
         ret = __fd_ctx_get(fd, this, &tmp_val);
         if ((ret < 0) || (tmp_val == 0)) {
-            UNLOCK(&fd->lock);
-            goto out;
+            goto unlock;
         }
 
         fd_ctx = (dht_fd_ctx_t *)(uintptr_t)tmp_val;
         GF_REF_GET(fd_ctx);
     }
+unlock:
     UNLOCK(&fd->lock);
 
 out:
@@ -365,10 +365,27 @@ dht_check_and_open_fd_on_subvol_complete(int ret, call_frame_t *frame,
 
             break;
 
+        case GF_FOP_FXATTROP:
+            STACK_WIND(frame, dht_common_xattrop_cbk, subvol,
+                       subvol->fops->fxattrop, local->fd,
+                       local->rebalance.flags, local->rebalance.xattr,
+                       local->xattr_req);
+            break;
+
+        case GF_FOP_FGETXATTR:
+            STACK_WIND(frame, dht_getxattr_cbk, subvol, subvol->fops->fgetxattr,
+                       local->fd, local->key, NULL);
+            break;
+
+        case GF_FOP_FINODELK:
+            STACK_WIND(frame, dht_finodelk_cbk, subvol, subvol->fops->finodelk,
+                       local->key, local->fd, local->rebalance.lock_cmd,
+                       &local->rebalance.flock, local->xattr_req);
+            break;
         default:
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UNKNOWN_FOP,
-                   "Unknown FOP on fd (%p) on file %s @ %s", fd,
-                   uuid_utoa(fd->inode->gfid), subvol->name);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UNKNOWN_FOP, "fd=%p",
+                    fd, "gfid=%s", uuid_utoa(fd->inode->gfid), "name=%s",
+                    subvol->name, NULL);
             break;
     }
 
@@ -428,10 +445,22 @@ handle_err:
             DHT_STACK_UNWIND(fremovexattr, frame, -1, op_errno, NULL);
             break;
 
+        case GF_FOP_FXATTROP:
+            DHT_STACK_UNWIND(fxattrop, frame, -1, op_errno, NULL, NULL);
+            break;
+
+        case GF_FOP_FGETXATTR:
+            DHT_STACK_UNWIND(fgetxattr, frame, -1, op_errno, NULL, NULL);
+            break;
+
+        case GF_FOP_FINODELK:
+            DHT_STACK_UNWIND(finodelk, frame, -1, op_errno, NULL);
+            break;
+
         default:
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UNKNOWN_FOP,
-                   "Unknown FOP on fd (%p) on file %s @ %s", fd,
-                   uuid_utoa(fd->inode->gfid), subvol->name);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_UNKNOWN_FOP, "fd=%p",
+                    fd, "gfid=%s", uuid_utoa(fd->inode->gfid), "name=%s",
+                    subvol->name, NULL);
             break;
     }
 
@@ -484,10 +513,9 @@ dht_check_and_open_fd_on_subvol_task(void *data)
                       fd, NULL, NULL);
 
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_OPEN_FD_ON_DST_FAILED,
-               "Failed to open the fd"
-               " (%p, flags=0%o) on file %s @ %s",
-               fd, fd->flags, uuid_utoa(fd->inode->gfid), subvol->name);
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_OPEN_FD_ON_DST_FAILED,
+                "fd=%p", fd, "flags=0%o", fd->flags, "gfid=%s",
+                uuid_utoa(fd->inode->gfid), "name=%s", subvol->name, NULL);
         /* This can happen if the cached subvol was updated in the
          * inode_ctx and the fd was opened on the new cached suvol
          * after this fop was wound on the old cached subvol.
@@ -533,10 +561,8 @@ dht_check_and_open_fd_on_subvol(xlator_t *this, call_frame_t *frame)
                        dht_check_and_open_fd_on_subvol_complete, frame, frame);
 
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, 0,
-               "Failed to create synctask"
-               " to check and open fd=%p",
-               local->fd);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SYNCTASK_CREATE_FAILED,
+                "to-check-and-open fd=%p", local->fd, NULL);
     }
 
     return ret;
@@ -645,9 +671,7 @@ dht_get_subvol_from_id(xlator_t *this, int client_id)
 
     ret = gf_asprintf(&sid, "%d", client_id);
     if (ret == -1) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_ASPRINTF_FAILED,
-               "asprintf failed while "
-               "fetching subvol from the id");
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_ASPRINTF_FAILED, NULL);
         goto out;
     }
 
@@ -1260,6 +1284,7 @@ dht_migration_complete_check_task(void *data)
     fd_t *tmp = NULL;
     uint64_t tmp_miginfo = 0;
     dht_migrate_info_t *miginfo = NULL;
+    gf_boolean_t skip_open = _gf_false;
     int open_failed = 0;
 
     this = THIS;
@@ -1306,9 +1331,9 @@ dht_migration_complete_check_task(void *data)
              * migrated by two different layers. Raise
              * a warning here.
              */
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_HAS_MIGINFO,
-                   "%s: Found miginfo in the inode ctx",
-                   tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid));
+            gf_smsg(
+                this->name, GF_LOG_WARNING, 0, DHT_MSG_HAS_MIGINFO, "tmp=%s",
+                tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid), NULL);
 
             miginfo = (void *)(uintptr_t)tmp_miginfo;
             GF_REF_PUT(miginfo);
@@ -1329,10 +1354,9 @@ dht_migration_complete_check_task(void *data)
 
     ret = syncop_lookup(this, &tmp_loc, &stbuf, 0, 0, 0);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_FILE_LOOKUP_FAILED,
-               "%s: failed to lookup the file on %s",
-               tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
-               this->name);
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_FILE_LOOKUP_FAILED,
+                "tmp=%s", tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
+                "name=%s", this->name, NULL);
         local->op_errno = -ret;
         ret = -1;
         goto out;
@@ -1340,18 +1364,15 @@ dht_migration_complete_check_task(void *data)
 
     dst_node = dht_subvol_get_cached(this, tmp_loc.inode);
     if (linkto_target && dst_node != linkto_target) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_INVALID_LINKFILE,
-               "linkto target (%s) is "
-               "different from cached-subvol (%s). Treating %s as "
-               "destination subvol",
-               linkto_target->name, dst_node->name, dst_node->name);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_INVALID_LINKFILE,
+                "linkto_target_name=%s", linkto_target->name, "dst_name=%s",
+                dst_node->name, NULL);
     }
 
     if (gf_uuid_compare(stbuf.ia_gfid, tmp_loc.inode->gfid)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_GFID_MISMATCH,
-               "%s: gfid different on the target file on %s",
-               tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
-               dst_node->name);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_GFID_MISMATCH, "tmp=%s",
+                tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
+                "dst_name=%s", dst_node->name, NULL);
         ret = -1;
         local->op_errno = EIO;
         goto out;
@@ -1398,24 +1419,34 @@ dht_migration_complete_check_task(void *data)
      * the loop will cause the destruction of the fd. So we need to
      * iterate the list safely because iter_fd cannot be trusted.
      */
-    list_for_each_entry_safe(iter_fd, tmp, &inode->fd_list, inode_list)
-    {
-        if (fd_is_anonymous(iter_fd))
-            continue;
-
-        if (dht_fd_open_on_dst(this, iter_fd, dst_node))
-            continue;
-
+    iter_fd = list_entry((&inode->fd_list)->next, typeof(*iter_fd), inode_list);
+    while (&iter_fd->inode_list != (&inode->fd_list)) {
+        if (fd_is_anonymous(iter_fd) ||
+            (dht_fd_open_on_dst(this, iter_fd, dst_node))) {
+            if (!tmp) {
+                iter_fd = list_entry(iter_fd->inode_list.next, typeof(*iter_fd),
+                                     inode_list);
+                continue;
+            }
+            skip_open = _gf_true;
+        }
         /* We need to release the inode->lock before calling
          * syncop_open() to avoid possible deadlocks. However this
          * can cause the iter_fd to be released by other threads.
          * To avoid this, we take a reference before releasing the
          * lock.
          */
-        __fd_ref(iter_fd);
+        fd_ref(iter_fd);
 
         UNLOCK(&inode->lock);
 
+        if (tmp) {
+            fd_unref(tmp);
+            tmp = NULL;
+        }
+        if (skip_open)
+            goto next;
+
         /* flags for open are stripped down to allow following the
          * new location of the file, otherwise we can get EEXIST or
          * truncate the file again as rebalance is moving the data */
@@ -1423,12 +1454,10 @@ dht_migration_complete_check_task(void *data)
                           (iter_fd->flags & ~(O_CREAT | O_EXCL | O_TRUNC)),
                           iter_fd, NULL, NULL);
         if (ret < 0) {
-            gf_msg(this->name, GF_LOG_ERROR, -ret,
-                   DHT_MSG_OPEN_FD_ON_DST_FAILED,
-                   "failed"
-                   " to open the fd"
-                   " (%p, flags=0%o) on file %s @ %s",
-                   iter_fd, iter_fd->flags, path, dst_node->name);
+            gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                    DHT_MSG_OPEN_FD_ON_DST_FAILED, "id=%p", iter_fd,
+                    "flags=0%o", iter_fd->flags, "path=%s", path, "name=%s",
+                    dst_node->name, NULL);
 
             open_failed = 1;
             local->op_errno = -ret;
@@ -1437,9 +1466,11 @@ dht_migration_complete_check_task(void *data)
             dht_fd_ctx_set(this, iter_fd, dst_node);
         }
 
-        fd_unref(iter_fd);
-
+    next:
         LOCK(&inode->lock);
+        skip_open = _gf_false;
+        tmp = iter_fd;
+        iter_fd = list_entry(tmp->inode_list.next, typeof(*tmp), inode_list);
     }
 
     SYNCTASK_SETID(frame->root->uid, frame->root->gid);
@@ -1452,6 +1483,10 @@ dht_migration_complete_check_task(void *data)
 
 unlock:
     UNLOCK(&inode->lock);
+    if (tmp) {
+        fd_unref(tmp);
+        tmp = NULL;
+    }
 
 out:
     if (dict) {
@@ -1533,6 +1568,7 @@ dht_rebalance_inprogress_task(void *data)
     int open_failed = 0;
     uint64_t tmp_miginfo = 0;
     dht_migrate_info_t *miginfo = NULL;
+    gf_boolean_t skip_open = _gf_false;
 
     this = THIS;
     frame = data;
@@ -1575,9 +1611,9 @@ dht_rebalance_inprogress_task(void *data)
              * migrated by two different layers. Raise
              * a warning here.
              */
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_HAS_MIGINFO,
-                   "%s: Found miginfo in the inode ctx",
-                   tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid));
+            gf_smsg(
+                this->name, GF_LOG_WARNING, 0, DHT_MSG_HAS_MIGINFO, "tmp=%s",
+                tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid), NULL);
             miginfo = (void *)(uintptr_t)tmp_miginfo;
             GF_REF_PUT(miginfo);
         }
@@ -1586,17 +1622,16 @@ dht_rebalance_inprogress_task(void *data)
     }
 
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_GET_XATTR_FAILED,
-               "%s: failed to get the 'linkto' xattr", local->loc.path);
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_GET_XATTR_FAILED,
+                "path=%s", local->loc.path, NULL);
         ret = -1;
         goto out;
     }
 
     dst_node = dht_linkfile_subvol(this, NULL, NULL, dict);
     if (!dst_node) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SUBVOL_NOT_FOUND,
-               "%s: failed to get the 'linkto' xattr from dict",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_GET_XATTR_FAILED,
+                "path=%s", local->loc.path, NULL);
         ret = -1;
         goto out;
     }
@@ -1613,20 +1648,17 @@ dht_rebalance_inprogress_task(void *data)
     /* lookup on dst */
     ret = syncop_lookup(dst_node, &tmp_loc, &stbuf, NULL, NULL, NULL);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret,
-               DHT_MSG_FILE_LOOKUP_ON_DST_FAILED,
-               "%s: failed to lookup the file on %s",
-               tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
-               dst_node->name);
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_FILE_LOOKUP_FAILED,
+                "tmp=%s", tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
+                "name=%s", dst_node->name, NULL);
         ret = -1;
         goto out;
     }
 
     if (gf_uuid_compare(stbuf.ia_gfid, tmp_loc.inode->gfid)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_GFID_MISMATCH,
-               "%s: gfid different on the target file on %s",
-               tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
-               dst_node->name);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_GFID_MISMATCH, "tmp=%s",
+                tmp_loc.path ? tmp_loc.path : uuid_utoa(tmp_loc.gfid),
+                "name=%s", dst_node->name, NULL);
         ret = -1;
         goto out;
     }
@@ -1653,24 +1685,40 @@ dht_rebalance_inprogress_task(void *data)
      * the loop will cause the destruction of the fd. So we need to
      * iterate the list safely because iter_fd cannot be trusted.
      */
-    list_for_each_entry_safe(iter_fd, tmp, &inode->fd_list, inode_list)
-    {
-        if (fd_is_anonymous(iter_fd))
-            continue;
-
-        if (dht_fd_open_on_dst(this, iter_fd, dst_node))
-            continue;
-
+    iter_fd = list_entry((&inode->fd_list)->next, typeof(*iter_fd), inode_list);
+    while (&iter_fd->inode_list != (&inode->fd_list)) {
         /* We need to release the inode->lock before calling
          * syncop_open() to avoid possible deadlocks. However this
          * can cause the iter_fd to be released by other threads.
          * To avoid this, we take a reference before releasing the
          * lock.
          */
-        __fd_ref(iter_fd);
 
+        if (fd_is_anonymous(iter_fd) ||
+            (dht_fd_open_on_dst(this, iter_fd, dst_node))) {
+            if (!tmp) {
+                iter_fd = list_entry(iter_fd->inode_list.next, typeof(*iter_fd),
+                                     inode_list);
+                continue;
+            }
+            skip_open = _gf_true;
+        }
+
+        /* Yes, this is ugly but there isn't a cleaner way to do this
+         * the fd_ref is an atomic increment so not too bad. We want to
+         * reduce the number of inode locks and unlocks.
+         */
+
+        fd_ref(iter_fd);
         UNLOCK(&inode->lock);
 
+        if (tmp) {
+            fd_unref(tmp);
+            tmp = NULL;
+        }
+        if (skip_open)
+            goto next;
+
         /* flags for open are stripped down to allow following the
          * new location of the file, otherwise we can get EEXIST or
          * truncate the file again as rebalance is moving the data */
@@ -1678,11 +1726,10 @@ dht_rebalance_inprogress_task(void *data)
                           (iter_fd->flags & ~(O_CREAT | O_EXCL | O_TRUNC)),
                           iter_fd, NULL, NULL);
         if (ret < 0) {
-            gf_msg(this->name, GF_LOG_ERROR, -ret,
-                   DHT_MSG_OPEN_FD_ON_DST_FAILED,
-                   "failed to send open "
-                   "the fd (%p, flags=0%o) on file %s @ %s",
-                   iter_fd, iter_fd->flags, path, dst_node->name);
+            gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                    DHT_MSG_OPEN_FD_ON_DST_FAILED, "fd=%p", iter_fd,
+                    "flags=0%o", iter_fd->flags, "path=%s", path, "name=%s",
+                    dst_node->name, NULL);
             ret = -1;
             open_failed = 1;
         } else {
@@ -1691,9 +1738,11 @@ dht_rebalance_inprogress_task(void *data)
             dht_fd_ctx_set(this, iter_fd, dst_node);
         }
 
-        fd_unref(iter_fd);
-
+    next:
         LOCK(&inode->lock);
+        skip_open = _gf_false;
+        tmp = iter_fd;
+        iter_fd = list_entry(tmp->inode_list.next, typeof(*tmp), inode_list);
     }
 
     SYNCTASK_SETID(frame->root->uid, frame->root->gid);
@@ -1701,6 +1750,10 @@ dht_rebalance_inprogress_task(void *data)
 unlock:
     UNLOCK(&inode->lock);
 
+    if (tmp) {
+        fd_unref(tmp);
+        tmp = NULL;
+    }
     if (open_failed) {
         ret = -1;
         goto out;
@@ -1708,9 +1761,8 @@ unlock:
 
     ret = dht_inode_ctx_set_mig_info(this, inode, src_node, dst_node);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SET_INODE_CTX_FAILED,
-               "%s: failed to set inode-ctx target file at %s", local->loc.path,
-               dst_node->name);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SET_INODE_CTX_FAILED,
+                "path=%s", local->loc.path, "name=%s", dst_node->name, NULL);
         goto out;
     }
 
@@ -1803,12 +1855,16 @@ dht_inode_ctx_time_update(inode_t *inode, xlator_t *this, struct iatt *stat,
 
     time = &ctx->time;
 
-    DHT_UPDATE_TIME(time->mtime, time->mtime_nsec, stat->ia_mtime,
-                    stat->ia_mtime_nsec, inode, post);
-    DHT_UPDATE_TIME(time->ctime, time->ctime_nsec, stat->ia_ctime,
-                    stat->ia_ctime_nsec, inode, post);
-    DHT_UPDATE_TIME(time->atime, time->atime_nsec, stat->ia_atime,
-                    stat->ia_atime_nsec, inode, post);
+    LOCK(&inode->lock);
+    {
+        DHT_UPDATE_TIME(time->mtime, time->mtime_nsec, stat->ia_mtime,
+                        stat->ia_mtime_nsec, post);
+        DHT_UPDATE_TIME(time->ctime, time->ctime_nsec, stat->ia_ctime,
+                        stat->ia_ctime_nsec, post);
+        DHT_UPDATE_TIME(time->atime, time->atime_nsec, stat->ia_atime,
+                        stat->ia_atime_nsec, post);
+    }
+    UNLOCK(&inode->lock);
 
     ret = dht_inode_ctx_set(inode, this, ctx);
 out:
@@ -1877,9 +1933,7 @@ dht_heal_path(xlator_t *this, char *path, inode_table_t *itable)
     };
     char *bname = NULL;
     char *save_ptr = NULL;
-    uuid_t gfid = {
-        0,
-    };
+    static uuid_t gfid = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1};
     char *tmp_path = NULL;
 
     tmp_path = gf_strdup(path);
@@ -1887,9 +1941,6 @@ dht_heal_path(xlator_t *this, char *path, inode_table_t *itable)
         goto out;
     }
 
-    memset(gfid, 0, 16);
-    gfid[15] = 1;
-
     gf_uuid_copy(loc.pargfid, gfid);
     loc.parent = inode_ref(itable->root);
 
@@ -1918,10 +1969,10 @@ dht_heal_path(xlator_t *this, char *path, inode_table_t *itable)
              */
             linked_inode = loc.inode;
             bname = strtok_r(NULL, "/", &save_ptr);
-            inode_unref(loc.parent);
             if (!bname) {
                 goto out;
             }
+            inode_unref(loc.parent);
             loc.parent = loc.inode;
             gf_uuid_copy(loc.pargfid, loc.inode->gfid);
             loc.inode = NULL;
@@ -1933,10 +1984,9 @@ dht_heal_path(xlator_t *this, char *path, inode_table_t *itable)
 
         ret = syncop_lookup(this, &loc, &iatt, NULL, NULL, NULL);
         if (ret) {
-            gf_msg(this->name, GF_LOG_INFO, -ret, DHT_MSG_DIR_SELFHEAL_FAILED,
-                   "Healing of path %s failed on subvolume %s for "
-                   "directory %s",
-                   path, this->name, bname);
+            gf_smsg(this->name, GF_LOG_INFO, -ret, DHT_MSG_DIR_SELFHEAL_FAILED,
+                    "path=%s", path, "subvolume=%s", this->name, "bname=%s",
+                    bname, NULL);
             goto out;
         }
 
@@ -1994,10 +2044,8 @@ dht_heal_full_path(void *data)
     ret = syncop_getxattr(source, &loc, &dict, GET_ANCESTRY_PATH_KEY, NULL,
                           NULL);
     if (ret) {
-        gf_msg(this->name, GF_LOG_INFO, -ret, DHT_MSG_DIR_SELFHEAL_FAILED,
-               "Failed to get path from subvol %s. Aborting "
-               "directory healing.",
-               source->name);
+        gf_smsg(this->name, GF_LOG_INFO, -ret, DHT_MSG_DIR_HEAL_ABORT,
+                "subvol=%s", source->name, NULL);
         goto out;
     }
 
@@ -2035,6 +2083,7 @@ dht_heal_full_path_done(int op_ret, call_frame_t *heal_frame, void *data)
     dht_local_t *local = NULL;
     xlator_t *this = NULL;
     int ret = -1;
+    int op_errno = 0;
 
     local = heal_frame->local;
     main_frame = local->main_frame;
@@ -2044,10 +2093,12 @@ dht_heal_full_path_done(int op_ret, call_frame_t *heal_frame, void *data)
     dht_set_fixed_dir_stat(&local->postparent);
     if (local->need_xattr_heal) {
         local->need_xattr_heal = 0;
-        ret = dht_dir_xattr_heal(this, local);
-        if (ret)
-            gf_msg(this->name, GF_LOG_ERROR, ret, DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                   "xattr heal failed for directory  %s ", local->loc.path);
+        ret = dht_dir_xattr_heal(this, local, &op_errno);
+        if (ret) {
+            gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                    DHT_MSG_DIR_XATTR_HEAL_FAILED, "path=%s", local->loc.path,
+                    NULL);
+        }
     }
 
     DHT_STACK_UNWIND(lookup, main_frame, 0, 0, local->inode, &local->stbuf,
@@ -2134,16 +2185,15 @@ dht_get_lock_subvolume(xlator_t *this, struct gf_flock *lock,
         ret = __dht_lock_subvol_set(inode, this, cached_subvol);
         if (ret) {
             gf_uuid_unparse(inode->gfid, gfid);
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_SET_INODE_CTX_FAILED,
-                   "Failed to set lock_subvol in "
-                   "inode ctx for gfid %s",
-                   gfid);
-            goto unlock;
+            UNLOCK(&inode->lock);
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_SET_INODE_CTX_FAILED,
+                    "lock_subvol gfid=%s", gfid, NULL);
+            goto post_unlock;
         }
         subvol = cached_subvol;
     }
-unlock:
     UNLOCK(&inode->lock);
+post_unlock:
     if (!subvol && inode && lock->l_type != F_UNLCK) {
         inode_unref(inode);
     }
@@ -2167,8 +2217,8 @@ dht_lk_inode_unref(call_frame_t *frame, int32_t op_ret)
         inode = local->loc.inode ? local->loc.inode : local->fd->inode;
     }
     if (!inode) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LOCK_INODE_UNREF_FAILED,
-               "Found a NULL inode. Failed to unref the inode");
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LOCK_INODE_UNREF_FAILED,
+                NULL);
         goto out;
     }
 
@@ -2194,11 +2244,8 @@ dht_lk_inode_unref(call_frame_t *frame, int32_t op_ret)
                 inode_unref(inode);
             } else {
                 gf_uuid_unparse(inode->gfid, gfid);
-                gf_msg(this->name, GF_LOG_WARNING, 0,
-                       DHT_MSG_LOCK_INODE_UNREF_FAILED,
-                       "Unlock request failed for gfid %s."
-                       "Failed to unref the inode",
-                       gfid);
+                gf_smsg(this->name, GF_LOG_WARNING, 0,
+                        DHT_MSG_LOCK_INODE_UNREF_FAILED, "gfid=%s", gfid, NULL);
                 goto out;
             }
         default:
@@ -2220,12 +2267,11 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
     int luret = -1;
     int luflag = -1;
     int i = 0;
+    char **xattrs_to_heal;
 
     if (!src || !dst) {
-        gf_msg(this->name, GF_LOG_WARNING, EINVAL, DHT_MSG_DICT_SET_FAILED,
-               "src or dst is NULL. Failed to set "
-               " dictionary value for path %s",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_WARNING, EINVAL, DHT_MSG_DST_NULL_SET_FAILED,
+                "path=%s", local->loc.path, NULL);
         return;
     }
     /* Check if any user xattr present in src dict and set
@@ -2236,17 +2282,18 @@ dht_dir_set_heal_xattr(xlator_t *this, dht_local_t *local, dict_t *dst,
        and set it to dst dict, here index start from 1 because
        user xattr already checked in previous statement
     */
+
+    xattrs_to_heal = get_xattrs_to_heal();
+
     for (i = 1; xattrs_to_heal[i]; i++) {
         keyval = dict_get(src, xattrs_to_heal[i]);
         if (keyval) {
             luflag = 1;
             ret = dict_set(dst, xattrs_to_heal[i], keyval);
             if (ret)
-                gf_msg(this->name, GF_LOG_WARNING, ENOMEM,
-                       DHT_MSG_DICT_SET_FAILED,
-                       "Failed to set dictionary value:key = %s for "
-                       "path %s",
-                       xattrs_to_heal[i], local->loc.path);
+                gf_smsg(this->name, GF_LOG_WARNING, ENOMEM,
+                        DHT_MSG_DICT_SET_FAILED, "key=%s", xattrs_to_heal[i],
+                        "path=%s", local->loc.path, NULL);
             keyval = NULL;
         }
     }
diff --git a/xlators/cluster/dht/src/dht-inode-read.c b/xlators/cluster/dht/src/dht-inode-read.c
index f46370a9208..dbb8070b0da 100644
--- a/xlators/cluster/dht/src/dht-inode-read.c
+++ b/xlators/cluster/dht/src/dht-inode-read.c
@@ -10,25 +10,25 @@
 
 #include "dht-common.h"
 
-int
+static int
 dht_access2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_readv2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_attr2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_open2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_flush2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_lk2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_fsync2(xlator_t *this, xlator_t *dst_node, call_frame_t *frame, int ret);
-int
+static int
 dht_common_xattrop2(xlator_t *this, xlator_t *subvol, call_frame_t *frame,
                     int ret);
 
-int
+static int
 dht_open_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
              int op_errno, fd_t *fd, dict_t *xdata)
 {
@@ -67,7 +67,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_open2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -162,8 +162,8 @@ dht_file_attr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     local = frame->local;
     prev = cookie;
 
-    if ((local->fop == GF_FOP_FSTAT) && (op_ret == -1) && (op_errno == EBADF) &&
-        !(local->fd_checked)) {
+    if ((local->fop == GF_FOP_FSTAT) &&
+        dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -216,7 +216,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_attr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -258,7 +258,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_attr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
              int op_errno, struct iatt *stbuf, dict_t *xdata)
 {
@@ -272,19 +272,19 @@ dht_attr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     {
         if (op_ret == -1) {
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
                          prev->name);
 
-            goto unlock;
+            goto post_unlock;
         }
 
         dht_iatt_merge(this, &local->stbuf, stbuf);
 
         local->op_ret = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt)) {
         DHT_STACK_UNWIND(stat, frame, local->op_ret, local->op_errno,
@@ -431,7 +431,7 @@ dht_readv_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     if (local->call_cnt != 1)
         goto out;
 
-    if (op_ret == -1 && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -473,7 +473,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_readv2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -558,7 +558,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_access_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                int op_errno, dict_t *xdata)
 {
@@ -606,7 +606,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_access2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -703,7 +703,7 @@ dht_flush_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     if (local->call_cnt != 1)
         goto out;
 
-    if (op_ret == -1 && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -735,7 +735,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_flush2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -820,7 +820,7 @@ dht_fsync_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
 
     local->op_errno = op_errno;
 
-    if (op_ret == -1 && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -881,7 +881,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_fsync2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -959,7 +959,7 @@ err:
 /* TODO: for 'lk()' call, we need some other special error, may be ESTALE to
    indicate that lock migration happened on the fd, so we can consider it as
    phase 2 of migration */
-int
+static int
 dht_lk_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
            int op_errno, struct gf_flock *flock, dict_t *xdata)
 {
@@ -1006,7 +1006,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_lk2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -1087,7 +1087,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_lease_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
               int op_errno, struct gf_lease *lease, dict_t *xdata)
 {
@@ -1129,7 +1129,7 @@ err:
 }
 
 /* Symlinks are currently not migrated, so no need for any check here */
-int
+static int
 dht_readlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
                  int op_errno, const char *path, struct iatt *stbuf,
                  dict_t *xdata)
@@ -1192,6 +1192,29 @@ err:
     return 0;
 }
 
+/* Get both DHT_IATT_IN_XDATA_KEY and DHT_MODE_IN_XDATA_KEY
+ * Use DHT_MODE_IN_XDATA_KEY if available, else fall back to
+ * DHT_IATT_IN_XDATA_KEY
+ * This will return a dummy iatt with only the mode and type set
+ */
+static int
+dht_read_iatt_from_xdata(dict_t *xdata, struct iatt *stbuf)
+{
+    int ret = -1;
+    int32_t mode = 0;
+
+    ret = dict_get_int32(xdata, DHT_MODE_IN_XDATA_KEY, &mode);
+
+    if (ret) {
+        ret = dict_get_bin(xdata, DHT_IATT_IN_XDATA_KEY, (void **)&stbuf);
+    } else {
+        stbuf->ia_prot = ia_prot_from_st_mode(mode);
+        stbuf->ia_type = ia_type_from_st_mode(mode);
+    }
+
+    return ret;
+}
+
 int
 dht_common_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                        int32_t op_ret, int32_t op_errno, dict_t *dict,
@@ -1223,7 +1246,14 @@ dht_common_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     if (local->call_cnt != 1)
         goto out;
 
-    ret = dht_read_iatt_from_xdata(this, xdata, &stbuf);
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
+        ret = dht_check_and_open_fd_on_subvol(this, frame);
+        if (ret)
+            goto out;
+        return 0;
+    }
+
+    ret = dht_read_iatt_from_xdata(xdata, &stbuf);
 
     if ((!op_ret) && (ret)) {
         /* This is a potential problem and can cause corruption
@@ -1275,7 +1305,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_common_xattrop2(xlator_t *this, xlator_t *subvol, call_frame_t *frame,
                     int ret)
 {
@@ -1334,7 +1364,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                 int32_t op_ret, int32_t op_errno, dict_t *dict, dict_t *xdata)
 {
@@ -1342,6 +1372,22 @@ dht_xattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     return 0;
 }
 
+/* Set both DHT_IATT_IN_XDATA_KEY and DHT_MODE_IN_XDATA_KEY
+ * Use DHT_MODE_IN_XDATA_KEY if available. Else fall back to
+ * DHT_IATT_IN_XDATA_KEY
+ */
+static int
+dht_request_iatt_in_xdata(dict_t *xattr_req)
+{
+    int ret = -1;
+
+    ret = dict_set_int8(xattr_req, DHT_MODE_IN_XDATA_KEY, 1);
+    ret = dict_set_int8(xattr_req, DHT_IATT_IN_XDATA_KEY, 1);
+
+    /* At least one call succeeded */
+    return ret;
+}
+
 int
 dht_xattrop(call_frame_t *frame, xlator_t *this, loc_t *loc,
             gf_xattrop_flags_t flags, dict_t *dict, dict_t *xdata)
@@ -1384,7 +1430,7 @@ dht_xattrop(call_frame_t *frame, xlator_t *this, loc_t *loc,
         local->rebalance.xattr = dict_ref(dict);
         local->rebalance.flags = flags;
 
-        ret = dht_request_iatt_in_xdata(this, local->xattr_req);
+        ret = dht_request_iatt_in_xdata(local->xattr_req);
 
         if (ret) {
             gf_msg_debug(this->name, 0,
@@ -1406,7 +1452,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_fxattrop_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                  int32_t op_ret, int32_t op_errno, dict_t *dict, dict_t *xdata)
 {
@@ -1454,7 +1500,7 @@ dht_fxattrop(call_frame_t *frame, xlator_t *this, fd_t *fd,
         local->rebalance.xattr = dict_ref(dict);
         local->rebalance.flags = flags;
 
-        ret = dht_request_iatt_in_xdata(this, local->xattr_req);
+        ret = dht_request_iatt_in_xdata(local->xattr_req);
 
         if (ret) {
             gf_msg_debug(this->name, 0, "Failed to set dictionary key %s fd=%p",
@@ -1479,7 +1525,7 @@ err:
  * below fops, hence not implementing 'migration' related checks
  */
 
-int
+static int
 dht_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                 int32_t op_ret, int32_t op_errno, dict_t *xdata)
 
@@ -1535,8 +1581,26 @@ dht_finodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                  int32_t op_ret, int32_t op_errno, dict_t *xdata)
 
 {
+    dht_local_t *local = NULL;
+    int ret = 0;
+
+    GF_VALIDATE_OR_GOTO("dht", frame, out);
+    GF_VALIDATE_OR_GOTO("dht", this, out);
+    GF_VALIDATE_OR_GOTO("dht", frame->local, out);
+
+    local = frame->local;
+
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
+        ret = dht_check_and_open_fd_on_subvol(this, frame);
+        if (ret)
+            goto out;
+        return 0;
+    }
+
+out:
     dht_lk_inode_unref(frame, op_ret);
     DHT_STACK_UNWIND(finodelk, frame, op_ret, op_errno, xdata);
+
     return 0;
 }
 
@@ -1574,6 +1638,13 @@ dht_finodelk(call_frame_t *frame, xlator_t *this, const char *volume, fd_t *fd,
             if (ret)
                     goto err;
     */
+    local->rebalance.flock = *lock;
+    local->rebalance.lock_cmd = cmd;
+    local->key = gf_strdup(volume);
+
+    if (xdata)
+        local->xattr_req = dict_ref(xdata);
+
     STACK_WIND(frame, dht_finodelk_cbk, lock_subvol,
                lock_subvol->fops->finodelk, volume, fd, cmd, lock, xdata);
 
diff --git a/xlators/cluster/dht/src/dht-inode-write.c b/xlators/cluster/dht/src/dht-inode-write.c
index d0d12fd7658..2f23ce90fbd 100644
--- a/xlators/cluster/dht/src/dht-inode-write.c
+++ b/xlators/cluster/dht/src/dht-inode-write.c
@@ -10,17 +10,17 @@
 
 #include "dht-common.h"
 
-int
+static int
 dht_writev2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
-int
+static int
 dht_truncate2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
-int
+static int
 dht_setattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
-int
+static int
 dht_fallocate2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
-int
+static int
 dht_discard2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
-int
+static int
 dht_zerofill2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret);
 
 int
@@ -49,7 +49,7 @@ dht_writev_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * We only check once as this could be a valid bad fd error.
      */
 
-    if (op_ret == -1 && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -93,30 +93,28 @@ dht_writev_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
 
     /* Check if the rebalance phase1 is true */
     if (IS_DHT_MIGRATION_PHASE1(postbuf)) {
-        if (!dht_is_tier_xlator(this)) {
+        if (!local->xattr_req) {
+            local->xattr_req = dict_new();
             if (!local->xattr_req) {
-                local->xattr_req = dict_new();
-                if (!local->xattr_req) {
-                    gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, ENOMEM,
-                           "insufficient memory");
-                    local->op_errno = ENOMEM;
-                    local->op_ret = -1;
-                    goto out;
-                }
-            }
-
-            ret = dict_set_uint32(local->xattr_req,
-                                  GF_PROTECT_FROM_EXTERNAL_WRITES, 1);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_DICT_SET_FAILED, 0,
-                       "Failed to set key %s in dictionary",
-                       GF_PROTECT_FROM_EXTERNAL_WRITES);
+                gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, ENOMEM,
+                       "insufficient memory");
                 local->op_errno = ENOMEM;
                 local->op_ret = -1;
                 goto out;
             }
         }
 
+        ret = dict_set_uint32(local->xattr_req, GF_PROTECT_FROM_EXTERNAL_WRITES,
+                              1);
+        if (ret) {
+            gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_DICT_SET_FAILED, 0,
+                   "Failed to set key %s in dictionary",
+                   GF_PROTECT_FROM_EXTERNAL_WRITES);
+            local->op_errno = ENOMEM;
+            local->op_ret = -1;
+            goto out;
+        }
+
         dht_iatt_merge(this, &local->stbuf, postbuf);
         dht_iatt_merge(this, &local->prebuf, prebuf);
 
@@ -142,7 +140,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_writev2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -262,8 +260,8 @@ dht_truncate_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * We only check once as this could actually be a valid error.
      */
 
-    if ((local->fop == GF_FOP_FTRUNCATE) && (op_ret == -1) &&
-        ((op_errno == EBADF) || (op_errno == EINVAL)) && !(local->fd_checked)) {
+    if ((local->fop == GF_FOP_FTRUNCATE) &&
+        dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -336,7 +334,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_truncate2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -489,7 +487,7 @@ dht_fallocate_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * We only check once as this could actually be a valid error.
      */
 
-    if ((op_ret == -1) && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -555,7 +553,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_fallocate2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -666,7 +664,7 @@ dht_discard_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * and a lookup updated the cached subvol in the inode ctx.
      * We only check once as this could actually be a valid error.
      */
-    if ((op_ret == -1) && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -731,7 +729,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_discard2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -838,7 +836,7 @@ dht_zerofill_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
      * and a lookup updated the cached subvol in the inode ctx.
      * We only check once as this could actually be a valid error.
      */
-    if ((op_ret == -1) && (op_errno == EBADF) && !(local->fd_checked)) {
+    if (dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -902,7 +900,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_zerofill2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -1005,8 +1003,8 @@ dht_file_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     local->op_errno = op_errno;
 
-    if ((local->fop == GF_FOP_FSETATTR) && (op_ret == -1) &&
-        (op_errno == EBADF) && !(local->fd_checked)) {
+    if ((local->fop == GF_FOP_FSETATTR) &&
+        dht_check_remote_fd_failed_error(local, op_ret, op_errno)) {
         ret = dht_check_and_open_fd_on_subvol(this, frame);
         if (ret)
             goto out;
@@ -1049,7 +1047,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_setattr2(xlator_t *this, xlator_t *subvol, call_frame_t *frame, int ret)
 {
     dht_local_t *local = NULL;
@@ -1113,9 +1111,10 @@ dht_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
     {
         if (op_ret == -1) {
             local->op_errno = op_errno;
+            UNLOCK(&frame->lock);
             gf_msg_debug(this->name, op_errno, "subvolume %s returned -1",
                          prev->name);
-            goto unlock;
+            goto post_unlock;
         }
 
         dht_iatt_merge(this, &local->prebuf, statpre);
@@ -1124,9 +1123,8 @@ dht_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
         local->op_ret = 0;
         local->op_errno = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt)) {
         if (local->op_ret == 0)
@@ -1151,24 +1149,22 @@ dht_non_mds_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     local = frame->local;
     prev = cookie;
 
+    if (op_ret == -1) {
+        gf_msg(this->name, op_errno, 0, 0, "subvolume %s returned -1",
+               prev->name);
+        goto post_unlock;
+    }
+
     LOCK(&frame->lock);
     {
-        if (op_ret == -1) {
-            gf_msg(this->name, op_errno, 0, 0, "subvolume %s returned -1",
-                   prev->name);
-
-            goto unlock;
-        }
-
         dht_iatt_merge(this, &local->prebuf, statpre);
         dht_iatt_merge(this, &local->stbuf, statpost);
 
         local->op_ret = 0;
         local->op_errno = 0;
     }
-unlock:
     UNLOCK(&frame->lock);
-
+post_unlock:
     this_call_cnt = dht_frame_return(frame);
     if (is_last_call(this_call_cnt)) {
         dht_inode_ctx_time_set(local->loc.inode, this, &local->stbuf);
diff --git a/xlators/cluster/dht/src/dht-layout.c b/xlators/cluster/dht/src/dht-layout.c
index 544b9638104..fda904c92c9 100644
--- a/xlators/cluster/dht/src/dht-layout.c
+++ b/xlators/cluster/dht/src/dht-layout.c
@@ -8,11 +8,8 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "glusterfs.h"
-#include "xlator.h"
 #include "dht-common.h"
-#include "byte-order.h"
-#include "dht-messages.h"
+#include <glusterfs/byte-order.h>
 #include "unittest/unittest.h"
 
 #define layout_base_size (sizeof(dht_layout_t))
@@ -134,9 +131,8 @@ dht_layout_search(xlator_t *this, dht_layout_t *layout, const char *name)
 
     ret = dht_hash_compute(this, layout->type, name, &hash);
     if (ret != 0) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_COMPUTE_HASH_FAILED,
-               "hash computation failed for type=%d name=%s", layout->type,
-               name);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_COMPUTE_HASH_FAILED,
+                "type=%d", layout->type, "name=%s", name, NULL);
         goto out;
     }
 
@@ -148,8 +144,8 @@ dht_layout_search(xlator_t *this, dht_layout_t *layout, const char *name)
     }
 
     if (!subvol) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_HASHED_SUBVOL_GET_FAILED,
-               "no subvolume for hash (value) = %u", hash);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_HASHED_SUBVOL_GET_FAILED,
+                "hash-value=0x%x", hash, NULL);
     }
 
 out:
@@ -258,7 +254,7 @@ dht_disk_layout_extract_for_subvol(xlator_t *this, dht_layout_t *layout,
     return dht_disk_layout_extract(this, layout, i, disk_layout_p);
 }
 
-int
+static int
 dht_disk_layout_merge(xlator_t *this, dht_layout_t *layout, int pos,
                       void *disk_layout_raw, int disk_layout_len)
 {
@@ -269,8 +265,8 @@ dht_disk_layout_merge(xlator_t *this, dht_layout_t *layout, int pos,
     int disk_layout[4];
 
     if (!disk_layout_raw) {
-        gf_msg(this->name, GF_LOG_CRITICAL, 0, DHT_MSG_LAYOUT_MERGE_FAILED,
-               "error no layout on disk for merge");
+        gf_smsg(this->name, GF_LOG_CRITICAL, 0, DHT_MSG_LAYOUT_MERGE_FAILED,
+                NULL);
         return -1;
     }
 
@@ -287,10 +283,8 @@ dht_disk_layout_merge(xlator_t *this, dht_layout_t *layout, int pos,
         case DHT_HASH_TYPE_DM:
             break;
         default:
-            gf_msg(this->name, GF_LOG_CRITICAL, 0, DHT_MSG_INVALID_DISK_LAYOUT,
-                   "Invalid disk layout: "
-                   "Catastrophic error layout with unknown type found %d",
-                   disk_layout[1]);
+            gf_smsg(this->name, GF_LOG_CRITICAL, 0, DHT_MSG_INVALID_DISK_LAYOUT,
+                    "layout=%d", disk_layout[1], NULL);
             return -1;
     }
 
@@ -302,9 +296,10 @@ dht_disk_layout_merge(xlator_t *this, dht_layout_t *layout, int pos,
     layout->list[pos].start = start_off;
     layout->list[pos].stop = stop_off;
 
-    gf_msg_trace(
-        this->name, 0, "merged to layout: %u - %u (type %d, hash %d) from %s",
-        start_off, stop_off, commit_hash, type, layout->list[pos].xlator->name);
+    gf_msg_trace(this->name, 0,
+                 "merged to layout: 0x%x - 0x%x (hash 0x%x, type %d) from %s",
+                 start_off, stop_off, commit_hash, type,
+                 layout->list[pos].xlator->name);
 
     return 0;
 }
@@ -357,8 +352,8 @@ dht_layout_merge(xlator_t *this, dht_layout_t *layout, xlator_t *subvol,
     ret = dht_disk_layout_merge(this, layout, i, disk_layout_raw,
                                 disk_layout_len);
     if (ret != 0) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_MERGE_FAILED,
-               "layout merge from subvolume %s failed", subvol->name);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_MERGE_FAILED,
+                "subvolume=%s", subvol->name, NULL);
         goto out;
     }
 
@@ -417,8 +412,7 @@ dht_layout_range_swap(dht_layout_t *layout, int i, int j)
     layout->list[j].start = start_swap;
     layout->list[j].stop = stop_swap;
 }
-
-int64_t
+static int64_t
 dht_layout_entry_cmp_volname(dht_layout_t *layout, int i, int j)
 {
     return (strcmp(layout->list[i].xlator->name, layout->list[j].xlator->name));
@@ -441,7 +435,7 @@ dht_is_subvol_in_layout(dht_layout_t *layout, xlator_t *xlator)
     return _gf_false;
 }
 
-int64_t
+static int64_t
 dht_layout_entry_cmp(dht_layout_t *layout, int i, int j)
 {
     int64_t diff = 0;
@@ -477,7 +471,7 @@ dht_layout_sort(dht_layout_t *layout)
     return 0;
 }
 
-int
+void
 dht_layout_sort_volname(dht_layout_t *layout)
 {
     int i = 0;
@@ -493,8 +487,6 @@ dht_layout_sort_volname(dht_layout_t *layout)
                 dht_layout_entry_swap(layout, i, j);
         }
     }
-
-    return 0;
 }
 
 void
@@ -627,8 +619,8 @@ dht_layout_normalize(xlator_t *this, loc_t *loc, dht_layout_t *layout)
 
     ret = dht_layout_sort(layout);
     if (ret == -1) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_SORT_FAILED,
-               "sort failed?! how the ....");
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_SORT_FAILED,
+                NULL);
         goto out;
     }
 
@@ -644,10 +636,9 @@ dht_layout_normalize(xlator_t *this, loc_t *loc, dht_layout_t *layout)
                          " gfid = %s",
                          loc->path, gfid);
         } else {
-            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_ANOMALIES_INFO,
-                   "Found anomalies in %s (gfid = %s). "
-                   "Holes=%d overlaps=%d",
-                   loc->path, gfid, holes, overlaps);
+            gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_ANOMALIES_INFO,
+                    "path=%s", loc->path, "gfid=%s", gfid, "holes=%d", holes,
+                    "overlaps=%d", overlaps, NULL);
         }
         ret = -1;
     }
@@ -714,12 +705,11 @@ dht_layout_dir_mismatch(xlator_t *this, dht_layout_t *layout, xlator_t *subvol,
     if (!xattr) {
         if (err == 0) {
             if (loc) {
-                gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_DICT_GET_FAILED,
-                       "%s: xattr dictionary is NULL", loc->path);
+                gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_XATTR_DICT_NULL,
+                        "path=%s", loc->path, NULL);
             } else {
-                gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_DICT_GET_FAILED,
-                       "path not found: "
-                       "xattr dictionary is NULL");
+                gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_XATTR_DICT_NULL,
+                        "path not found", NULL);
             }
             ret = -1;
         }
@@ -731,13 +721,13 @@ dht_layout_dir_mismatch(xlator_t *this, dht_layout_t *layout, xlator_t *subvol,
     if (dict_ret < 0) {
         if (err == 0 && layout->list[pos].stop) {
             if (loc) {
-                gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_DISK_LAYOUT_MISSING,
-                       "%s: Disk layout missing, gfid = %s", loc->path, gfid);
+                gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_DISK_LAYOUT_MISSING,
+                        "path=%s", loc->path, "gfid=%s", gfid, NULL);
             } else {
-                gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_DISK_LAYOUT_MISSING,
-                       "path not found: "
-                       "Disk layout missing, gfid = %s",
-                       gfid);
+                gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_DISK_LAYOUT_MISSING,
+                        "path not found"
+                        "gfid=%s",
+                        gfid, NULL);
             }
             ret = -1;
         }
@@ -753,13 +743,13 @@ dht_layout_dir_mismatch(xlator_t *this, dht_layout_t *layout, xlator_t *subvol,
     if ((layout->list[pos].start != start_off) ||
         (layout->list[pos].stop != stop_off) ||
         (layout->list[pos].commit_hash != commit_hash)) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LAYOUT_INFO,
-               "subvol: %s; inode layout - %" PRIu32 " - %" PRIu32 " - %" PRIu32
-               "; "
-               "disk layout - %" PRIu32 " - %" PRIu32 " - %" PRIu32,
-               layout->list[pos].xlator->name, layout->list[pos].start,
-               layout->list[pos].stop, layout->list[pos].commit_hash, start_off,
-               stop_off, commit_hash);
+        gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_LAYOUT_INFO, "subvol=%s",
+                layout->list[pos].xlator->name, "inode-layout:start=0x%x",
+                layout->list[pos].start, "inode-layout:stop=0x%x",
+                layout->list[pos].stop, "layout-commit-hash=0x%x; ",
+                layout->list[pos].commit_hash, "disk-layout:start-off=0x%x",
+                start_off, "disk-layout:top-off=0x%x", stop_off,
+                "commit-hash=0x%x", commit_hash, NULL);
         ret = 1;
     } else {
         ret = 0;
@@ -781,9 +771,8 @@ dht_layout_preset(xlator_t *this, xlator_t *subvol, inode_t *inode)
 
     layout = dht_layout_for_subvol(this, subvol);
     if (!layout) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_SUBVOL_NO_LAYOUT_INFO,
-               "no pre-set layout for subvolume %s",
-               subvol ? subvol->name : "<nil>");
+        gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_SUBVOL_NO_LAYOUT_INFO,
+                "subvolume=%s", subvol ? subvol->name : "<nil>", NULL);
         ret = -1;
         goto out;
     }
diff --git a/xlators/cluster/dht/src/dht-linkfile.c b/xlators/cluster/dht/src/dht-linkfile.c
index eb1695f7e05..89ec6cca56e 100644
--- a/xlators/cluster/dht/src/dht-linkfile.c
+++ b/xlators/cluster/dht/src/dht-linkfile.c
@@ -8,13 +8,10 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "glusterfs.h"
-#include "xlator.h"
-#include "compat.h"
+#include <glusterfs/compat.h>
 #include "dht-common.h"
-#include "dht-messages.h"
 
-int
+static int
 dht_linkfile_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                         int op_ret, int op_errno, inode_t *inode,
                         struct iatt *stbuf, dict_t *xattr,
@@ -37,17 +34,16 @@ dht_linkfile_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     is_linkfile = check_is_linkfile(inode, stbuf, xattr, conf->link_xattr_name);
     if (!is_linkfile)
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NOT_LINK_FILE_ERROR,
-               "got non-linkfile %s:%s, gfid = %s", prev->name, local->loc.path,
-               gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NOT_LINK_FILE_ERROR,
+                "name=%s", prev->name, "path=%s", local->loc.path, "gfid=%s",
+                gfid, NULL);
 out:
     local->linkfile.linkfile_cbk(frame, cookie, this, op_ret, op_errno, inode,
                                  stbuf, postparent, postparent, xattr);
     return 0;
 }
 
-#define is_equal(a, b) ((a) == (b))
-int
+static int
 dht_linkfile_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                         int op_ret, int op_errno, inode_t *inode,
                         struct iatt *stbuf, struct iatt *preparent,
@@ -76,9 +72,8 @@ dht_linkfile_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
             goto out;
         ret = dict_set_uint32(xattrs, conf->link_xattr_name, 256);
         if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-                   "Failed to set dictionary value. key : %s",
-                   conf->link_xattr_name);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+                    "mame=%s", conf->link_xattr_name, NULL);
             goto out;
         }
 
@@ -128,27 +123,23 @@ dht_linkfile_create(call_frame_t *frame, fop_mknod_cbk_t linkfile_cbk,
 
         ret = dict_set_gfuuid(dict, "gfid-req", local->gfid, true);
         if (ret)
-            gf_msg("dht-linkfile", GF_LOG_INFO, 0, DHT_MSG_DICT_SET_FAILED,
-                   "%s: Failed to set dictionary value: "
-                   "key = gfid-req, gfid = %s ",
-                   loc->path, gfid);
+            gf_smsg("dht-linkfile", GF_LOG_INFO, 0, DHT_MSG_DICT_SET_FAILED,
+                    "path=%s", loc->path, "gfid=%s", gfid, NULL);
     } else {
         gf_uuid_unparse(loc->gfid, gfid);
     }
 
     ret = dict_set_str(dict, GLUSTERFS_INTERNAL_FOP_KEY, "yes");
     if (ret)
-        gf_msg("dht-linkfile", GF_LOG_INFO, 0, DHT_MSG_DICT_SET_FAILED,
-               "%s: Failed to set dictionary value: key = %s,"
-               " gfid = %s",
-               loc->path, GLUSTERFS_INTERNAL_FOP_KEY, gfid);
+        gf_smsg("dht-linkfile", GF_LOG_INFO, 0, DHT_MSG_DICT_SET_FAILED,
+                "path=%s", loc->path, "key=%s", GLUSTERFS_INTERNAL_FOP_KEY,
+                "gfid=%s", gfid, NULL);
 
     ret = dict_set_str(dict, conf->link_xattr_name, tovol->name);
 
     if (ret < 0) {
-        gf_msg(frame->this->name, GF_LOG_INFO, 0, DHT_MSG_CREATE_LINK_FAILED,
-               "%s: failed to initialize linkfile data, gfid = %s", loc->path,
-               gfid);
+        gf_smsg(frame->this->name, GF_LOG_INFO, 0, DHT_MSG_CREATE_LINK_FAILED,
+                "path=%s", loc->path, "gfid=%s", gfid, NULL);
         goto out;
     }
 
@@ -189,10 +180,9 @@ dht_linkfile_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     if (op_ret == -1) {
         gf_uuid_unparse(local->loc.gfid, gfid);
-        gf_msg(this->name, GF_LOG_INFO, op_errno, DHT_MSG_UNLINK_FAILED,
-               "Unlinking linkfile %s (gfid = %s)on "
-               "subvolume %s failed ",
-               local->loc.path, gfid, subvol->name);
+        gf_smsg(this->name, GF_LOG_INFO, op_errno, DHT_MSG_UNLINK_FAILED,
+                "path=%s", local->loc.path, "gfid=%s", gfid, "subvolume=%s",
+                subvol->name, NULL);
     }
 
     DHT_STACK_DESTROY(frame);
@@ -260,7 +250,7 @@ out:
     return subvol;
 }
 
-int
+static int
 dht_linkfile_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                          int op_ret, int op_errno, struct iatt *statpre,
                          struct iatt *statpost, dict_t *xdata)
@@ -272,10 +262,9 @@ dht_linkfile_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     loc = &local->loc;
 
     if (op_ret)
-        gf_msg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_SETATTR_FAILED,
-               "Failed to set attr uid/gid on %s"
-               " :<gfid:%s> ",
-               (loc->path ? loc->path : "NULL"), uuid_utoa(local->gfid));
+        gf_smsg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_SETATTR_FAILED,
+                "path=%s", (loc->path ? loc->path : "NULL"), "gfid=%s",
+                uuid_utoa(local->gfid), NULL);
 
     DHT_STACK_DESTROY(frame);
 
diff --git a/xlators/cluster/dht/src/dht-lock.c b/xlators/cluster/dht/src/dht-lock.c
index f9bac4f97c8..638821ccee5 100644
--- a/xlators/cluster/dht/src/dht-lock.c
+++ b/xlators/cluster/dht/src/dht-lock.c
@@ -44,7 +44,8 @@ dht_log_lk_array(char *name, gf_loglevel_t log_level, dht_lock_t **lk_array,
         if (!lk_buf)
             goto out;
 
-        gf_msg(name, log_level, 0, DHT_MSG_LK_ARRAY_INFO, "%d. %s", i, lk_buf);
+        gf_smsg(name, log_level, 0, DHT_MSG_LK_ARRAY_INFO, "index=%d", i,
+                "lk_buf=%s", lk_buf, NULL);
         GF_FREE(lk_buf);
     }
 
@@ -313,11 +314,9 @@ dht_unlock_entrylk_done(call_frame_t *frame, void *cookie, xlator_t *this,
                     gfid);
 
     if (op_ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, op_errno,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "unlock failed on gfid: %s, stale lock might be left "
-               "in DHT_LAYOUT_HEAL_DOMAIN",
-               gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, op_errno,
+                DHT_MSG_UNLOCK_GFID_FAILED, "gfid=%s", gfid,
+                "DHT_LAYOUT_HEAL_DOMAIN", NULL);
     }
 
     DHT_STACK_DESTROY(frame);
@@ -339,9 +338,10 @@ dht_unlock_entrylk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     uuid_utoa_r(local->lock[0].ns.directory_ns.locks[lk_index]->loc.gfid, gfid);
 
     if (op_ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_UNLOCKING_FAILED,
-               "unlocking failed on %s:%s",
-               local->lock[0].ns.directory_ns.locks[lk_index]->xl->name, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_UNLOCKING_FAILED,
+                "name=%s",
+                local->lock[0].ns.directory_ns.locks[lk_index]->xl->name,
+                "gfid=%s", gfid, NULL);
     } else {
         local->lock[0].ns.directory_ns.locks[lk_index]->locked = 0;
     }
@@ -375,9 +375,9 @@ dht_unlock_entrylk(call_frame_t *frame, dht_lock_t **lk_array, int lk_count,
 
     lock_frame = dht_lock_frame(frame);
     if (lock_frame == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_UNLOCKING_FAILED,
-               "cannot allocate a frame, not unlocking following "
-               "entrylks:");
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_ALLOC_FRAME_FAILED_NOT_UNLOCKING_FOLLOWING_ENTRYLKS,
+                NULL);
 
         dht_log_lk_array(frame->this->name, GF_LOG_WARNING, lk_array, lk_count);
         goto done;
@@ -385,9 +385,9 @@ dht_unlock_entrylk(call_frame_t *frame, dht_lock_t **lk_array, int lk_count,
 
     ret = dht_local_entrylk_init(lock_frame, lk_array, lk_count, entrylk_cbk);
     if (ret < 0) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_UNLOCKING_FAILED,
-               "storing locks in local failed, not unlocking "
-               "following entrylks:");
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_LOCAL_LOCKS_STORE_FAILED_UNLOCKING_FOLLOWING_ENTRYLK,
+                NULL);
 
         dht_log_lk_array(frame->this->name, GF_LOG_WARNING, lk_array, lk_count);
 
@@ -446,21 +446,17 @@ dht_unlock_entrylk_wrapper(call_frame_t *frame, dht_elock_wrap_t *entrylk)
 
     lock_frame = copy_frame(frame);
     if (lock_frame == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, ENOMEM,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "mkdir (%s/%s) (path: %s): "
-               "copy frame failed",
-               pgfid, local->loc.name, local->loc.path);
+        gf_smsg(frame->this->name, GF_LOG_WARNING, ENOMEM,
+                DHT_MSG_COPY_FRAME_FAILED, "pgfid=%s", pgfid, "name=%s",
+                local->loc.name, "path=%s", local->loc.path, NULL);
         goto done;
     }
 
     lock_local = dht_local_init(lock_frame, NULL, NULL, 0);
     if (lock_local == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, ENOMEM,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "mkdir (%s/%s) (path: %s): "
-               "local creation failed",
-               pgfid, local->loc.name, local->loc.path);
+        gf_smsg(frame->this->name, GF_LOG_WARNING, ENOMEM,
+                DHT_MSG_CREATE_FAILED, "local", "pgfid=%s", pgfid, "name=%s",
+                local->loc.name, "path=%s", local->loc.path, NULL);
         goto done;
     }
 
@@ -700,9 +696,10 @@ dht_unlock_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         uuid_utoa_r(local->lock[0].layout.my_layout.locks[lk_index]->loc.gfid,
                     gfid);
 
-        gf_msg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_UNLOCKING_FAILED,
-               "unlocking failed on %s:%s",
-               local->lock[0].layout.my_layout.locks[lk_index]->xl->name, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_UNLOCKING_FAILED,
+                "name=%s",
+                local->lock[0].layout.my_layout.locks[lk_index]->xl->name,
+                "gfid=%s", gfid, NULL);
     } else {
         local->lock[0].layout.my_layout.locks[lk_index]->locked = 0;
     }
@@ -727,11 +724,9 @@ dht_unlock_inodelk_done(call_frame_t *frame, void *cookie, xlator_t *this,
                     gfid);
 
     if (op_ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, op_errno,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "unlock failed on gfid: %s, stale lock might be left "
-               "in DHT_LAYOUT_HEAL_DOMAIN",
-               gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, op_errno,
+                DHT_MSG_UNLOCK_GFID_FAILED, "DHT_LAYOUT_HEAL_DOMAIN gfid=%s",
+                gfid, NULL);
     }
 
     DHT_STACK_DESTROY(frame);
@@ -762,9 +757,9 @@ dht_unlock_inodelk(call_frame_t *frame, dht_lock_t **lk_array, int lk_count,
 
     lock_frame = dht_lock_frame(frame);
     if (lock_frame == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_UNLOCKING_FAILED,
-               "cannot allocate a frame, not unlocking following "
-               "locks:");
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_ALLOC_FRAME_FAILED_NOT_UNLOCKING_FOLLOWING_ENTRYLKS,
+                NULL);
 
         dht_log_lk_array(frame->this->name, GF_LOG_WARNING, lk_array, lk_count);
         goto done;
@@ -772,9 +767,9 @@ dht_unlock_inodelk(call_frame_t *frame, dht_lock_t **lk_array, int lk_count,
 
     ret = dht_local_inodelk_init(lock_frame, lk_array, lk_count, inodelk_cbk);
     if (ret < 0) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_UNLOCKING_FAILED,
-               "storing locks in local failed, not unlocking "
-               "following locks:");
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_LOCAL_LOCKS_STORE_FAILED_UNLOCKING_FOLLOWING_ENTRYLK,
+                NULL);
 
         dht_log_lk_array(frame->this->name, GF_LOG_WARNING, lk_array, lk_count);
 
@@ -834,21 +829,17 @@ dht_unlock_inodelk_wrapper(call_frame_t *frame, dht_ilock_wrap_t *inodelk)
 
     lock_frame = copy_frame(frame);
     if (lock_frame == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, ENOMEM,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "mkdir (%s/%s) (path: %s): "
-               "copy frame failed",
-               pgfid, local->loc.name, local->loc.path);
+        gf_smsg(frame->this->name, GF_LOG_WARNING, ENOMEM,
+                DHT_MSG_COPY_FRAME_FAILED, "pgfid=%s", pgfid, "name=%s",
+                local->loc.name, "path=%s", local->loc.path, NULL);
         goto done;
     }
 
     lock_local = dht_local_init(lock_frame, NULL, NULL, 0);
     if (lock_local == NULL) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, ENOMEM,
-               DHT_MSG_PARENT_LAYOUT_CHANGED,
-               "mkdir (%s/%s) (path: %s): "
-               "local creation failed",
-               pgfid, local->loc.name, local->loc.path);
+        gf_smsg(frame->this->name, GF_LOG_WARNING, ENOMEM,
+                DHT_MSG_CREATE_FAILED, "local", "gfid=%s", pgfid, "name=%s",
+                local->loc.name, "path=%s", local->loc.path, NULL);
         goto done;
     }
 
@@ -1039,13 +1030,12 @@ dht_blocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                                     gfid);
                     local->lock[0].layout.my_layout.op_ret = -1;
                     local->lock[0].layout.my_layout.op_errno = op_errno;
-                    gf_msg(this->name, GF_LOG_ERROR, op_errno,
-                           DHT_MSG_INODELK_FAILED,
-                           "inodelk failed on subvol %s. gfid:%s",
-                           local->lock[0]
-                               .layout.my_layout.locks[lk_index]
-                               ->xl->name,
-                           gfid);
+                    gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                            DHT_MSG_INODELK_FAILED, "subvol=%s",
+                            local->lock[0]
+                                .layout.my_layout.locks[lk_index]
+                                ->xl->name,
+                            "gfid=%s", gfid, NULL);
                     goto cleanup;
                 }
                 break;
@@ -1060,13 +1050,12 @@ dht_blocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                                     gfid);
                     local->lock[0].layout.my_layout.op_ret = -1;
                     local->lock[0].layout.my_layout.op_errno = op_errno;
-                    gf_msg(this->name, GF_LOG_ERROR, op_errno,
-                           DHT_MSG_INODELK_FAILED,
-                           "inodelk failed on subvol %s. gfid:%s",
-                           local->lock[0]
-                               .layout.my_layout.locks[lk_index]
-                               ->xl->name,
-                           gfid);
+                    gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                            DHT_MSG_INODELK_FAILED, "subvol=%s",
+                            local->lock[0]
+                                .layout.my_layout.locks[lk_index]
+                                ->xl->name,
+                            "gfid=%s", gfid, NULL);
                     goto cleanup;
                 }
                 break;
@@ -1077,11 +1066,11 @@ dht_blocking_inodelk_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                     gfid);
                 local->lock[0].layout.my_layout.op_ret = -1;
                 local->lock[0].layout.my_layout.op_errno = op_errno;
-                gf_msg(
+                gf_smsg(
                     this->name, GF_LOG_ERROR, op_errno, DHT_MSG_INODELK_FAILED,
-                    "inodelk failed on subvol %s, gfid:%s",
+                    "subvol=%s",
                     local->lock[0].layout.my_layout.locks[lk_index]->xl->name,
-                    gfid);
+                    "gfid=%s", gfid, NULL);
                 goto cleanup;
         }
     }
@@ -1153,19 +1142,16 @@ dht_blocking_inodelk(call_frame_t *frame, dht_lock_t **lk_array, int lk_count,
     lock_frame = dht_lock_frame(frame);
     if (lock_frame == NULL) {
         gf_uuid_unparse(tmp_local->loc.gfid, gfid);
-        gf_msg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_LOCK_FRAME_FAILED,
-               "memory allocation failed for lock_frame. gfid:%s"
-               " path:%s",
-               gfid, tmp_local->loc.path);
+        gf_smsg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_LOCK_FRAME_FAILED,
+                "gfid=%s", gfid, "path=%s", tmp_local->loc.path, NULL);
         goto out;
     }
 
     ret = dht_local_inodelk_init(lock_frame, lk_array, lk_count, inodelk_cbk);
     if (ret < 0) {
         gf_uuid_unparse(tmp_local->loc.gfid, gfid);
-        gf_msg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_LOCAL_LOCK_INIT_FAILED,
-               "dht_local_lock_init failed, gfid: %s path:%s", gfid,
-               tmp_local->loc.path);
+        gf_smsg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_LOCAL_LOCK_INIT_FAILED,
+                "gfid=%s", gfid, "path=%s", tmp_local->loc.path, NULL);
         goto out;
     }
 
@@ -1246,11 +1232,10 @@ dht_blocking_entrylk_after_inodelk(call_frame_t *frame, void *cookie,
     if (ret < 0) {
         local->op_ret = -1;
         local->op_errno = EIO;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
-               DHT_MSG_ENTRYLK_ERROR,
-               "%s (%s/%s): "
-               "dht_blocking_entrylk failed after taking inodelk",
-               gf_fop_list[local->fop], pgfid, entrylk->locks[0]->basename);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_ENTRYLK_FAILED_AFT_INODELK, "fop=%s",
+                gf_fop_list[local->fop], "pgfid=%s", pgfid, "basename=%s",
+                entrylk->locks[0]->basename, NULL);
         goto err;
     }
 
@@ -1310,10 +1295,9 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
 
     ret = dht_build_parent_loc(this, &parent, loc, &op_errno);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_LOC_FAILED,
-               "gfid:%s (name:%s) (path: %s): "
-               "parent loc build failed",
-               loc->gfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_ERROR, op_errno, DHT_MSG_LOC_FAILED,
+                "gfid=%s", loc->gfid, "name=%s", loc->name, "path=%s",
+                loc->path, NULL);
         goto out;
     }
     gf_uuid_unparse(parent.gfid, pgfid);
@@ -1322,10 +1306,10 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
     inodelk->locks = GF_CALLOC(count, sizeof(*lk_array), gf_common_mt_pointer);
     if (inodelk->locks == NULL) {
         local->op_errno = ENOMEM;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno, DHT_MSG_NO_MEMORY,
-               "%s (%s/%s) (path: %s): "
-               "calloc failure",
-               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_CALLOC_FAILED, "fop=%s", gf_fop_list[local->fop],
+                "pgfid=%s", pgfid, "name=%s", loc->name, "path=%s", loc->path,
+                NULL);
         goto out;
     }
 
@@ -1334,10 +1318,10 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
                                      FAIL_ON_ANY_ERROR);
     if (inodelk->locks[0] == NULL) {
         local->op_errno = ENOMEM;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno, DHT_MSG_NO_MEMORY,
-               "%s (%s/%s) (path: %s): "
-               "inodelk: lock allocation failed",
-               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_LOCK_ALLOC_FAILED, "inodelk-fop=%s",
+                gf_fop_list[local->fop], "pgfid=%s", pgfid, "name=%s",
+                loc->name, "path=%s", loc->path, NULL);
         goto err;
     }
     inodelk->lk_count = count;
@@ -1346,10 +1330,10 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
     entrylk->locks = GF_CALLOC(count, sizeof(*lk_array), gf_common_mt_pointer);
     if (entrylk->locks == NULL) {
         local->op_errno = ENOMEM;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno, DHT_MSG_NO_MEMORY,
-               "%s (%s/%s) (path: %s): "
-               "entrylk: calloc failure",
-               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_CALLOC_FAILED, "entrylk-fop=%s",
+                gf_fop_list[local->fop], "pgfid=%s", pgfid, "name=%s",
+                loc->name, "path=%s", loc->path, NULL);
 
         goto err;
     }
@@ -1359,10 +1343,10 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
                                      FAIL_ON_ANY_ERROR);
     if (entrylk->locks[0] == NULL) {
         local->op_errno = ENOMEM;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno, DHT_MSG_NO_MEMORY,
-               "%s (%s/%s) (path: %s): "
-               "entrylk: lock allocation failed",
-               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_LOCK_ALLOC_FAILED, "entrylk-fop=%s",
+                gf_fop_list[local->fop], "pgfid=%s", pgfid, "name=%s",
+                loc->name, "path=%s", loc->path, NULL);
 
         goto err;
     }
@@ -1376,11 +1360,11 @@ dht_protect_namespace(call_frame_t *frame, loc_t *loc, xlator_t *subvol,
                                dht_blocking_entrylk_after_inodelk);
     if (ret < 0) {
         local->op_errno = EIO;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
-               DHT_MSG_INODELK_ERROR,
-               "%s (%s/%s) (path: %s): "
-               "dht_blocking_inodelk failed",
-               gf_fop_list[local->fop], pgfid, loc->name, loc->path);
+        gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                DHT_MSG_BLOCK_INODELK_FAILED, "fop=%s", gf_fop_list[local->fop],
+                "pgfid=%s", pgfid, "name=%s", loc->name, "path=%s", loc->path,
+                NULL);
+
         goto err;
     }
 
diff --git a/xlators/cluster/dht/src/dht-lock.h b/xlators/cluster/dht/src/dht-lock.h
index 802970adb3b..6485c03fb6e 100644
--- a/xlators/cluster/dht/src/dht-lock.h
+++ b/xlators/cluster/dht/src/dht-lock.h
@@ -11,7 +11,6 @@
 #ifndef _DHT_LOCK_H
 #define _DHT_LOCK_H
 
-#include "xlator.h"
 #include "dht-common.h"
 
 void
diff --git a/xlators/cluster/dht/src/dht-mem-types.h b/xlators/cluster/dht/src/dht-mem-types.h
index c4430741ff7..e3c4471334a 100644
--- a/xlators/cluster/dht/src/dht-mem-types.h
+++ b/xlators/cluster/dht/src/dht-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __DHT_MEM_TYPES_H__
 #define __DHT_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_dht_mem_types_ {
     gf_dht_mt_dht_du_t = gf_common_mt_end + 1,
@@ -30,10 +30,7 @@ enum gf_dht_mem_types_ {
     gf_dht_mt_container_t,
     gf_dht_mt_octx_t,
     gf_dht_mt_miginfo_t,
-    gf_tier_mt_bricklist_t,
-    gf_tier_mt_ipc_ctr_params_t,
     gf_dht_mt_fd_ctx_t,
-    gf_tier_mt_qfile_array_t,
     gf_dht_ret_cache_t,
     gf_dht_nodeuuids_t,
     gf_dht_mt_end
diff --git a/xlators/cluster/dht/src/dht-messages.h b/xlators/cluster/dht/src/dht-messages.h
index 005ab57b505..601f8dad78b 100644
--- a/xlators/cluster/dht/src/dht-messages.h
+++ b/xlators/cluster/dht/src/dht-messages.h
@@ -10,7 +10,7 @@
 #ifndef _DHT_MESSAGES_H_
 #define _DHT_MESSAGES_H_
 
-#include "glfs-message-id.h"
+#include <glusterfs/glfs-message-id.h>
 
 /* To add new message IDs, append new identifiers at the end of the list.
  *
@@ -38,12 +38,11 @@ GLFS_MSGID(
     DHT_MSG_REBALANCE_STATUS, DHT_MSG_REBALANCE_STOPPED, DHT_MSG_RENAME_FAILED,
     DHT_MSG_SETATTR_FAILED, DHT_MSG_SUBVOL_INSUFF_INODES,
     DHT_MSG_SUBVOL_INSUFF_SPACE, DHT_MSG_UNLINK_FAILED,
-    DHT_MSG_LAYOUT_SET_FAILED, DHT_MSG_LOG_FIXED_LAYOUT, DHT_MSG_LOG_TIER_ERROR,
-    DHT_MSG_LOG_TIER_STATUS, DHT_MSG_GET_XATTR_FAILED,
-    DHT_MSG_FILE_LOOKUP_FAILED, DHT_MSG_OPEN_FD_FAILED,
-    DHT_MSG_SET_INODE_CTX_FAILED, DHT_MSG_UNLOCKING_FAILED,
-    DHT_MSG_DISK_LAYOUT_NULL, DHT_MSG_SUBVOL_INFO, DHT_MSG_CHUNK_SIZE_INFO,
-    DHT_MSG_LAYOUT_FORM_FAILED, DHT_MSG_SUBVOL_ERROR,
+    DHT_MSG_LAYOUT_SET_FAILED, DHT_MSG_LOG_FIXED_LAYOUT,
+    DHT_MSG_GET_XATTR_FAILED, DHT_MSG_FILE_LOOKUP_FAILED,
+    DHT_MSG_OPEN_FD_FAILED, DHT_MSG_SET_INODE_CTX_FAILED,
+    DHT_MSG_UNLOCKING_FAILED, DHT_MSG_DISK_LAYOUT_NULL, DHT_MSG_SUBVOL_INFO,
+    DHT_MSG_CHUNK_SIZE_INFO, DHT_MSG_LAYOUT_FORM_FAILED, DHT_MSG_SUBVOL_ERROR,
     DHT_MSG_LAYOUT_SORT_FAILED, DHT_MSG_REGEX_INFO, DHT_MSG_FOPEN_FAILED,
     DHT_MSG_SET_HOSTNAME_FAILED, DHT_MSG_BRICK_ERROR, DHT_MSG_SYNCOP_FAILED,
     DHT_MSG_MIGRATE_INFO, DHT_MSG_SOCKET_ERROR, DHT_MSG_CREATE_FD_FAILED,
@@ -69,8 +68,7 @@ GLFS_MSGID(
     DHT_MSG_INIT_LOCAL_SUBVOL_FAILED, DHT_MSG_SYS_CALL_GET_TIME_FAILED,
     DHT_MSG_NO_DISK_USAGE_STATUS, DHT_MSG_SUBVOL_DOWN_ERROR,
     DHT_MSG_REBAL_THROTTLE_INFO, DHT_MSG_COMMIT_HASH_INFO,
-    DHT_MSG_REBAL_STRUCT_SET, DHT_MSG_HAS_MIGINFO, DHT_MSG_LOG_IPC_TIER_ERROR,
-    DHT_MSG_TIER_PAUSED, DHT_MSG_TIER_RESUME, DHT_MSG_SETTLE_HASH_FAILED,
+    DHT_MSG_REBAL_STRUCT_SET, DHT_MSG_HAS_MIGINFO, DHT_MSG_SETTLE_HASH_FAILED,
     DHT_MSG_DEFRAG_PROCESS_DIR_FAILED, DHT_MSG_FD_CTX_SET_FAILED,
     DHT_MSG_STALE_LOOKUP, DHT_MSG_PARENT_LAYOUT_CHANGED,
     DHT_MSG_LOCK_MIGRATION_FAILED, DHT_MSG_LOCK_INODE_UNREF_FAILED,
@@ -79,6 +77,310 @@ GLFS_MSGID(
     DHT_MSG_ENTRYLK_ERROR, DHT_MSG_INODELK_ERROR, DHT_MSG_LOC_FAILED,
     DHT_MSG_UNKNOWN_FOP, DHT_MSG_MIGRATE_FILE_SKIPPED,
     DHT_MSG_DIR_XATTR_HEAL_FAILED, DHT_MSG_HASHED_SUBVOL_DOWN,
-    DHT_MSG_NON_HASHED_SUBVOL_DOWN);
+    DHT_MSG_NON_HASHED_SUBVOL_DOWN, DHT_MSG_SYNCTASK_CREATE_FAILED,
+    DHT_MSG_DIR_HEAL_ABORT, DHT_MSG_MIGRATE_SKIP, DHT_MSG_FD_CREATE_FAILED,
+    DHT_MSG_DICT_NEW_FAILED, DHT_MSG_FAILED_TO_OPEN, DHT_MSG_CREATE_FAILED,
+    DHT_MSG_FILE_NOT_EXIST, DHT_MSG_CHOWN_FAILED, DHT_MSG_FALLOCATE_FAILED,
+    DHT_MSG_FTRUNCATE_FAILED, DHT_MSG_STATFS_FAILED, DHT_MSG_WRITE_CROSS,
+    DHT_MSG_NEW_TARGET_FOUND, DHT_MSG_INSUFF_MEMORY, DHT_MSG_SET_XATTR_FAILED,
+    DHT_MSG_SET_MODE_FAILED, DHT_MSG_FILE_EXISTS_IN_DEST,
+    DHT_MSG_SYMLINK_FAILED, DHT_MSG_LINKFILE_DEL_FAILED, DHT_MSG_MKNOD_FAILED,
+    DHT_MSG_MIGRATE_CLEANUP_FAILED, DHT_MSG_LOCK_MIGRATE,
+    DHT_MSG_PARENT_BUILD_FAILED, DHT_MSG_HASHED_SUBVOL_NOT_FOUND,
+    DHT_MSG_ACQUIRE_ENTRYLK_FAILED, DHT_MSG_CREATE_DST_FAILED,
+    DHT_MSG_MIGRATION_EXIT, DHT_MSG_CHANGED_DST, DHT_MSG_TRACE_FAILED,
+    DHT_MSG_WRITE_LOCK_FAILED, DHT_MSG_GETACTIVELK_FAILED, DHT_MSG_STAT_FAILED,
+    DHT_MSG_UNLINK_PERFORM_FAILED, DHT_MSG_CLANUP_SOURCE_FILE_FAILED,
+    DHT_MSG_UNLOCK_FILE_FAILED, DHT_MSG_REMOVE_XATTR_FAILED,
+    DHT_MSG_DATA_MIGRATE_ABORT, DHT_MSG_DEFRAG_NULL, DHT_MSG_PARENT_NULL,
+    DHT_MSG_GFID_NOT_PRESENT, DHT_MSG_CHILD_LOC_FAILED,
+    DHT_MSG_SET_LOOKUP_FAILED, DHT_MSG_DIR_REMOVED, DHT_MSG_FIX_NOT_COMP,
+    DHT_MSG_SUBVOL_DETER_FAILED, DHT_MSG_LOCAL_SUBVOL, DHT_MSG_NODE_UUID,
+    DHT_MSG_SIZE_FILE, DHT_MSG_GET_DATA_SIZE_FAILED,
+    DHT_MSG_PTHREAD_JOIN_FAILED, DHT_MSG_COUNTER_THREAD_CREATE_FAILED,
+    DHT_MSG_MIGRATION_INIT_QUEUE_FAILED, DHT_MSG_PAUSED_TIMEOUT, DHT_MSG_WOKE,
+    DHT_MSG_ABORT_REBALANCE, DHT_MSG_CREATE_TASK_REBAL_FAILED,
+    DHT_MSG_REBAL_ESTIMATE_NOT_AVAIL, DHT_MSG_ADD_CHOICES_ERROR,
+    DHT_MSG_GET_CHOICES_ERROR, DHT_MSG_PREPARE_STATUS_ERROR,
+    DHT_MSG_SET_CHOICE_FAILED, DHT_MSG_SET_HASHED_SUBVOL_FAILED,
+    DHT_MSG_XATTR_HEAL_NOT_POSS, DHT_MSG_LINKTO_FILE_FAILED,
+    DHT_MSG_STALE_LINKFILE_DELETE, DHT_MSG_NO_SUBVOL_FOR_LINKTO,
+    DHT_MSG_SUBVOL_RETURNED, DHT_MSG_UNKNOWN_LOCAL_XSEL, DHT_MSG_GET_XATTR_ERR,
+    DHT_MSG_ALLOC_OR_FILL_FAILED, DHT_MSG_GET_REAL_NAME_FAILED,
+    DHT_MSG_COPY_UUID_FAILED, DHT_MSG_MDS_DETER_FAILED,
+    DHT_MSG_CREATE_REBAL_FAILED, DHT_MSG_LINK_LAYOUT_FAILED,
+    DHT_MSG_NO_SUBVOL_IN_LAYOUT, DHT_MSG_MEM_ALLOC_FAILED,
+    DHT_MSG_SET_IN_PARAMS_DICT_FAILED, DHT_MSG_LOC_COPY_FAILED,
+    DHT_MSG_PARENT_LOC_FAILED, DHT_MSG_CREATE_LOCK_FAILED,
+    DHT_MSG_PREV_ATTEMPT_FAILED, DHT_MSG_REFRESH_ATTEMPT,
+    DHT_MSG_ACQUIRE_LOCK_FAILED, DHT_MSG_CREATE_STUB_FAILED,
+    DHT_MSG_WIND_LOCK_REQ_FAILED, DHT_MSG_REFRESH_FAILED,
+    DHT_MSG_CACHED_SUBVOL_ERROR, DHT_MSG_NO_LINK_SUBVOL, DHT_MSG_SET_KEY_FAILED,
+    DHT_MSG_REMOVE_LINKTO_FAILED, DHT_MSG_LAYOUT_DICT_SET_FAILED,
+    DHT_MSG_XATTR_DICT_NULL, DHT_MSG_DUMMY_ALLOC_FAILED, DHT_MSG_DICT_IS_NULL,
+    DHT_MSG_LINK_INODE_FAILED, DHT_MSG_SELFHEAL_FAILED, DHT_MSG_NO_MDS_SUBVOL,
+    DHT_MSG_LIST_XATTRS_FAILED, DHT_MSG_RESET_INTER_XATTR_FAILED,
+    DHT_MSG_MDS_DOWN_UNABLE_TO_SET, DHT_MSG_WIND_UNLOCK_FAILED,
+    DHT_MSG_COMMIT_HASH_FAILED, DHT_MSG_UNLOCK_GFID_FAILED,
+    DHT_MSG_UNLOCK_FOLLOW_ENTRYLK, DHT_MSG_COPY_FRAME_FAILED,
+    DHT_MSG_UNLOCK_FOLLOW_LOCKS, DHT_MSG_ENTRYLK_FAILED_AFT_INODELK,
+    DHT_MSG_CALLOC_FAILED, DHT_MSG_LOCK_ALLOC_FAILED,
+    DHT_MSG_BLOCK_INODELK_FAILED,
+    DHT_MSG_LOCAL_LOCKS_STORE_FAILED_UNLOCKING_FOLLOWING_ENTRYLK,
+    DHT_MSG_ALLOC_FRAME_FAILED_NOT_UNLOCKING_FOLLOWING_ENTRYLKS,
+    DHT_MSG_DST_NULL_SET_FAILED);
+
+#define DHT_MSG_FD_CTX_SET_FAILED_STR "Failed to set fd ctx"
+#define DHT_MSG_INVALID_VALUE_STR "Different dst found in the fd ctx"
+#define DHT_MSG_UNKNOWN_FOP_STR "Unknown FOP on file"
+#define DHT_MSG_OPEN_FD_ON_DST_FAILED_STR "Failed to open the fd on file"
+#define DHT_MSG_SYNCTASK_CREATE_FAILED_STR "Failed to create synctask"
+#define DHT_MSG_ASPRINTF_FAILED_STR                                            \
+    "asprintf failed while fetching subvol from the id"
+#define DHT_MSG_HAS_MIGINFO_STR "Found miginfo in the inode ctx"
+#define DHT_MSG_FILE_LOOKUP_FAILED_STR "failed to lookup the file"
+#define DHT_MSG_INVALID_LINKFILE_STR                                           \
+    "linkto target is different from cached-subvol. treating as destination "  \
+    "subvol"
+#define DHT_MSG_GFID_MISMATCH_STR "gfid different on the target file"
+#define DHT_MSG_GET_XATTR_FAILED_STR "failed to get 'linkto' xattr"
+#define DHT_MSG_SET_INODE_CTX_FAILED_STR "failed to set inode-ctx target file"
+#define DHT_MSG_DIR_SELFHEAL_FAILED_STR "Healing of path failed"
+#define DHT_MSG_DIR_HEAL_ABORT_STR                                             \
+    "Failed to get path from subvol. Aborting directory healing"
+#define DHT_MSG_DIR_XATTR_HEAL_FAILED_STR "xattr heal failed for directory"
+#define DHT_MSG_LOCK_INODE_UNREF_FAILED_STR                                    \
+    "Found a NULL inode. Failed to unref the inode"
+#define DHT_MSG_DICT_SET_FAILED_STR "Failed to set dictionary value"
+#define DHT_MSG_NOT_LINK_FILE_ERROR_STR "got non-linkfile"
+#define DHT_MSG_CREATE_LINK_FAILED_STR "failed to initialize linkfile data"
+#define DHT_MSG_UNLINK_FAILED_STR "Unlinking linkfile on subvolume failed"
+#define DHT_MSG_MIGRATE_FILE_FAILED_STR "Migrate file failed"
+#define DHT_MSG_NO_MEMORY_STR "could not allocate memory for dict"
+#define DHT_MSG_SUBVOL_ERROR_STR "Failed to get linkto subvol"
+#define DHT_MSG_MIGRATE_HARDLINK_FILE_FAILED_STR "link failed on subvol"
+#define DHT_MSG_MIGRATE_FILE_SKIPPED_STR "Migration skipped"
+#define DHT_MSG_FD_CREATE_FAILED_STR "fd create failed"
+#define DHT_MSG_DICT_NEW_FAILED_STR "dict_new failed"
+#define DHT_MSG_FAILED_TO_OPEN_STR "failed to open"
+#define DHT_MSG_CREATE_FAILED_STR "failed to create"
+#define DHT_MSG_FILE_NOT_EXIST_STR "file does not exist"
+#define DHT_MSG_CHOWN_FAILED_STR "chown failed"
+#define DHT_MSG_FALLOCATE_FAILED_STR "fallocate failed"
+#define DHT_MSG_FTRUNCATE_FAILED_STR "ftruncate failed"
+#define DHT_MSG_STATFS_FAILED_STR "failed to get statfs"
+#define DHT_MSG_WRITE_CROSS_STR                                                \
+    "write will cross min-fre-disk for file on subvol. looking for new subvol"
+#define DHT_MSG_SUBVOL_INSUFF_SPACE_STR                                        \
+    "Could not find any subvol with space accommodating the file. Cosider "    \
+    "adding bricks"
+#define DHT_MSG_NEW_TARGET_FOUND_STR "New target found for file"
+#define DHT_MSG_INSUFF_MEMORY_STR "insufficient memory"
+#define DHT_MSG_SET_XATTR_FAILED_STR "failed to set xattr"
+#define DHT_MSG_SET_MODE_FAILED_STR "failed to set mode"
+#define DHT_MSG_FILE_EXISTS_IN_DEST_STR "file exists in destination"
+#define DHT_MSG_LINKFILE_DEL_FAILED_STR "failed to delete the linkfile"
+#define DHT_MSG_SYMLINK_FAILED_STR "symlink failed"
+#define DHT_MSG_MKNOD_FAILED_STR "mknod failed"
+#define DHT_MSG_SETATTR_FAILED_STR "failed to perform setattr"
+#define DHT_MSG_MIGRATE_CLEANUP_FAILED_STR                                     \
+    "Migrate file cleanup failed: failed to fstat file"
+#define DHT_MSG_LOCK_MIGRATE_STR "locks will be migrated for file"
+#define DHT_MSG_PARENT_BUILD_FAILED_STR                                        \
+    "failed to build parent loc, which is needed to acquire entrylk to "       \
+    "synchronize with renames on this path. Skipping migration"
+#define DHT_MSG_HASHED_SUBVOL_NOT_FOUND_STR                                    \
+    "cannot find hashed subvol which is needed to synchronize with renames "   \
+    "on this path. Skipping migration"
+#define DHT_MSG_ACQUIRE_ENTRYLK_FAILED_STR "failed to acquire entrylk on subvol"
+#define DHT_MSG_CREATE_DST_FAILED_STR "create dst failed for file"
+#define DHT_MSG_MIGRATION_EXIT_STR "Exiting migration"
+#define DHT_MSG_CHANGED_DST_STR "destination changed fo file"
+#define DHT_MSG_TRACE_FAILED_STR "Trace failed"
+#define DHT_MSG_WRITE_LOCK_FAILED_STR "write lock failed"
+#define DHT_MSG_GETACTIVELK_FAILED_STR "getactivelk failed for file"
+#define DHT_MSG_STAT_FAILED_STR "failed to do a stat"
+#define DHT_MSG_UNLINK_PERFORM_FAILED_STR "failed to perform unlink"
+#define DHT_MSG_MIGRATE_FILE_COMPLETE_STR "completed migration"
+#define DHT_MSG_CLANUP_SOURCE_FILE_FAILED_STR "failed to cleanup source file"
+#define DHT_MSG_UNLOCK_FILE_FAILED_STR "failed to unlock file"
+#define DHT_MSG_REMOVE_XATTR_FAILED_STR "remove xattr failed"
+#define DHT_MSG_SOCKET_ERROR_STR "Failed to unlink listener socket"
+#define DHT_MSG_HASHED_SUBVOL_GET_FAILED_STR "Failed to get hashed subvolume"
+#define DHT_MSG_CACHED_SUBVOL_GET_FAILED_STR "Failed to get cached subvolume"
+#define DHT_MSG_MIGRATE_DATA_FAILED_STR "migrate-data failed"
+#define DHT_MSG_DEFRAG_NULL_STR "defrag is NULL"
+#define DHT_MSG_DATA_MIGRATE_ABORT_STR                                         \
+    "Readdirp failed. Aborting data migration for dict"
+#define DHT_MSG_LAYOUT_FIX_FAILED_STR "fix layout failed"
+#define DHT_MSG_PARENT_NULL_STR "parent is NULL"
+#define DHT_MSG_GFID_NOT_PRESENT_STR "gfid not present"
+#define DHT_MSG_CHILD_LOC_FAILED_STR "Child loc build failed"
+#define DHT_MSG_SET_LOOKUP_FAILED_STR "Failed to set lookup"
+#define DHT_MSG_DIR_LOOKUP_FAILED_STR "lookup failed"
+#define DHT_MSG_DIR_REMOVED_STR "Dir renamed or removed. Skipping"
+#define DHT_MSG_READDIR_ERROR_STR "readdir failed, Aborting fix-layout"
+#define DHT_MSG_SETTLE_HASH_FAILED_STR "Settle hash failed"
+#define DHT_MSG_DEFRAG_PROCESS_DIR_FAILED_STR "gf_defrag_process_dir failed"
+#define DHT_MSG_FIX_NOT_COMP_STR                                               \
+    "Unable to retrieve fixlayout xattr. Assume background fix layout not "    \
+    "complete"
+#define DHT_MSG_SUBVOL_DETER_FAILED_STR                                        \
+    "local subvolume determination failed with error"
+#define DHT_MSG_LOCAL_SUBVOL_STR "local subvol"
+#define DHT_MSG_NODE_UUID_STR "node uuid"
+#define DHT_MSG_SIZE_FILE_STR "Total size files"
+#define DHT_MSG_GET_DATA_SIZE_FAILED_STR                                       \
+    "Failed to get the total data size. Unable to estimate time to complete "  \
+    "rebalance"
+#define DHT_MSG_PTHREAD_JOIN_FAILED_STR                                        \
+    "file_counter_thread: pthread_join failed"
+#define DHT_MSG_COUNTER_THREAD_CREATE_FAILED_STR                               \
+    "Failed to create the file counter thread"
+#define DHT_MSG_MIGRATION_INIT_QUEUE_FAILED_STR                                \
+    "Failed to initialise migration queue"
+#define DHT_MSG_REBALANCE_STOPPED_STR "Received stop command on rebalance"
+#define DHT_MSG_PAUSED_TIMEOUT_STR "Request pause timer timeout"
+#define DHT_MSG_WOKE_STR "woken"
+#define DHT_MSG_ABORT_REBALANCE_STR "Aborting rebalance"
+#define DHT_MSG_REBALANCE_START_FAILED_STR                                     \
+    "Failed to start rebalance: look up on / failed"
+#define DHT_MSG_CREATE_TASK_REBAL_FAILED_STR                                   \
+    "Could not create task for rebalance"
+#define DHT_MSG_REBAL_ESTIMATE_NOT_AVAIL_STR                                   \
+    "Rebalance estimates will not be available"
+#define DHT_MSG_REBALANCE_STATUS_STR "Rebalance status"
+#define DHT_MSG_DATA_NULL_STR "data value is NULL"
+#define DHT_MSG_ADD_CHOICES_ERROR_STR "Error to add choices in buffer"
+#define DHT_MSG_GET_CHOICES_ERROR_STR "Error to get choices"
+#define DHT_MSG_PREPARE_STATUS_ERROR_STR "Error to prepare status"
+#define DHT_MSG_SET_CHOICE_FAILED_STR "Failed to set full choice"
+#define DHT_MSG_AGGREGATE_QUOTA_XATTR_FAILED_STR                               \
+    "Failed to aggregate quota xattr"
+#define DHT_MSG_FILE_TYPE_MISMATCH_STR                                         \
+    "path exists as a file on one subvolume and directory on another. Please " \
+    "fix it manually"
+#define DHT_MSG_LAYOUT_SET_FAILED_STR "failed to set layout for subvolume"
+#define DHT_MSG_LAYOUT_MERGE_FAILED_STR "failed to merge layouts for subvolume"
+#define DHT_MSG_SET_HASHED_SUBVOL_FAILED_STR "Failed to set hashed subvolume"
+#define DHT_MSG_XATTR_HEAL_NOT_POSS_STR                                        \
+    "No gfid exists for path. so healing xattr is not possible"
+#define DHT_MSG_REVALIDATE_CBK_INFO_STR "Revalidate: subvolume returned -1"
+#define DHT_MSG_LAYOUT_MISMATCH_STR "Mismatching layouts"
+#define DHT_MSG_UNLINK_LOOKUP_INFO_STR "lookup_unlink retuened"
+#define DHT_MSG_LINKTO_FILE_FAILED_STR                                         \
+    "Could not unlink the linkto file as either fd is open and/or linkto "     \
+    "xattr is set"
+#define DHT_MSG_LAYOUT_PRESET_FAILED_STR                                       \
+    "Could not set pre-set layout for subvolume"
+#define DHT_MSG_FILE_ON_MULT_SUBVOL_STR                                        \
+    "multiple subvolumes have file (preferably rename the file in the "        \
+    "backend, and do a fresh lookup"
+#define DHT_MSG_STALE_LINKFILE_DELETE_STR                                      \
+    "attempting deletion of stale linkfile"
+#define DHT_MSG_LINK_FILE_LOOKUP_INFO_STR "Lookup on following linkfile"
+#define DHT_MSG_NO_SUBVOL_FOR_LINKTO_STR "No link subvolume for linkto"
+#define DHT_MSG_SUBVOL_RETURNED_STR "Subvolume returned -1"
+#define DHT_MSG_UNKNOWN_LOCAL_XSEL_STR "Unknown local->xsel"
+#define DHT_MSG_DICT_GET_FAILED_STR "Failed to get"
+#define DHT_MSG_UUID_PARSE_ERROR_STR "Failed to parse uuid"
+#define DHT_MSG_GET_XATTR_ERR_STR "getxattr err for dir"
+#define DHT_MSG_ALLOC_OR_FILL_FAILED_STR "alloc or fill failed"
+#define DHT_MSG_UPGRADE_BRICKS_STR                                             \
+    "At least one of the bricks does not support this operation. Please "      \
+    "upgrade all bricks"
+#define DHT_MSG_GET_REAL_NAME_FAILED_STR "Failed to get real filename"
+#define DHT_MSG_LAYOUT_NULL_STR "Layout is NULL"
+#define DHT_MSG_COPY_UUID_FAILED_STR "Failed to copy node uuid key"
+#define DHT_MSG_MDS_DETER_FAILED_STR                                           \
+    "Cannot determine MDS, fetching xattr randomly from a subvol"
+#define DHT_MSG_HASHED_SUBVOL_DOWN_STR                                         \
+    "MDS is down for path, so fetching xattr randomly from subvol"
+#define DHT_MSG_CREATE_REBAL_FAILED_STR                                        \
+    "failed to create a new rebalance synctask"
+#define DHT_MSG_FIX_LAYOUT_INFO_STR "fixing the layout"
+#define DHT_MSG_OPERATION_NOT_SUP_STR "wrong directory-spread-count value"
+#define DHT_MSG_LINK_LAYOUT_FAILED_STR "failed to link the layout in inode"
+#define DHT_MSG_NO_SUBVOL_IN_LAYOUT_STR "no subvolume in layout for path"
+#define DHT_MSG_INODE_LK_ERROR_STR "mknod lock failed for file"
+#define DHT_MSG_MEM_ALLOC_FAILED_STR "mem allocation failed"
+#define DHT_MSG_PARENT_LAYOUT_CHANGED_STR                                      \
+    "extracting in-memory layout of parent failed"
+#define DHT_MSG_SET_IN_PARAMS_DICT_FAILED_STR                                  \
+    "setting in params dictionary failed"
+#define DHT_MSG_LOC_COPY_FAILED_STR "loc_copy failed"
+#define DHT_MSG_LOC_FAILED_STR "parent loc build failed"
+#define DHT_MSG_PARENT_LOC_FAILED_STR "locking parent failed"
+#define DHT_MSG_CREATE_LOCK_FAILED_STR "Create lock failed"
+#define DHT_MSG_PREV_ATTEMPT_FAILED_STR                                        \
+    "mkdir loop detected. parent layout didn't change even though previous "   \
+    "attempt of mkdir failed because of in-memory layout not matching with "   \
+    "that on disk."
+#define DHT_MSG_REFRESH_ATTEMPT_STR                                            \
+    "mkdir parent layout changed. Attempting a refresh and then a retry"
+#define DHT_MSG_ACQUIRE_LOCK_FAILED_STR                                        \
+    "Acquiring lock on parent to guard against layout-change failed"
+#define DHT_MSG_CREATE_STUB_FAILED_STR "creating stub failed"
+#define DHT_MSG_WIND_LOCK_REQ_FAILED_STR                                       \
+    "cannot wind lock request to guard parent layout"
+#define DHT_MSG_REFRESH_FAILED_STR "refreshing parent layout failed."
+#define DHT_MSG_CACHED_SUBVOL_ERROR_STR "On cached subvol"
+#define DHT_MSG_NO_LINK_SUBVOL_STR "Linkfile does not have link subvolume"
+#define DHT_MSG_SET_KEY_FAILED_STR "failed to set key"
+#define DHT_MSG_CHILD_DOWN_STR "Received CHILD_DOWN. Exiting"
+#define DHT_MSG_LOG_FIXED_LAYOUT_STR "log layout fixed"
+#define DHT_MSG_REBAL_STRUCT_SET_STR "local->rebalance already set"
+#define DHT_MSG_REMOVE_LINKTO_FAILED_STR "Removal of linkto failed at subvol"
+#define DHT_MSG_LAYOUT_DICT_SET_FAILED_STR "dht layout dict set failed"
+#define DHT_MSG_SUBVOL_INFO_STR "creating subvolume"
+#define DHT_MSG_COMPUTE_HASH_FAILED_STR "hash computation failed"
+#define DHT_MSG_INVALID_DISK_LAYOUT_STR                                        \
+    "Invalid disk layout: Catastrophic error layout with unknown type found"
+#define DHT_MSG_LAYOUT_SORT_FAILED_STR "layout sort failed"
+#define DHT_MSG_ANOMALIES_INFO_STR "Found anomalies"
+#define DHT_MSG_XATTR_DICT_NULL_STR "xattr dictionary is NULL"
+#define DHT_MSG_DISK_LAYOUT_MISSING_STR "Disk layout missing"
+#define DHT_MSG_LAYOUT_INFO_STR "layout info"
+#define DHT_MSG_SUBVOL_NO_LAYOUT_INFO_STR "no pre-set layout for subvol"
+#define DHT_MSG_SELFHEAL_XATTR_FAILED_STR "layout setxattr failed"
+#define DHT_MSG_DIR_SELFHEAL_XATTR_FAILED_STR "Directory self heal xattr failed"
+#define DHT_MSG_DUMMY_ALLOC_FAILED_STR "failed to allocate dummy layout"
+#define DHT_MSG_DICT_IS_NULL_STR                                               \
+    "dict is NULL, need to make sure gfids are same"
+#define DHT_MSG_ENTRYLK_ERROR_STR "acquiring entrylk after inodelk failed"
+#define DHT_MSG_NO_DISK_USAGE_STATUS_STR "no du stats"
+#define DHT_MSG_LINK_INODE_FAILED_STR "linking inode failed"
+#define DHT_MSG_SELFHEAL_FAILED_STR "Directory selfheal failed"
+#define DHT_MSG_NO_MDS_SUBVOL_STR "No mds subvol"
+#define DHT_MSG_LIST_XATTRS_FAILED_STR "failed to list xattrs"
+#define DHT_MSG_RESET_INTER_XATTR_FAILED_STR "Failed to reset internal xattr"
+#define DHT_MSG_MDS_DOWN_UNABLE_TO_SET_STR                                     \
+    "mds subvol is down, unable to set xattr"
+#define DHT_MSG_DIR_ATTR_HEAL_FAILED_STR                                       \
+    "Directory attr heal failed. Failed to set uid/gid"
+#define DHT_MSG_WIND_UNLOCK_FAILED_STR                                         \
+    "Winding unlock failed: stale locks left on brick"
+#define DHT_MSG_COMMIT_HASH_FAILED_STR "Directory commit hash updaten failed"
+#define DHT_MSG_LK_ARRAY_INFO_STR "lk info"
+#define DHT_MSG_UNLOCK_GFID_FAILED_STR                                         \
+    "unlock failed on gfid: stale lock might be left"
+#define DHT_MSG_UNLOCKING_FAILED_STR "unlocking failed"
+#define DHT_MSG_UNLOCK_FOLLOW_ENTRYLK_STR "not unlocking following entrylks"
+#define DHT_MSG_COPY_FRAME_FAILED_STR "copy frame failed"
+#define DHT_MSG_UNLOCK_FOLLOW_LOCKS_STR "not unlocking following locks"
+#define DHT_MSG_INODELK_FAILED_STR "inodelk failed on subvol"
+#define DHT_MSG_LOCK_FRAME_FAILED_STR "memory allocation failed for lock_frame"
+#define DHT_MSG_LOCAL_LOCK_INIT_FAILED_STR "dht_local_lock_init failed"
+#define DHT_MSG_ENTRYLK_FAILED_AFT_INODELK_STR                                 \
+    "dht_blocking_entrylk failed after taking inodelk"
+#define DHT_MSG_BLOCK_INODELK_FAILED_STR "dht_blocking_inodelk failed"
+#define DHT_MSG_CALLOC_FAILED_STR "calloc failed"
+#define DHT_MSG_LOCK_ALLOC_FAILED_STR "lock allocation failed"
+#define DHT_MSG_ALLOC_FRAME_FAILED_NOT_UNLOCKING_FOLLOWING_ENTRYLKS_STR        \
+    "cannot allocate a frame, not unlocking following entrylks"
+#define DHT_MSG_LOCAL_LOCKS_STORE_FAILED_UNLOCKING_FOLLOWING_ENTRYLK_STR       \
+    "storing locks in local failed, not unlocking following entrylks"
+#define DHT_MSG_DST_NULL_SET_FAILED_STR                                        \
+    "src or dst is NULL, Failed to set dictionary value"
 
 #endif /* _DHT_MESSAGES_H_ */
diff --git a/xlators/cluster/dht/src/dht-rebalance.c b/xlators/cluster/dht/src/dht-rebalance.c
index 41a587823a9..8ba8082bd86 100644
--- a/xlators/cluster/dht/src/dht-rebalance.c
+++ b/xlators/cluster/dht/src/dht-rebalance.c
@@ -9,16 +9,15 @@
 */
 
 #include "dht-common.h"
-#include "xlator.h"
-#include "syscall.h"
-#include <signal.h>
+#include <glusterfs/syscall.h>
 #include <fnmatch.h>
 #include <signal.h>
-#include "events.h"
+#include <glusterfs/events.h>
+#include "glusterfs/compat-errno.h"  // for ENODATA on BSD
 
 #define GF_DISK_SECTOR_SIZE 512
-#define DHT_REBALANCE_PID 4242              /* Change it if required */
-#define DHT_REBALANCE_BLKSIZE (1024 * 1024) /* 1 MB */
+#define DHT_REBALANCE_PID 4242        /* Change it if required */
+#define DHT_REBALANCE_BLKSIZE 1048576 /* 1 MB */
 #define MAX_MIGRATE_QUEUE_COUNT 500
 #define MIN_MIGRATE_QUEUE_COUNT 200
 #define MAX_REBAL_TYPE_SIZE 16
@@ -46,7 +45,10 @@ gf_defrag_free_dir_dfmeta(struct dir_dfmeta *meta, int local_subvols_cnt)
 
     if (meta) {
         for (i = 0; i < local_subvols_cnt; i++) {
-            gf_dirent_free(&meta->equeue[i]);
+            if (meta->equeue)
+                gf_dirent_free(&meta->equeue[i]);
+            if (meta->lfd && meta->lfd[i])
+                fd_unref(meta->lfd[i]);
         }
 
         GF_FREE(meta->equeue);
@@ -54,6 +56,7 @@ gf_defrag_free_dir_dfmeta(struct dir_dfmeta *meta, int local_subvols_cnt)
         GF_FREE(meta->iterator);
         GF_FREE(meta->offset_var);
         GF_FREE(meta->fetch_entries);
+        GF_FREE(meta->lfd);
         GF_FREE(meta);
     }
 }
@@ -85,26 +88,6 @@ dht_set_global_defrag_error(gf_defrag_info_t *defrag, int ret)
     return;
 }
 
-static gf_boolean_t
-dht_is_tier_command(int cmd)
-{
-    gf_boolean_t is_tier = _gf_false;
-
-    switch (cmd) {
-        case GF_DEFRAG_CMD_START_TIER:
-        case GF_DEFRAG_CMD_STATUS_TIER:
-        case GF_DEFRAG_CMD_START_DETACH_TIER:
-        case GF_DEFRAG_CMD_STOP_DETACH_TIER:
-        case GF_DEFRAG_CMD_PAUSE_TIER:
-        case GF_DEFRAG_CMD_RESUME_TIER:
-            is_tier = _gf_true;
-            break;
-        default:
-            break;
-    }
-    return is_tier;
-}
-
 static int
 dht_send_rebalance_event(xlator_t *this, int cmd, gf_defrag_status_t status)
 {
@@ -113,8 +96,6 @@ dht_send_rebalance_event(xlator_t *this, int cmd, gf_defrag_status_t status)
     char *tmpstr = NULL;
     char *ptr = NULL;
     char *suffix = "-dht";
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
     int len = 0;
 
     eventtypes_t event = EVENT_LAST;
@@ -133,21 +114,14 @@ dht_send_rebalance_event(xlator_t *this, int cmd, gf_defrag_status_t status)
             break;
     }
 
-    if (dht_is_tier_command(cmd)) {
-        /* We should have the tier volume name*/
-        conf = this->private;
-        defrag = conf->defrag;
-        volname = defrag->tier_conf.volname;
-    } else {
-        /* DHT volume */
-        len = strlen(this->name) - strlen(suffix);
-        tmpstr = gf_strdup(this->name);
-        if (tmpstr) {
-            ptr = tmpstr + len;
-            if (!strcmp(ptr, suffix)) {
-                tmpstr[len] = '\0';
-                volname = tmpstr;
-            }
+    /* DHT volume */
+    len = strlen(this->name) - strlen(suffix);
+    tmpstr = gf_strdup(this->name);
+    if (tmpstr) {
+        ptr = tmpstr + len;
+        if (!strcmp(ptr, suffix)) {
+            tmpstr[len] = '\0';
+            volname = tmpstr;
         }
     }
 
@@ -173,75 +147,6 @@ dht_strip_out_acls(dict_t *dict)
     }
 }
 
-static int
-dht_write_with_holes(xlator_t *to, fd_t *fd, struct iovec *vec, int count,
-                     int32_t size, off_t offset, struct iobref *iobref,
-                     int *fop_errno)
-{
-    int i = 0;
-    int ret = -1;
-    int start_idx = 0;
-    int tmp_offset = 0;
-    int write_needed = 0;
-    int buf_len = 0;
-    int size_pending = 0;
-    char *buf = NULL;
-
-    /* loop through each vector */
-    for (i = 0; i < count; i++) {
-        buf = vec[i].iov_base;
-        buf_len = vec[i].iov_len;
-
-        for (start_idx = 0; (start_idx + GF_DISK_SECTOR_SIZE) <= buf_len;
-             start_idx += GF_DISK_SECTOR_SIZE) {
-            if (mem_0filled(buf + start_idx, GF_DISK_SECTOR_SIZE) != 0) {
-                write_needed = 1;
-                continue;
-            }
-
-            if (write_needed) {
-                ret = syncop_write(
-                    to, fd, (buf + tmp_offset), (start_idx - tmp_offset),
-                    (offset + tmp_offset), iobref, 0, NULL, NULL);
-                /* 'path' will be logged in calling function */
-                if (ret < 0) {
-                    gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
-                           strerror(-ret));
-                    *fop_errno = -ret;
-                    ret = -1;
-                    goto out;
-                }
-
-                write_needed = 0;
-            }
-            tmp_offset = start_idx + GF_DISK_SECTOR_SIZE;
-        }
-
-        if ((start_idx < buf_len) || write_needed) {
-            /* This means, last chunk is not yet written.. write it */
-            ret = syncop_write(to, fd, (buf + tmp_offset),
-                               (buf_len - tmp_offset), (offset + tmp_offset),
-                               iobref, 0, NULL, NULL);
-            if (ret < 0) {
-                /* 'path' will be logged in calling function */
-                gf_log(THIS->name, GF_LOG_WARNING, "failed to write (%s)",
-                       strerror(-ret));
-                *fop_errno = -ret;
-                ret = -1;
-                goto out;
-            }
-        }
-
-        size_pending = (size - buf_len);
-        if (!size_pending)
-            break;
-    }
-
-    ret = size;
-out:
-    return ret;
-}
-
 /*
    return values:
    -1 : failure
@@ -649,7 +554,7 @@ out:
 static int
 __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
                                 loc_t *loc, struct iatt *stbuf, fd_t **dst_fd,
-                                int *fop_errno)
+                                int *fop_errno, int file_has_holes)
 {
     int ret = -1;
     int ret2 = -1;
@@ -704,26 +609,23 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
         goto out;
     }
 
-    if (!!dht_is_tier_xlator(this)) {
-        xdata = dict_new();
-        if (!xdata) {
-            *fop_errno = ENOMEM;
-            ret = -1;
-            gf_msg(this->name, GF_LOG_ERROR, ENOMEM,
-                   DHT_MSG_MIGRATE_FILE_FAILED, "%s: dict_new failed)",
-                   loc->path);
-            goto out;
-        }
+    xdata = dict_new();
+    if (!xdata) {
+        *fop_errno = ENOMEM;
+        ret = -1;
+        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MIGRATE_FILE_FAILED,
+               "%s: dict_new failed)", loc->path);
+        goto out;
+    }
 
-        ret = dict_set_int32(xdata, GF_CLEAN_WRITE_PROTECTION, 1);
-        if (ret) {
-            *fop_errno = ENOMEM;
-            ret = -1;
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-                   "%s: failed to set dictionary value: key = %s ", loc->path,
-                   GF_CLEAN_WRITE_PROTECTION);
-            goto out;
-        }
+    ret = dict_set_int32_sizen(xdata, GF_CLEAN_WRITE_PROTECTION, 1);
+    if (ret) {
+        *fop_errno = ENOMEM;
+        ret = -1;
+        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+               "%s: failed to set dictionary value: key = %s ", loc->path,
+               GF_CLEAN_WRITE_PROTECTION);
+        goto out;
     }
 
     ret = syncop_lookup(to, loc, &new_stbuf, NULL, xdata, NULL);
@@ -818,7 +720,7 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
 
     /* No need to bother about 0 byte size files */
     if (stbuf->ia_size > 0) {
-        if (conf->use_fallocate) {
+        if (conf->use_fallocate && !file_has_holes) {
             ret = syncop_fallocate(to, fd, 0, 0, stbuf->ia_size, NULL, NULL);
             if (ret < 0) {
                 if (ret == -EOPNOTSUPP || ret == -EINVAL || ret == -ENOSYS) {
@@ -845,9 +747,7 @@ __dht_rebalance_create_dst_file(xlator_t *this, xlator_t *to, xlator_t *from,
                     goto out;
                 }
             }
-        }
-
-        if (!conf->use_fallocate) {
+        } else {
             ret = syncop_ftruncate(to, fd, stbuf->ia_size, NULL, NULL, NULL,
                                    NULL);
             if (ret < 0) {
@@ -875,7 +775,7 @@ out:
         dict_unref(dict);
 
     if (xdata)
-        dict_unref(dict);
+        dict_unref(xdata);
 
     return ret;
 }
@@ -1098,32 +998,103 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
     int ret = 0;
     int count = 0;
     off_t offset = 0;
+    off_t data_offset = 0;
+    off_t hole_offset = 0;
     struct iovec *vector = NULL;
     struct iobref *iobref = NULL;
     uint64_t total = 0;
     size_t read_size = 0;
+    size_t data_block_size = 0;
     dict_t *xdata = NULL;
     dht_conf_t *conf = NULL;
 
     conf = this->private;
+
     /* if file size is '0', no need to enter this loop */
     while (total < ia_size) {
-        read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
-                         ? DHT_REBALANCE_BLKSIZE
-                         : (ia_size - total));
+        /* This is a regular file - read it sequentially */
+        if (!hole_exists) {
+            read_size = (((ia_size - total) > DHT_REBALANCE_BLKSIZE)
+                             ? DHT_REBALANCE_BLKSIZE
+                             : (ia_size - total));
+        } else {
+            /* This is a sparse file - read only the data segments in the file
+             */
+
+            /* If the previous data block is fully copied, find the next data
+             * segment
+             * starting at the offset of the last read and written byte,  */
+            if (data_block_size <= 0) {
+                ret = syncop_seek(from, src, offset, GF_SEEK_DATA, NULL,
+                                  &data_offset);
+                if (ret) {
+                    if (ret == -ENXIO)
+                        ret = 0; /* No more data segments */
+                    else
+                        *fop_errno = -ret; /* Error occurred */
+
+                    break;
+                }
+
+                /* If the position of the current data segment is greater than
+                 * the position of the next hole, find the next hole in order to
+                 * calculate the length of the new data segment */
+                if (data_offset > hole_offset) {
+                    /* Starting at the offset of the last data segment, find the
+                     * next hole */
+                    ret = syncop_seek(from, src, data_offset, GF_SEEK_HOLE,
+                                      NULL, &hole_offset);
+                    if (ret) {
+                        /* If an error occurred here it's a real error because
+                         * if the seek for a data segment was successful then
+                         * necessarily another hole must exist (EOF is a hole)
+                         */
+                        *fop_errno = -ret;
+                        break;
+                    }
+
+                    /* Calculate the total size of the current data block */
+                    data_block_size = hole_offset - data_offset;
+                }
+            } else {
+                /* There is still data in the current segment, move the
+                 * data_offset to the position of the last written byte */
+                data_offset = offset;
+            }
+
+            /* Calculate how much data needs to be read and written. If the data
+             * segment's length is bigger than DHT_REBALANCE_BLKSIZE, read and
+             * write DHT_REBALANCE_BLKSIZE data length and the rest in the
+             * next iteration(s) */
+            read_size = ((data_block_size > DHT_REBALANCE_BLKSIZE)
+                             ? DHT_REBALANCE_BLKSIZE
+                             : data_block_size);
+
+            /* Calculate the remaining size of the data block - maybe there's no
+             * need to seek for data in the next iteration */
+            data_block_size -= read_size;
+
+            /* Set offset to the offset of the data segment so read and write
+             * will have the correct position */
+            offset = data_offset;
+        }
 
         ret = syncop_readv(from, src, read_size, offset, 0, &vector, &count,
                            &iobref, NULL, NULL, NULL);
+
         if (!ret || (ret < 0)) {
-            *fop_errno = -ret;
+            if (!ret) {
+                /* File was probably truncated*/
+                ret = -1;
+                *fop_errno = ENOSPC;
+            } else {
+                *fop_errno = -ret;
+            }
             break;
         }
 
-        if (hole_exists) {
-            ret = dht_write_with_holes(to, dst, vector, count, ret, offset,
-                                       iobref, fop_errno);
-        } else {
-            if (!conf->force_migration && !dht_is_tier_xlator(this)) {
+        if (!conf->force_migration) {
+            if (!xdata) {
                 xdata = dict_new();
                 if (!xdata) {
                     gf_msg("dht", GF_LOG_ERROR, 0, DHT_MSG_MIGRATE_FILE_FAILED,
@@ -1143,7 +1114,7 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
                  * https://github.com/gluster/glusterfs/issues/308
                  * for more details.
                  */
-                ret = dict_set_int32(xdata, GF_AVOID_OVERWRITE, 1);
+                ret = dict_set_int32_sizen(xdata, GF_AVOID_OVERWRITE, 1);
                 if (ret) {
                     gf_msg("dht", GF_LOG_ERROR, 0, ENOMEM,
                            "failed to set dict");
@@ -1152,22 +1123,12 @@ __dht_rebalance_migrate_data(xlator_t *this, gf_defrag_info_t *defrag,
                     break;
                 }
             }
-
-            ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
-                                NULL, xdata, NULL);
-            if (ret < 0) {
-                *fop_errno = -ret;
-            }
-        }
-
-        if ((defrag && defrag->cmd == GF_DEFRAG_CMD_START_TIER) &&
-            (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING)) {
-            gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_TIER_PAUSED,
-                   "Migrate file paused");
-            ret = -1;
         }
 
+        ret = syncop_writev(to, dst, vector, count, offset, iobref, 0, NULL,
+                            NULL, xdata, NULL);
         if (ret < 0) {
+            *fop_errno = -ret;
             break;
         }
 
@@ -1561,6 +1522,7 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
     xlator_t *old_target = NULL;
     xlator_t *hashed_subvol = NULL;
     fd_t *linkto_fd = NULL;
+    dict_t *xdata = NULL;
 
     if (from == to) {
         gf_msg_debug(this->name, 0,
@@ -1571,20 +1533,6 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
         goto out;
     }
 
-    /* If defrag is NULL, it should be assumed that migration is triggered
-     * from client */
-    defrag = conf->defrag;
-
-    /* migration of files from clients is restricted to non-tiered clients
-     * for now */
-    if (!defrag && dht_is_tier_xlator(this)) {
-        ret = ENOTSUP;
-        goto out;
-    }
-
-    if (defrag && defrag->tier_conf.is_tier)
-        log_level = GF_LOG_TRACE;
-
     gf_log(this->name, log_level, "%s: attempting to move from %s to %s",
            loc->path, from->name, to->name);
 
@@ -1627,6 +1575,10 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
                loc->path);
     }
 
+    /* The file is locked to prevent a rename during a migration. Renames
+     * and migrations on the file at the same time can lead to data loss.
+     */
+
     ret = dht_build_parent_loc(this, &parent_loc, loc, fop_errno);
     if (ret < 0) {
         ret = -1;
@@ -1727,9 +1679,13 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
         goto out;
     }
 
+    /* Try to preserve 'holes' while migrating data */
+    if (stbuf.ia_size > (stbuf.ia_blocks * GF_DISK_SECTOR_SIZE))
+        file_has_holes = 1;
+
     /* create the destination, with required modes/xattr */
     ret = __dht_rebalance_create_dst_file(this, to, from, loc, &stbuf, &dst_fd,
-                                          fop_errno);
+                                          fop_errno, file_has_holes);
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, 0, 0,
                "Create dst failed"
@@ -1773,8 +1729,8 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
          * destination. We need to do update this only post migration
          * as in case of failure the linkto needs to point to the source
          * subvol */
-        ret = __dht_rebalance_create_dst_file(this, to, from, loc, &stbuf,
-                                              &dst_fd, fop_errno);
+        ret = __dht_rebalance_create_dst_file(
+            this, to, from, loc, &stbuf, &dst_fd, fop_errno, file_has_holes);
         if (ret) {
             gf_log(this->name, GF_LOG_ERROR,
                    "Create dst failed"
@@ -1861,9 +1817,6 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
             ret = 0;
         goto out;
     }
-    /* Try to preserve 'holes' while migrating data */
-    if (stbuf.ia_size > (stbuf.ia_blocks * GF_DISK_SECTOR_SIZE))
-        file_has_holes = 1;
 
     ret = __dht_rebalance_migrate_data(this, defrag, from, to, src_fd, dst_fd,
                                        stbuf.ia_size, file_has_holes,
@@ -1878,7 +1831,15 @@ dht_migrate_file(xlator_t *this, loc_t *loc, xlator_t *from, xlator_t *to,
 
     /* TODO: Sync the locks */
 
-    ret = syncop_fsync(to, dst_fd, 0, NULL, NULL, NULL, NULL);
+    xdata = dict_new();
+    if (!xdata || dict_set_int8(xdata, "last-fsync", 1)) {
+        gf_log(this->name, GF_LOG_ERROR,
+               "%s: failed to set last-fsync flag on "
+               "%s (%s)",
+               loc->path, to->name, strerror(ENOMEM));
+    }
+
+    ret = syncop_fsync(to, dst_fd, 0, NULL, NULL, xdata, NULL);
     if (ret) {
         gf_log(this->name, GF_LOG_WARNING, "%s: failed to fsync on %s (%s)",
                loc->path, to->name, strerror(-ret));
@@ -2321,14 +2282,12 @@ out:
         }
     }
 
-    if (!dht_is_tier_xlator(this)) {
-        lk_ret = syncop_removexattr(to, loc, GF_PROTECT_FROM_EXTERNAL_WRITES,
-                                    NULL, NULL);
-        if (lk_ret && (lk_ret != -ENODATA) && (lk_ret != -ENOATTR)) {
-            gf_msg(this->name, GF_LOG_WARNING, -lk_ret, 0,
-                   "%s: removexattr failed key %s", loc->path,
-                   GF_PROTECT_FROM_EXTERNAL_WRITES);
-        }
+    lk_ret = syncop_removexattr(to, loc, GF_PROTECT_FROM_EXTERNAL_WRITES, NULL,
+                                NULL);
+    if (lk_ret && (lk_ret != -ENODATA) && (lk_ret != -ENOATTR)) {
+        gf_msg(this->name, GF_LOG_WARNING, -lk_ret, 0,
+               "%s: removexattr failed key %s", loc->path,
+               GF_PROTECT_FROM_EXTERNAL_WRITES);
     }
 
     if (dict)
@@ -2341,11 +2300,15 @@ out:
 
     if (dst_fd)
         syncop_close(dst_fd);
+
     if (src_fd)
         syncop_close(src_fd);
     if (linkto_fd)
         syncop_close(linkto_fd);
 
+    if (xdata)
+        dict_unref(xdata);
+
     loc_wipe(&tmp_loc);
     loc_wipe(&parent_loc);
 
@@ -2440,15 +2403,12 @@ void
 dht_build_root_inode(xlator_t *this, inode_t **inode)
 {
     inode_table_t *itable = NULL;
-    uuid_t root_gfid = {
-        0,
-    };
+    static uuid_t root_gfid = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1};
 
     itable = inode_table_new(0, this);
     if (!itable)
         return;
 
-    root_gfid[15] = 1;
     *inode = inode_find(itable, root_gfid);
 }
 
@@ -2578,10 +2538,10 @@ out:
  * all hardlinks.
  */
 
-int
+gf_boolean_t
 gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
 {
-    int ret = 0;
+    gf_boolean_t ret = _gf_false;
     int i = local_subvol_index;
     char *str = NULL;
     uint32_t hashval = 0;
@@ -2603,12 +2563,11 @@ gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
     }
 
     str = uuid_utoa_r(gfid, buf);
-    ret = dht_hash_compute(this, 0, str, &hashval);
-    if (ret == 0) {
+    if (dht_hash_compute(this, 0, str, &hashval) == 0) {
         index = (hashval % entry->count);
         if (entry->elements[index].info == REBAL_NODEUUID_MINE) {
             /* Index matches this node's nodeuuid.*/
-            ret = 1;
+            ret = _gf_true;
             goto out;
         }
 
@@ -2621,12 +2580,12 @@ gf_defrag_should_i_migrate(xlator_t *this, int local_subvol_index, uuid_t gfid)
                 /* None of the bricks in the subvol are up.
                  * CHILD_DOWN will kill the process soon */
 
-                return 0;
+                return _gf_false;
             }
 
             if (entry->elements[index].info == REBAL_NODEUUID_MINE) {
                 /* Index matches this node's nodeuuid.*/
-                ret = 1;
+                ret = _gf_true;
                 goto out;
             }
         }
@@ -2675,6 +2634,7 @@ gf_defrag_migrate_single_file(void *opaque)
     struct iatt *iatt_ptr = NULL;
     gf_boolean_t update_skippedcount = _gf_true;
     int i = 0;
+    gf_boolean_t should_i_migrate = 0;
 
     rebal_entry = (struct dht_container *)opaque;
     if (!rebal_entry) {
@@ -2729,17 +2689,29 @@ gf_defrag_migrate_single_file(void *opaque)
         goto out;
     }
 
-    if (!gf_defrag_should_i_migrate(this, rebal_entry->local_subvol_index,
-                                    entry->d_stat.ia_gfid)) {
-        gf_msg_debug(this->name, 0, "Don't migrate %s ", entry_loc.path);
-        goto out;
-    }
+    should_i_migrate = gf_defrag_should_i_migrate(
+        this, rebal_entry->local_subvol_index, entry->d_stat.ia_gfid);
 
     gf_uuid_copy(entry_loc.gfid, entry->d_stat.ia_gfid);
 
     gf_uuid_copy(entry_loc.pargfid, loc->gfid);
 
     ret = syncop_lookup(this, &entry_loc, &iatt, NULL, NULL, NULL);
+
+    if (!should_i_migrate) {
+        /* this node isn't supposed to migrate the file. suppressing any
+         * potential error from lookup as this file is under migration by
+         * another node */
+        if (ret) {
+            gf_msg_debug(this->name, -ret,
+                         "Ignoring lookup failure: node isn't migrating %s",
+                         entry_loc.path);
+            ret = 0;
+        }
+        gf_msg_debug(this->name, 0, "Don't migrate %s ", entry_loc.path);
+        goto out;
+    }
+
     if (ret) {
         gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_MIGRATE_FILE_FAILED,
                "Migrate file failed: %s lookup failed", entry_loc.path);
@@ -2902,8 +2874,7 @@ gf_defrag_migrate_single_file(void *opaque)
 
     if (defrag->stats == _gf_true) {
         gettimeofday(&end, NULL);
-        elapsed = (end.tv_sec - start.tv_sec) * 1e6 +
-                  (end.tv_usec - start.tv_usec);
+        elapsed = gf_tvdiff(&start, &end);
         gf_log(this->name, GF_LOG_INFO,
                "Migration of "
                "file:%s size:%" PRIu64
@@ -3082,9 +3053,9 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
                                dht_conf_t *conf, gf_defrag_info_t *defrag,
                                fd_t *fd, dict_t *migrate_data,
                                struct dir_dfmeta *dir_dfmeta, dict_t *xattr_req,
-                               int *should_commit_hash, int *perrno)
+                               int *perrno)
 {
-    int ret = -1;
+    int ret = 0;
     char is_linkfile = 0;
     gf_dirent_t *df_entry = NULL;
     struct dht_container *tmp_container = NULL;
@@ -3100,6 +3071,13 @@ int static gf_defrag_get_entry(xlator_t *this, int i,
     }
 
     if (dir_dfmeta->fetch_entries[i] == 1) {
+        if (!fd) {
+            dir_dfmeta->fetch_entries[i] = 0;
+            dir_dfmeta->offset_var[i].readdir_done = 1;
+            ret = 0;
+            goto out;
+        }
+
         ret = syncop_readdirp(conf->local_subvols[i], fd, 131072,
                               dir_dfmeta->offset_var[i].offset,
                               &(dir_dfmeta->equeue[i]), xattr_req, NULL);
@@ -3259,7 +3237,6 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                       dict_t *migrate_data, int *perrno)
 {
     int ret = -1;
-    fd_t *fd = NULL;
     dht_conf_t *conf = NULL;
     gf_dirent_t entries;
     dict_t *xattr_req = NULL;
@@ -3280,7 +3257,7 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
     int dfc_index = 0;
     int throttle_up = 0;
     struct dir_dfmeta *dir_dfmeta = NULL;
-    int should_commit_hash = 1;
+    xlator_t *old_THIS = NULL;
 
     gf_log(this->name, GF_LOG_INFO, "migrate data called on %s", loc->path);
     gettimeofday(&dir_start, NULL);
@@ -3293,28 +3270,53 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
         goto out;
     }
 
-    fd = fd_create(loc->inode, defrag->pid);
-    if (!fd) {
-        gf_log(this->name, GF_LOG_ERROR, "Failed to create fd");
+    old_THIS = THIS;
+    THIS = this;
+
+    dir_dfmeta = GF_CALLOC(1, sizeof(*dir_dfmeta), gf_common_mt_pointer);
+    if (!dir_dfmeta) {
+        gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta is NULL");
         ret = -1;
         goto out;
     }
 
-    ret = syncop_opendir(this, loc, fd, NULL, NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_MIGRATE_DATA_FAILED,
-               "Migrate data failed: Failed to open dir %s", loc->path);
-        *perrno = -ret;
+    dir_dfmeta->lfd = GF_CALLOC(local_subvols_cnt, sizeof(fd_t *),
+                                gf_common_mt_pointer);
+    if (!dir_dfmeta->lfd) {
+        gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_INSUFF_MEMORY,
+                "for dir_dfmeta", NULL);
         ret = -1;
+        *perrno = ENOMEM;
         goto out;
     }
 
-    fd_bind(fd);
-    dir_dfmeta = GF_CALLOC(1, sizeof(*dir_dfmeta), gf_common_mt_pointer);
-    if (!dir_dfmeta) {
-        gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta is NULL");
-        ret = -1;
-        goto out;
+    for (i = 0; i < local_subvols_cnt; i++) {
+        dir_dfmeta->lfd[i] = fd_create(loc->inode, defrag->pid);
+        if (!dir_dfmeta->lfd[i]) {
+            gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_FD_CREATE_FAILED,
+                    NULL);
+            *perrno = ENOMEM;
+            ret = -1;
+            goto out;
+        }
+
+        ret = syncop_opendir(conf->local_subvols[i], loc, dir_dfmeta->lfd[i],
+                             NULL, NULL);
+        if (ret) {
+            fd_unref(dir_dfmeta->lfd[i]);
+            dir_dfmeta->lfd[i] = NULL;
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_FAILED_TO_OPEN,
+                    "dir: %s", loc->path, "subvol: %s",
+                    conf->local_subvols[i]->name, NULL);
+
+            if (conf->decommission_in_progress) {
+                *perrno = -ret;
+                ret = -1;
+                goto out;
+            }
+        } else {
+            fd_bind(dir_dfmeta->lfd[i]);
+        }
     }
 
     dir_dfmeta->head = GF_CALLOC(local_subvols_cnt, sizeof(*(dir_dfmeta->head)),
@@ -3349,6 +3351,7 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
         ret = -1;
         goto out;
     }
+
     ret = gf_defrag_ctx_subvols_init(dir_dfmeta->offset_var, this);
     if (ret) {
         gf_log(this->name, GF_LOG_ERROR,
@@ -3361,7 +3364,8 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
     dir_dfmeta->fetch_entries = GF_CALLOC(local_subvols_cnt, sizeof(int),
                                           gf_common_mt_int);
     if (!dir_dfmeta->fetch_entries) {
-        gf_log(this->name, GF_LOG_ERROR, "dir_dfmeta->fetch_entries is NULL");
+        gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_INSUFF_MEMORY,
+                "for dir_dfmeta->fetch_entries", NULL);
         ret = -1;
         goto out;
     }
@@ -3431,8 +3435,13 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
             ldfq_count <= MAX_MIGRATE_QUEUE_COUNT &&
             !dht_dfreaddirp_done(dir_dfmeta->offset_var, local_subvols_cnt)) {
             ret = gf_defrag_get_entry(this, dfc_index, &container, loc, conf,
-                                      defrag, fd, migrate_data, dir_dfmeta,
-                                      xattr_req, &should_commit_hash, perrno);
+                                      defrag, dir_dfmeta->lfd[dfc_index],
+                                      migrate_data, dir_dfmeta, xattr_req,
+                                      perrno);
+
+            if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED) {
+                goto out;
+            }
 
             if (ret) {
                 gf_log(this->name, GF_LOG_WARNING,
@@ -3472,27 +3481,19 @@ gf_defrag_process_dir(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
     }
 
     gettimeofday(&end, NULL);
-    elapsed = (end.tv_sec - dir_start.tv_sec) * 1e6 +
-              (end.tv_usec - dir_start.tv_usec);
+    elapsed = gf_tvdiff(&dir_start, &end);
     gf_log(this->name, GF_LOG_INFO,
            "Migration operation on dir %s took "
            "%.2f secs",
            loc->path, elapsed / 1e6);
     ret = 0;
 out:
-
+    THIS = old_THIS;
     gf_defrag_free_dir_dfmeta(dir_dfmeta, local_subvols_cnt);
 
     if (xattr_req)
         dict_unref(xattr_req);
 
-    if (fd)
-        fd_unref(fd);
-
-    if (ret == 0 && should_commit_hash == 0) {
-        ret = 2;
-    }
-
     /* It does not matter if it errored out - this number is
      * used to calculate rebalance estimated time to complete.
      * No locking required as dirs are processed by a single thread.
@@ -3500,6 +3501,7 @@ out:
     defrag->num_dirs_processed++;
     return ret;
 }
+
 int
 gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                       dict_t *fix_layout)
@@ -3514,7 +3516,6 @@ gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
      * rebalance is complete.
      */
     if (defrag->cmd == GF_DEFRAG_CMD_START_LAYOUT_FIX ||
-        defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER ||
         defrag->cmd == GF_DEFRAG_CMD_DETACH_START) {
         return 0;
     }
@@ -3560,114 +3561,6 @@ gf_defrag_settle_hash(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
     return 0;
 }
 
-/* Function for doing a named lookup on file inodes during an attach tier
- * So that a hardlink lookup heal i.e gfid to parent gfid lookup heal
- * happens on pre-existing data. This is required so that the ctr database has
- * hardlinks of all the exisitng file in the volume. CTR xlator on the
- * brick/server side does db update/insert of the hardlink on a namelookup.
- * Currently the namedlookup is done synchronous to the fixlayout that is
- * triggered by attach tier. This is not performant, adding more time to
- * fixlayout. The performant approach is record the hardlinks on a compressed
- * datastore and then do the namelookup asynchronously later, giving the ctr db
- * eventual consistency
- * */
-int
-gf_fix_layout_tier_attach_lookup(xlator_t *this, loc_t *parent_loc,
-                                 gf_dirent_t *file_dentry)
-{
-    int ret = -1;
-    dict_t *lookup_xdata = NULL;
-    dht_conf_t *conf = NULL;
-    loc_t file_loc = {
-        0,
-    };
-    struct iatt iatt = {
-        0,
-    };
-
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, parent_loc, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, file_dentry, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, this->private, out);
-
-    if (!parent_loc->inode) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "%s/%s parent is NULL", parent_loc->path, file_dentry->d_name);
-        goto out;
-    }
-
-    conf = this->private;
-
-    loc_wipe(&file_loc);
-
-    if (gf_uuid_is_null(file_dentry->d_stat.ia_gfid)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "%s/%s gfid not present", parent_loc->path, file_dentry->d_name);
-        goto out;
-    }
-
-    gf_uuid_copy(file_loc.gfid, file_dentry->d_stat.ia_gfid);
-
-    if (gf_uuid_is_null(parent_loc->gfid)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "%s/%s"
-               " gfid not present",
-               parent_loc->path, file_dentry->d_name);
-        goto out;
-    }
-
-    gf_uuid_copy(file_loc.pargfid, parent_loc->gfid);
-
-    ret = dht_build_child_loc(this, &file_loc, parent_loc, file_dentry->d_name);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Child loc build failed");
-        ret = -1;
-        goto out;
-    }
-
-    lookup_xdata = dict_new();
-    if (!lookup_xdata) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed creating lookup dict for %s", file_dentry->d_name);
-        goto out;
-    }
-
-    ret = dict_set_int32(lookup_xdata, CTR_ATTACH_TIER_LOOKUP, 1);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to set lookup flag");
-        goto out;
-    }
-
-    gf_uuid_copy(file_loc.parent->gfid, parent_loc->gfid);
-
-    /* Sending lookup to cold tier only */
-    ret = syncop_lookup(conf->subvolumes[0], &file_loc, &iatt, NULL,
-                        lookup_xdata, NULL);
-    if (ret) {
-        /* If the file does not exist on the cold tier than it must */
-        /* have been discovered on the hot tier. This is not an error. */
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "%s lookup to cold tier on attach heal failed", file_loc.path);
-        goto out;
-    }
-
-    ret = 0;
-
-out:
-
-    loc_wipe(&file_loc);
-
-    if (lookup_xdata)
-        dict_unref(lookup_xdata);
-
-    return ret;
-}
-
 int
 gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                      dict_t *fix_layout, dict_t *migrate_data)
@@ -3687,7 +3580,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
     };
     inode_t *linked_inode = NULL, *inode = NULL;
     dht_conf_t *conf = NULL;
-    int should_commit_hash = 1;
     int perrno = 0;
 
     conf = this->private;
@@ -3790,16 +3682,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
             if (!strcmp(entry->d_name, ".") || !strcmp(entry->d_name, ".."))
                 continue;
             if (!IA_ISDIR(entry->d_stat.ia_type)) {
-                /* If its a fix layout during the attach
-                 * tier operation do lookups on files
-                 * on cold subvolume so that there is a
-                 * CTR DB Lookup Heal triggered on existing
-                 * data.
-                 * */
-                if (defrag->cmd == GF_DEFRAG_CMD_START_TIER) {
-                    gf_fix_layout_tier_attach_lookup(this, loc, entry);
-                }
-
                 continue;
             }
             loc_wipe(&entry_loc);
@@ -3816,8 +3698,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
 
                     goto out;
                 } else {
-                    should_commit_hash = 0;
-
                     continue;
                 }
             }
@@ -3880,7 +3760,6 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                         ret = -1;
                         goto out;
                     } else {
-                        should_commit_hash = 0;
                         continue;
                     }
                 }
@@ -3893,7 +3772,12 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
             ret = gf_defrag_fix_layout(this, defrag, &entry_loc, fix_layout,
                                        migrate_data);
 
-            if (ret && ret != 2) {
+            if (defrag->defrag_status == GF_DEFRAG_STATUS_STOPPED ||
+                defrag->defrag_status == GF_DEFRAG_STATUS_FAILED) {
+                goto out;
+            }
+
+            if (ret) {
                 gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LAYOUT_FIX_FAILED,
                        "Fix layout failed for %s", entry_loc.path);
 
@@ -3916,7 +3800,25 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
         INIT_LIST_HEAD(&entries.list);
     }
 
+    /* A directory layout is fixed only after its subdirs are healed to
+     * any newly added bricks. If the layout is fixed before subdirs are
+     * healed, the newly added brick will get a non-null layout.
+     * Any subdirs which hash to that layout will no longer show up
+     * in a directory listing until they are healed.
+     */
+
     ret = syncop_setxattr(this, loc, fix_layout, 0, NULL, NULL);
+
+    /* In case of a race where the directory is deleted just before
+     * layout setxattr, the errors are updated in the layout structure.
+     * We can use this information to make a decision whether the directory
+     * is deleted entirely.
+     */
+    if (ret == 0) {
+        ret = dht_dir_layout_error_check(this, loc->inode);
+        ret = -ret;
+    }
+
     if (ret) {
         if (-ret == ENOENT || -ret == ESTALE) {
             gf_msg(this->name, GF_LOG_INFO, -ret, DHT_MSG_LAYOUT_FIX_FAILED,
@@ -3927,6 +3829,7 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                 defrag->total_failures++;
             }
             ret = 0;
+            goto out;
         } else {
             gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LAYOUT_FIX_FAILED,
                    "Setxattr failed for %s", loc->path);
@@ -3941,11 +3844,10 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
         }
     }
 
-    if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
-        (defrag->cmd != GF_DEFRAG_CMD_START_LAYOUT_FIX)) {
+    if (defrag->cmd != GF_DEFRAG_CMD_START_LAYOUT_FIX) {
         ret = gf_defrag_process_dir(this, defrag, loc, migrate_data, &perrno);
 
-        if (ret && (ret != 2)) {
+        if (ret) {
             if (perrno == ENOENT || perrno == ESTALE) {
                 ret = 0;
                 goto out;
@@ -3961,18 +3863,13 @@ gf_defrag_fix_layout(xlator_t *this, gf_defrag_info_t *defrag, loc_t *loc,
                 if (conf->decommission_in_progress) {
                     goto out;
                 }
-
-                should_commit_hash = 0;
             }
-        } else if (ret == 2) {
-            should_commit_hash = 0;
         }
     }
 
     gf_msg_trace(this->name, 0, "fix layout called on %s", loc->path);
 
-    if (should_commit_hash &&
-        gf_defrag_settle_hash(this, defrag, loc, fix_layout) != 0) {
+    if (gf_defrag_settle_hash(this, defrag, loc, fix_layout) != 0) {
         defrag->total_failures++;
 
         gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SETTLE_HASH_FAILED,
@@ -3996,245 +3893,34 @@ out:
     if (fd)
         fd_unref(fd);
 
-    if (ret == 0 && should_commit_hash == 0) {
-        ret = 2;
-    }
-
-    return ret;
-}
-
-/******************************************************************************
- *                      Tier background Fix layout functions
- ******************************************************************************/
-/* This is the background tier fixlayout thread */
-void *
-gf_tier_do_fix_layout(void *args)
-{
-    gf_tier_fix_layout_arg_t *tier_fix_layout_arg = args;
-    int ret = -1;
-    xlator_t *this = NULL;
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    dict_t *dict = NULL;
-    loc_t loc = {
-        0,
-    };
-    struct iatt iatt = {
-        0,
-    };
-    struct iatt parent = {
-        0,
-    };
-
-    GF_VALIDATE_OR_GOTO("tier", tier_fix_layout_arg, out);
-    GF_VALIDATE_OR_GOTO("tier", tier_fix_layout_arg->this, out);
-    this = tier_fix_layout_arg->this;
-
-    conf = this->private;
-    GF_VALIDATE_OR_GOTO(this->name, conf, out);
-
-    defrag = conf->defrag;
-    GF_VALIDATE_OR_GOTO(this->name, defrag, out);
-    GF_VALIDATE_OR_GOTO(this->name, defrag->root_inode, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, tier_fix_layout_arg->fix_layout, out);
-
-    /* Get Root loc_t */
-    dht_build_root_loc(defrag->root_inode, &loc);
-    ret = syncop_lookup(this, &loc, &iatt, &parent, NULL, NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_REBALANCE_START_FAILED,
-               "Lookup on root failed.");
-        ret = -1;
-        goto out;
-    }
-
-    /* Start the crawl */
-    gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-           "Tiering Fixlayout started");
-
-    ret = gf_defrag_fix_layout(this, defrag, &loc,
-                               tier_fix_layout_arg->fix_layout, NULL);
-    if (ret && ret != 2) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_REBALANCE_FAILED,
-               "Tiering fixlayout failed.");
-        ret = -1;
-        goto out;
-    }
-
-    if (ret != 2 &&
-        gf_defrag_settle_hash(this, defrag, &loc,
-                              tier_fix_layout_arg->fix_layout) != 0) {
-        defrag->total_failures++;
-        ret = -1;
-        goto out;
-    }
-
-    dict = dict_new();
-    if (!dict) {
-        ret = -1;
-        goto out;
-    }
-
-    ret = dict_set_str(dict, GF_XATTR_TIER_LAYOUT_FIXED_KEY, "yes");
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_REBALANCE_FAILED,
-               "Failed to set dictionary value: key = %s",
-               GF_XATTR_TIER_LAYOUT_FIXED_KEY);
-        ret = -1;
-        goto out;
-    }
-
-    /* Marking the completion of tiering fix layout via a xattr on root */
-    ret = syncop_setxattr(this, &loc, dict, 0, NULL, NULL);
-    if (ret) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "Failed to set tiering fix "
-               "layout completed xattr on %s",
-               loc.path);
-        ret = -1;
-        goto out;
-    }
-
-    ret = 0;
-out:
-    if (ret && defrag)
-        defrag->total_failures++;
-
-    if (dict)
-        dict_unref(dict);
-
-    return NULL;
-}
-
-int
-gf_tier_start_fix_layout(xlator_t *this, loc_t *loc, gf_defrag_info_t *defrag,
-                         dict_t *fix_layout)
-{
-    int ret = -1;
-    dict_t *tier_dict = NULL;
-    gf_tier_fix_layout_arg_t *tier_fix_layout_arg = NULL;
-
-    tier_dict = dict_new();
-    if (!tier_dict) {
-        gf_log("tier", GF_LOG_ERROR,
-               "Tier fix layout failed :"
-               "Creation of tier_dict failed");
-        ret = -1;
-        goto out;
-    }
-
-    /* Check if layout is fixed already */
-    ret = syncop_getxattr(this, loc, &tier_dict, GF_XATTR_TIER_LAYOUT_FIXED_KEY,
-                          NULL, NULL);
-    if (ret != 0) {
-        tier_fix_layout_arg = &defrag->tier_conf.tier_fix_layout_arg;
-
-        /*Fill crawl arguments */
-        tier_fix_layout_arg->this = this;
-        tier_fix_layout_arg->fix_layout = fix_layout;
-
-        /* Spawn the fix layout thread so that its done in the
-         * background */
-        ret = gf_thread_create(&tier_fix_layout_arg->thread_id, NULL,
-                               gf_tier_do_fix_layout, tier_fix_layout_arg,
-                               "tierfixl");
-        if (ret) {
-            gf_log("tier", GF_LOG_ERROR,
-                   "Thread creation failed. "
-                   "Background fix layout for tiering will not "
-                   "work.");
-            defrag->total_failures++;
-            goto out;
-        }
-    }
-    ret = 0;
-out:
-    if (tier_dict)
-        dict_unref(tier_dict);
-
     return ret;
 }
 
-void
-gf_tier_clear_fix_layout(xlator_t *this, loc_t *loc, gf_defrag_info_t *defrag)
-{
-    int ret = -1;
-    dict_t *dict = NULL;
-
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-    GF_VALIDATE_OR_GOTO(this->name, loc, out);
-    GF_VALIDATE_OR_GOTO(this->name, defrag, out);
-
-    /* Check if background fixlayout is completed. This is not
-     * multi-process safe i.e there is a possibility that by the time
-     * we move to remove the xattr there it might have been cleared by some
-     * other detach process from other node. We ignore the error if such
-     * a thing happens */
-    ret = syncop_getxattr(this, loc, &dict, GF_XATTR_TIER_LAYOUT_FIXED_KEY,
-                          NULL, NULL);
-    if (ret) {
-        /* Background fixlayout not complete - nothing to clear*/
-        gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_LOG_TIER_STATUS,
-               "Unable to retrieve fixlayout xattr."
-               "Assume background fix layout not complete");
-        goto out;
-    }
-
-    ret = syncop_removexattr(this, loc, GF_XATTR_TIER_LAYOUT_FIXED_KEY, NULL,
-                             NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_LOG_TIER_STATUS,
-               "Failed removing tier fix layout "
-               "xattr from %s",
-               loc->path);
-        goto out;
-    }
-    ret = 0;
-out:
-    if (dict)
-        dict_unref(dict);
-}
-
-void
-gf_tier_wait_fix_lookup(gf_defrag_info_t *defrag)
-{
-    if (defrag->tier_conf.tier_fix_layout_arg.thread_id) {
-        pthread_join(defrag->tier_conf.tier_fix_layout_arg.thread_id, NULL);
-    }
-}
-/******************Tier background Fix layout functions END********************/
-
 int
 dht_init_local_subvols_and_nodeuuids(xlator_t *this, dht_conf_t *conf,
                                      loc_t *loc)
 {
     dict_t *dict = NULL;
-    gf_defrag_info_t *defrag = NULL;
     uuid_t *uuid_ptr = NULL;
     int ret = -1;
     int i = 0;
     int j = 0;
 
-    defrag = conf->defrag;
-
-    if (defrag->cmd != GF_DEFRAG_CMD_START_TIER) {
-        /* Find local subvolumes */
-        ret = syncop_getxattr(this, loc, &dict, GF_REBAL_FIND_LOCAL_SUBVOL,
-                              NULL, NULL);
-        if (ret && (ret != -ENODATA)) {
-            gf_msg(this->name, GF_LOG_ERROR, -ret, 0,
-                   "local "
-                   "subvolume determination failed with error: %d",
-                   -ret);
-            ret = -1;
-            goto out;
-        }
-
-        if (!ret)
-            goto out;
+    /* Find local subvolumes */
+    ret = syncop_getxattr(this, loc, &dict, GF_REBAL_FIND_LOCAL_SUBVOL, NULL,
+                          NULL);
+    if (ret && (ret != -ENODATA)) {
+        gf_msg(this->name, GF_LOG_ERROR, -ret, 0,
+               "local "
+               "subvolume determination failed with error: %d",
+               -ret);
+        ret = -1;
+        goto out;
     }
 
+    if (!ret)
+        goto out;
+
     ret = syncop_getxattr(this, loc, &dict, GF_REBAL_OLD_FIND_LOCAL_SUBVOL,
                           NULL, NULL);
     if (ret) {
@@ -4325,9 +4011,6 @@ dht_file_counter_thread(void *args)
     struct timespec time_to_wait = {
         0,
     };
-    struct timeval now = {
-        0,
-    };
     uint64_t tmp_size = 0;
 
     if (!args)
@@ -4337,9 +4020,8 @@ dht_file_counter_thread(void *args)
     dht_build_root_loc(defrag->root_inode, &root_loc);
 
     while (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED) {
-        gettimeofday(&now, NULL);
-        time_to_wait.tv_sec = now.tv_sec + 600;
-        time_to_wait.tv_nsec = 0;
+        timespec_now(&time_to_wait);
+        time_to_wait.tv_sec += 600;
 
         pthread_mutex_lock(&defrag->fc_mutex);
         pthread_cond_timedwait(&defrag->fc_wakeup_cond, &defrag->fc_mutex,
@@ -4412,7 +4094,7 @@ gf_defrag_estimates_init(xlator_t *this, loc_t *loc, pthread_t *filecnt_thread)
         goto out;
     }
 
-    ret = gf_thread_create(filecnt_thread, NULL, &dht_file_counter_thread,
+    ret = gf_thread_create(filecnt_thread, NULL, dht_file_counter_thread,
                            (void *)defrag, "dhtfcnt");
 
     if (ret) {
@@ -4436,9 +4118,6 @@ gf_defrag_parallel_migration_init(xlator_t *this, gf_defrag_info_t *defrag,
     int thread_spawn_count = 0;
     int index = 0;
     pthread_t *tid = NULL;
-    char thread_name[GF_THREAD_NAMEMAX] = {
-        0,
-    };
 
     if (!defrag)
         goto out;
@@ -4472,10 +4151,8 @@ gf_defrag_parallel_migration_init(xlator_t *this, gf_defrag_info_t *defrag,
 
     /*Spawn Threads Here*/
     while (index < thread_spawn_count) {
-        snprintf(thread_name, sizeof(thread_name), "dhtmig%d",
-                 ((index + 1) & 0x3ff));
-        ret = gf_thread_create(&(tid[index]), NULL, &gf_defrag_task,
-                               (void *)defrag, thread_name);
+        ret = gf_thread_create(&(tid[index]), NULL, gf_defrag_task,
+                               (void *)defrag, "dhtmig%d", (index + 1) & 0x3ff);
         if (ret != 0) {
             gf_msg("DHT", GF_LOG_ERROR, ret, 0, "Thread[%d] creation failed. ",
                    index);
@@ -4548,7 +4225,6 @@ gf_defrag_start_crawl(void *data)
     dict_t *migrate_data = NULL;
     dict_t *status = NULL;
     glusterfs_ctx_t *ctx = NULL;
-    dht_methods_t *methods = NULL;
     call_frame_t *statfs_frame = NULL;
     xlator_t *old_THIS = NULL;
     int ret = -1;
@@ -4564,7 +4240,6 @@ gf_defrag_start_crawl(void *data)
     int thread_index = 0;
     pthread_t *tid = NULL;
     pthread_t filecnt_thread;
-    gf_boolean_t is_tier_detach = _gf_false;
     gf_boolean_t fc_thread_started = _gf_false;
 
     this = data;
@@ -4583,7 +4258,8 @@ gf_defrag_start_crawl(void *data)
     if (!defrag)
         goto exit;
 
-    gettimeofday(&defrag->start_time, NULL);
+    defrag->start_time = gf_time();
+
     dht_build_root_inode(this, &defrag->root_inode);
     if (!defrag->root_inode)
         goto out;
@@ -4717,43 +4393,17 @@ gf_defrag_start_crawl(void *data)
         }
     }
 
-    if (defrag->cmd == GF_DEFRAG_CMD_START_TIER) {
-        /* Fix layout for attach tier */
-        ret = gf_tier_start_fix_layout(this, &loc, defrag, fix_layout);
-        if (ret) {
-            goto out;
-        }
-
-        methods = &(conf->methods);
-
-        /* Calling tier_start of tier.c */
-        methods->migration_other(this, defrag);
-        if (defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER ||
-            defrag->cmd == GF_DEFRAG_CMD_DETACH_START) {
-            ret = dict_set_str(migrate_data, GF_XATTR_FILE_MIGRATE_KEY,
-                               "force");
-            if (ret)
-                goto out;
-        }
-    } else {
-        ret = gf_defrag_fix_layout(this, defrag, &loc, fix_layout,
-                                   migrate_data);
-        if (ret && ret != 2) {
-            defrag->total_failures++;
-            ret = -1;
-            goto out;
-        }
-
-        if (ret != 2 &&
-            gf_defrag_settle_hash(this, defrag, &loc, fix_layout) != 0) {
-            defrag->total_failures++;
-            ret = -1;
-            goto out;
-        }
+    ret = gf_defrag_fix_layout(this, defrag, &loc, fix_layout, migrate_data);
+    if (ret) {
+        defrag->total_failures++;
+        ret = -1;
+        goto out;
+    }
 
-        if (defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER ||
-            defrag->cmd == GF_DEFRAG_CMD_DETACH_START)
-            is_tier_detach = _gf_true;
+    if (gf_defrag_settle_hash(this, defrag, &loc, fix_layout) != 0) {
+        defrag->total_failures++;
+        ret = -1;
+        goto out;
     }
 
     gf_log("DHT", GF_LOG_INFO, "crawling file-system completed");
@@ -4767,19 +4417,6 @@ out:
         defrag->defrag_status = GF_DEFRAG_STATUS_FAILED;
     }
 
-    if (defrag->cmd == GF_DEFRAG_CMD_START_TIER) {
-        /* Wait for the tier fixlayout to
-         * complete if its was started.*/
-        gf_tier_wait_fix_lookup(defrag);
-    }
-
-    if (is_tier_detach && ret == 0) {
-        /* If it was a detach remove the tier fix-layout
-         * xattr on root. Ignoring the failure, as nothing has to be
-         * done, logging is done in gf_tier_clear_fix_layout */
-        gf_tier_clear_fix_layout(this, &loc, defrag);
-    }
-
     gf_defrag_parallel_migration_cleanup(defrag, tid, thread_index);
 
     if ((defrag->defrag_status != GF_DEFRAG_STATUS_STOPPED) &&
@@ -4793,9 +4430,9 @@ out:
 
     dht_send_rebalance_event(this, defrag->cmd, defrag->defrag_status);
 
+    status = dict_new();
     LOCK(&defrag->lock);
     {
-        status = dict_new();
         gf_defrag_status_get(conf, status);
         if (ctx && ctx->notify)
             ctx->notify(GF_EN_DEFRAG_STATUS, status);
@@ -4878,9 +4515,6 @@ gf_defrag_get_estimates_based_on_size(dht_conf_t *conf)
     uint64_t total_processed = 0;
     uint64_t tmp_count = 0;
     uint64_t time_to_complete = 0;
-    struct timeval now = {
-        0,
-    };
     double elapsed = 0;
 
     defrag = conf->defrag;
@@ -4888,8 +4522,7 @@ gf_defrag_get_estimates_based_on_size(dht_conf_t *conf)
     if (!g_totalsize)
         goto out;
 
-    gettimeofday(&now, NULL);
-    elapsed = now.tv_sec - defrag->start_time.tv_sec;
+    elapsed = gf_time() - defrag->start_time;
 
     /* Don't calculate the estimates for the first 10 minutes.
      * It is unlikely to be accurate and estimates are not required
@@ -4939,13 +4572,8 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
     uint64_t lookup = 0;
     uint64_t failures = 0;
     uint64_t skipped = 0;
-    uint64_t promoted = 0;
-    uint64_t demoted = 0;
     char *status = "";
     double elapsed = 0;
-    struct timeval end = {
-        0,
-    };
     uint64_t time_to_complete = 0;
     uint64_t time_left = 0;
     gf_defrag_info_t *defrag = conf->defrag;
@@ -4962,17 +4590,12 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
     lookup = defrag->num_files_lookedup;
     failures = defrag->total_failures;
     skipped = defrag->skipped;
-    promoted = defrag->total_files_promoted;
-    demoted = defrag->total_files_demoted;
 
-    gettimeofday(&end, NULL);
-
-    elapsed = end.tv_sec - defrag->start_time.tv_sec;
+    elapsed = gf_time() - defrag->start_time;
 
     /* The rebalance is still in progress */
 
-    if ((defrag->cmd != GF_DEFRAG_CMD_START_TIER) &&
-        (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED)) {
+    if (defrag->defrag_status == GF_DEFRAG_STATUS_STARTED) {
         time_to_complete = gf_defrag_get_estimates_based_on_size(conf);
 
         if (time_to_complete && (time_to_complete > elapsed))
@@ -4987,14 +4610,6 @@ gf_defrag_status_get(dht_conf_t *conf, dict_t *dict)
     if (!dict)
         goto log;
 
-    ret = dict_set_uint64(dict, "promoted", promoted);
-    if (ret)
-        gf_log(THIS->name, GF_LOG_WARNING, "failed to set promoted count");
-
-    ret = dict_set_uint64(dict, "demoted", demoted);
-    if (ret)
-        gf_log(THIS->name, GF_LOG_WARNING, "failed to set demoted count");
-
     ret = dict_set_uint64(dict, "files", files);
     if (ret)
         gf_log(THIS->name, GF_LOG_WARNING, "failed to set file count");
@@ -5060,159 +4675,6 @@ out:
     return 0;
 }
 
-void
-gf_defrag_set_pause_state(gf_tier_conf_t *tier_conf, tier_pause_state_t state)
-{
-    pthread_mutex_lock(&tier_conf->pause_mutex);
-    tier_conf->pause_state = state;
-    pthread_mutex_unlock(&tier_conf->pause_mutex);
-}
-
-tier_pause_state_t
-gf_defrag_get_pause_state(gf_tier_conf_t *tier_conf)
-{
-    int state;
-
-    pthread_mutex_lock(&tier_conf->pause_mutex);
-    state = tier_conf->pause_state;
-    pthread_mutex_unlock(&tier_conf->pause_mutex);
-
-    return state;
-}
-
-tier_pause_state_t
-gf_defrag_check_pause_tier(gf_tier_conf_t *tier_conf)
-{
-    int woke = 0;
-    int state = -1;
-
-    pthread_mutex_lock(&tier_conf->pause_mutex);
-
-    if (tier_conf->pause_state == TIER_RUNNING)
-        goto out;
-
-    if (tier_conf->pause_state == TIER_PAUSED)
-        goto out;
-
-    if (tier_conf->promote_in_progress || tier_conf->demote_in_progress)
-        goto out;
-
-    tier_conf->pause_state = TIER_PAUSED;
-
-    if (tier_conf->pause_synctask) {
-        synctask_wake(tier_conf->pause_synctask);
-        tier_conf->pause_synctask = 0;
-        woke = 1;
-    }
-
-    gf_msg("tier", GF_LOG_DEBUG, 0, DHT_MSG_TIER_PAUSED, "woken %d", woke);
-
-    gf_event(EVENT_TIER_PAUSE, "vol=%s", tier_conf->volname);
-out:
-    state = tier_conf->pause_state;
-
-    pthread_mutex_unlock(&tier_conf->pause_mutex);
-
-    return state;
-}
-
-void
-gf_defrag_pause_tier_timeout(void *data)
-{
-    xlator_t *this = NULL;
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-
-    this = (xlator_t *)data;
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-
-    conf = this->private;
-    GF_VALIDATE_OR_GOTO(this->name, conf, out);
-
-    defrag = conf->defrag;
-    GF_VALIDATE_OR_GOTO(this->name, defrag, out);
-
-    gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_TIER_PAUSED,
-           "Request pause timer timeout");
-
-    gf_defrag_check_pause_tier(&defrag->tier_conf);
-
-out:
-    return;
-}
-
-int
-gf_defrag_pause_tier(xlator_t *this, gf_defrag_info_t *defrag)
-{
-    int ret = 0;
-    struct timespec delta = {
-        0,
-    };
-    int delay = 2;
-
-    if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED)
-        goto out;
-
-    /*
-     * Set flag requesting to pause tiering. Wait 'delay' seconds for
-     * tiering to actually stop as indicated by the pause state
-     * before returning success or failure.
-     */
-    gf_defrag_set_pause_state(&defrag->tier_conf, TIER_REQUEST_PAUSE);
-
-    /*
-     * If migration is not underway, can pause immediately.
-     */
-    gf_defrag_check_pause_tier(&defrag->tier_conf);
-    if (gf_defrag_get_pause_state(&defrag->tier_conf) == TIER_PAUSED)
-        goto out;
-
-    gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_TIER_PAUSED,
-           "Request pause tier");
-
-    defrag->tier_conf.pause_synctask = synctask_get();
-    delta.tv_sec = delay;
-    delta.tv_nsec = 0;
-    defrag->tier_conf.pause_timer = gf_timer_call_after(
-        this->ctx, delta, gf_defrag_pause_tier_timeout, this);
-
-    synctask_yield(defrag->tier_conf.pause_synctask);
-
-    if (gf_defrag_get_pause_state(&defrag->tier_conf) == TIER_PAUSED)
-        goto out;
-
-    gf_defrag_set_pause_state(&defrag->tier_conf, TIER_RUNNING);
-
-    ret = -1;
-out:
-
-    gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_TIER_PAUSED,
-           "Pause tiering ret=%d", ret);
-
-    return ret;
-}
-
-int
-gf_defrag_resume_tier(xlator_t *this, gf_defrag_info_t *defrag)
-{
-    gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_TIER_RESUME,
-           "Pause end. Resume tiering");
-
-    gf_defrag_set_pause_state(&defrag->tier_conf, TIER_RUNNING);
-
-    gf_event(EVENT_TIER_RESUME, "vol=%s", defrag->tier_conf.volname);
-
-    return 0;
-}
-
-int
-gf_defrag_start_detach_tier(gf_defrag_info_t *defrag)
-{
-    defrag->cmd = GF_DEFRAG_CMD_START_DETACH_TIER;
-
-    return 0;
-}
-
 int
 gf_defrag_stop(dht_conf_t *conf, gf_defrag_status_t status, dict_t *output)
 {
diff --git a/xlators/cluster/dht/src/dht-rename.c b/xlators/cluster/dht/src/dht-rename.c
index af342bdbe21..d9dbf50492f 100644
--- a/xlators/cluster/dht/src/dht-rename.c
+++ b/xlators/cluster/dht/src/dht-rename.c
@@ -11,11 +11,9 @@
 /* TODO: link(oldpath, newpath) fails if newpath already exists. DHT should
  *       delete the newpath if it gets EEXISTS from link() call.
  */
-#include "glusterfs.h"
-#include "xlator.h"
 #include "dht-common.h"
 #include "dht-lock.h"
-#include "defaults.h"
+#include <glusterfs/defaults.h>
 
 int
 dht_rename_unlock(call_frame_t *frame, xlator_t *this);
@@ -505,6 +503,8 @@ dht_order_rename_lock(call_frame_t *frame, loc_t **loc, xlator_t **subvol)
             uuid_utoa_r(local->loc.pargfid, src);
         else if (local->loc.parent)
             uuid_utoa_r(local->loc.parent->gfid, src);
+        else
+            src[0] = '\0';
 
         strcat(src, local->loc.name);
 
@@ -520,6 +520,8 @@ dht_order_rename_lock(call_frame_t *frame, loc_t **loc, xlator_t **subvol)
             uuid_utoa_r(local->loc2.pargfid, dst);
         else if (local->loc2.parent)
             uuid_utoa_r(local->loc2.parent->gfid, dst);
+        else
+            dst[0] = '\0';
 
         strcat(dst, local->loc2.name);
         ret = strcmp(src, dst);
@@ -1009,9 +1011,11 @@ dht_rename_links_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 {
     xlator_t *prev = NULL;
     dht_local_t *local = NULL;
+    call_frame_t *main_frame = NULL;
 
     prev = cookie;
     local = frame->local;
+    main_frame = local->main_frame;
 
     /* TODO: Handle this case in lookup-optimize */
     if (op_ret == -1) {
@@ -1024,7 +1028,8 @@ dht_rename_links_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         dht_linkfile_attr_heal(frame, this);
     }
 
-    dht_rename_unlink(frame, this);
+    dht_rename_unlink(main_frame, this);
+    DHT_STACK_DESTROY(frame);
     return 0;
 }
 
@@ -1040,7 +1045,8 @@ dht_rename_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
     xlator_t *src_cached = NULL;
     xlator_t *dst_hashed = NULL;
     xlator_t *dst_cached = NULL;
-    loc_t link_loc = {0};
+    call_frame_t *link_frame = NULL;
+    dht_local_t *link_local = NULL;
 
     local = frame->local;
     prev = cookie;
@@ -1110,18 +1116,36 @@ dht_rename_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     /* Create the linkto file for the dst file */
     if ((src_cached == dst_cached) && (dst_hashed != dst_cached)) {
-        loc_copy(&link_loc, &local->loc2);
-        if (link_loc.inode)
-            inode_unref(link_loc.inode);
-        link_loc.inode = inode_ref(local->loc.inode);
-        gf_uuid_copy(local->gfid, local->loc.inode->gfid);
-        gf_uuid_copy(link_loc.gfid, local->loc.inode->gfid);
-
-        dht_linkfile_create(frame, dht_rename_links_create_cbk, this,
-                            src_cached, dst_hashed, &link_loc);
+        link_frame = copy_frame(frame);
+        if (!link_frame) {
+            goto unlink;
+        }
+
+        /* fop value sent as maxvalue because it is not used
+         * anywhere in this case */
+        link_local = dht_local_init(link_frame, &local->loc2, NULL,
+                                    GF_FOP_MAXVALUE);
+        if (!link_local) {
+            goto unlink;
+        }
+
+        if (link_local->loc.inode)
+            inode_unref(link_local->loc.inode);
+        link_local->loc.inode = inode_ref(local->loc.inode);
+        link_local->main_frame = frame;
+        link_local->stbuf = local->stbuf;
+        gf_uuid_copy(link_local->gfid, local->loc.inode->gfid);
+
+        dht_linkfile_create(link_frame, dht_rename_links_create_cbk, this,
+                            src_cached, dst_hashed, &link_local->loc);
         return 0;
     }
 
+unlink:
+
+    if (link_frame) {
+        DHT_STACK_DESTROY(link_frame);
+    }
     dht_rename_unlink(frame, this);
     return 0;
 
diff --git a/xlators/cluster/dht/src/dht-selfheal.c b/xlators/cluster/dht/src/dht-selfheal.c
index 0e57eab5f7f..3e24065227c 100644
--- a/xlators/cluster/dht/src/dht-selfheal.c
+++ b/xlators/cluster/dht/src/dht-selfheal.c
@@ -8,12 +8,7 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "glusterfs.h"
-#include "xlator.h"
-#include "dht-common.h"
-#include "dht-messages.h"
 #include "dht-lock.h"
-#include "glusterfs-acl.h"
 
 #define DHT_SET_LAYOUT_RANGE(layout, i, srt, chunk, path)                      \
     do {                                                                       \
@@ -22,7 +17,7 @@
         layout->list[i].commit_hash = layout->commit_hash;                     \
                                                                                \
         gf_msg_trace(this->name, 0,                                            \
-                     "gave fix: %u - %u, with commit-hash %u"                  \
+                     "gave fix: 0x%x - 0x%x, with commit-hash 0x%x"            \
                      " on %s for %s",                                          \
                      layout->list[i].start, layout->list[i].stop,              \
                      layout->list[i].commit_hash,                              \
@@ -38,7 +33,7 @@
         }                                                                      \
     } while (0)
 
-int
+static int
 dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
                          gf_boolean_t newdir, dht_selfheal_layout_t healer,
                          dht_need_heal_t should_heal);
@@ -149,8 +144,8 @@ dht_refresh_layout_done(call_frame_t *frame)
 
     ret = dht_layout_sort(refreshed);
     if (ret == -1) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_SORT_FAILED,
-               "sorting the layout failed");
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_LAYOUT_SORT_FAILED, NULL);
         goto err;
     }
 
@@ -206,10 +201,9 @@ dht_refresh_layout_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         if (op_ret == -1) {
             gf_uuid_unparse(local->loc.gfid, gfid);
             local->op_errno = op_errno;
-            gf_msg(this->name, GF_LOG_ERROR, op_errno,
-                   DHT_MSG_FILE_LOOKUP_FAILED,
-                   "lookup of %s on %s returned error, gfid: %s",
-                   local->loc.path, prev->name, gfid);
+            gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                    DHT_MSG_FILE_LOOKUP_FAILED, "path=%s", local->loc.path,
+                    "name=%s", prev->name, "gfid=%s", gfid, NULL);
 
             goto unlock;
         }
@@ -270,9 +264,8 @@ dht_refresh_layout(call_frame_t *frame)
                                                       conf->subvolume_cnt);
     if (!local->selfheal.refreshed_layout) {
         gf_uuid_unparse(local->loc.gfid, gfid);
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-               "mem allocation for layout failed, path:%s gfid:%s",
-               local->loc.path, gfid);
+        gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MEM_ALLOC_FAILED,
+                "path=%s", local->loc.path, "gfid=%s", gfid, NULL);
         goto out;
     }
 
@@ -284,9 +277,8 @@ dht_refresh_layout(call_frame_t *frame)
         gf_uuid_unparse(local->loc.gfid, gfid);
         local->xattr_req = dict_new();
         if (local->xattr_req == NULL) {
-            gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                   "dict mem allocation failed, path:%s gfid:%s",
-                   local->loc.path, gfid);
+            gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
+                    "path=%s", local->loc.path, "gfid=%s", gfid, NULL);
             goto out;
         }
     }
@@ -294,9 +286,9 @@ dht_refresh_layout(call_frame_t *frame)
     if (dict_get(local->xattr_req, conf->xattr_name) == 0) {
         ret = dict_set_uint32(local->xattr_req, conf->xattr_name, 4 * 4);
         if (ret)
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-                   "%s: Failed to set dictionary value:key = %s",
-                   local->loc.path, conf->xattr_name);
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                    "path=%s", local->loc.path, "key=%s", conf->xattr_name,
+                    NULL);
     }
 
     for (i = 0; i < call_cnt; i++) {
@@ -529,7 +521,7 @@ out:
     return fixit;
 }
 
-int
+static int
 dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
                          gf_boolean_t newdir, dht_selfheal_layout_t healer,
                          dht_need_heal_t should_heal)
@@ -561,10 +553,8 @@ dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
         lk_array = GF_CALLOC(count, sizeof(*lk_array), gf_common_mt_char);
         if (lk_array == NULL) {
             gf_uuid_unparse(local->stbuf.ia_gfid, gfid);
-            gf_msg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                   "mem allocation failed for "
-                   "lk_array, gfid:%s path: %s",
-                   gfid, local->loc.path);
+            gf_smsg("dht", GF_LOG_ERROR, ENOMEM, DHT_MSG_MEM_ALLOC_FAILED,
+                    "lk_array-gfid=%s", gfid, "path=%s", local->loc.path, NULL);
             goto err;
         }
 
@@ -574,10 +564,9 @@ dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
                 DHT_LAYOUT_HEAL_DOMAIN, NULL, FAIL_ON_ANY_ERROR);
             if (lk_array[i] == NULL) {
                 gf_uuid_unparse(local->stbuf.ia_gfid, gfid);
-                gf_msg(THIS->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                       "mem allocation "
-                       "failed for lk_array, gfid:%s path:%s",
-                       gfid, local->loc.path);
+                gf_smsg(THIS->name, GF_LOG_ERROR, ENOMEM,
+                        DHT_MSG_MEM_ALLOC_FAILED, "lk_array-gfid=%s", gfid,
+                        "path=%s", local->loc.path, NULL);
                 goto err;
             }
         }
@@ -586,10 +575,8 @@ dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
         lk_array = GF_CALLOC(count, sizeof(*lk_array), gf_common_mt_char);
         if (lk_array == NULL) {
             gf_uuid_unparse(local->stbuf.ia_gfid, gfid);
-            gf_msg(THIS->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                   "mem allocation failed for "
-                   "lk_array, gfid:%s path:%s",
-                   gfid, local->loc.path);
+            gf_smsg(THIS->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MEM_ALLOC_FAILED,
+                    "lk_array-gfid=%s", gfid, "path=%s", local->loc.path, NULL);
             goto err;
         }
 
@@ -598,10 +585,8 @@ dht_selfheal_layout_lock(call_frame_t *frame, dht_layout_t *layout,
                                    NULL, FAIL_ON_ANY_ERROR);
         if (lk_array[0] == NULL) {
             gf_uuid_unparse(local->stbuf.ia_gfid, gfid);
-            gf_msg(THIS->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                   "mem allocation failed for "
-                   "lk_array, gfid:%s path:%s",
-                   gfid, local->loc.path);
+            gf_smsg(THIS->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MEM_ALLOC_FAILED,
+                    "lk_array-gfid=%s", gfid, "path=%s", local->loc.path, NULL);
             goto err;
         }
     }
@@ -627,7 +612,7 @@ err:
     return -1;
 }
 
-int
+static int
 dht_selfheal_dir_xattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                            int op_ret, int op_errno, dict_t *xdata)
 {
@@ -649,10 +634,9 @@ dht_selfheal_dir_xattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
         err = 0;
     } else {
         gf_uuid_unparse(local->loc.gfid, gfid);
-        gf_msg(this->name, GF_LOG_ERROR, op_errno,
-               DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-               "layout setxattr failed on %s, path:%s gfid:%s", subvol->name,
-               local->loc.path, gfid);
+        gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                DHT_MSG_DIR_SELFHEAL_XATTR_FAILED, "name=%s", subvol->name,
+                "path=%s", local->loc.path, "gfid=%s", gfid, NULL);
         err = op_errno;
     }
 
@@ -699,7 +683,7 @@ dht_set_user_xattr(dict_t *dict, char *k, data_t *v, void *data)
     return ret;
 }
 
-int
+static int
 dht_selfheal_dir_xattr_persubvol(call_frame_t *frame, loc_t *loc,
                                  dht_layout_t *layout, int i,
                                  xlator_t *req_subvol)
@@ -741,19 +725,17 @@ dht_selfheal_dir_xattr_persubvol(call_frame_t *frame, loc_t *loc,
 
     ret = dict_set_str(xdata, GLUSTERFS_INTERNAL_FOP_KEY, "yes");
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "%s: Failed to set dictionary value: key = %s,"
-               " gfid = %s",
-               loc->path, GLUSTERFS_INTERNAL_FOP_KEY, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                "path=%s", loc->path, "key=%s", GLUSTERFS_INTERNAL_FOP_KEY,
+                "gfid=%s", gfid, NULL);
         goto err;
     }
 
     ret = dict_set_int8(xdata, DHT_IATT_IN_XDATA_KEY, 1);
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "%s: Failed to set dictionary value: key = %s,"
-               " gfid = %s",
-               loc->path, DHT_IATT_IN_XDATA_KEY, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                "path=%s", loc->path, "key=%s", DHT_IATT_IN_XDATA_KEY,
+                "gfid=%s", gfid, NULL);
         goto err;
     }
 
@@ -761,27 +743,27 @@ dht_selfheal_dir_xattr_persubvol(call_frame_t *frame, loc_t *loc,
 
     ret = dht_disk_layout_extract(this, layout, i, &disk_layout);
     if (ret == -1) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-               "Directory self heal xattr failed:"
-               " %s: (subvol %s) Failed to extract disk layout,"
-               " gfid = %s",
-               loc->path, subvol->name, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
+                "extract-disk-layout-failed, path=%s", loc->path, "subvol=%s",
+                subvol->name, "gfid=%s", gfid, NULL);
         goto err;
     }
 
     ret = dict_set_bin(xattr, conf->xattr_name, disk_layout, 4 * 4);
     if (ret == -1) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-               "Directory self heal xattr failed:"
-               "%s: (subvol %s) Failed to set xattr dictionary,"
-               " gfid = %s",
-               loc->path, subvol->name, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0,
+                DHT_MSG_DIR_SELFHEAL_XATTR_FAILED, "path=%s", loc->path,
+                "subvol=%s", subvol->name,
+                "set-xattr-dictionary-failed"
+                "gfid=%s",
+                gfid, NULL);
         goto err;
     }
     disk_layout = NULL;
 
     gf_msg_trace(this->name, 0,
-                 "setting hash range %u - %u (type %d) on subvolume %s"
+                 "setting hash range 0x%x - 0x%x (type %d) on subvolume %s"
                  " for %s",
                  layout->list[i].start, layout->list[i].stop, layout->type,
                  subvol->name, loc->path);
@@ -791,20 +773,17 @@ dht_selfheal_dir_xattr_persubvol(call_frame_t *frame, loc_t *loc,
         if (data) {
             ret = dict_add(xattr, QUOTA_LIMIT_KEY, data);
             if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-                       "%s: Failed to set dictionary value:"
-                       " key = %s",
-                       loc->path, QUOTA_LIMIT_KEY);
+                gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+                        "path=%s", loc->path, "key=%s", QUOTA_LIMIT_KEY, NULL);
             }
         }
         data = dict_get(local->xattr, QUOTA_LIMIT_OBJECTS_KEY);
         if (data) {
             ret = dict_add(xattr, QUOTA_LIMIT_OBJECTS_KEY, data);
             if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-                       "%s: Failed to set dictionary value:"
-                       " key = %s",
-                       loc->path, QUOTA_LIMIT_OBJECTS_KEY);
+                gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+                        "path=%s", loc->path, "key=%s", QUOTA_LIMIT_OBJECTS_KEY,
+                        NULL);
             }
         }
     }
@@ -833,7 +812,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_fix_dir_xattr(call_frame_t *frame, loc_t *loc, dht_layout_t *layout)
 {
     dht_local_t *local = NULL;
@@ -882,7 +861,7 @@ out:
     return 0;
 }
 
-int
+static int
 dht_selfheal_dir_xattr(call_frame_t *frame, loc_t *loc, dht_layout_t *layout)
 {
     dht_local_t *local = NULL;
@@ -942,9 +921,8 @@ dht_selfheal_dir_xattr(call_frame_t *frame, loc_t *loc, dht_layout_t *layout)
     dummy = dht_layout_new(this, 1);
     if (!dummy) {
         gf_uuid_unparse(loc->gfid, gfid);
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-               "failed to allocate dummy layout, path:%s gfid:%s", loc->path,
-               gfid);
+        gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_DUMMY_ALLOC_FAILED,
+                "path=%s", loc->path, "gfid=%s", gfid, NULL);
         goto out;
     }
     for (i = 0; i < conf->subvolume_cnt && missing_xattr; i++) {
@@ -960,38 +938,6 @@ out:
     return 0;
 }
 
-gf_boolean_t
-dht_is_subvol_part_of_layout(dht_layout_t *layout, xlator_t *xlator)
-{
-    int i = 0;
-    gf_boolean_t ret = _gf_false;
-
-    for (i = 0; i < layout->cnt; i++) {
-        if (!strcmp(layout->list[i].xlator->name, xlator->name)) {
-            ret = _gf_true;
-            break;
-        }
-    }
-
-    return ret;
-}
-
-int
-dht_layout_index_from_conf(dht_layout_t *layout, xlator_t *xlator)
-{
-    int i = -1;
-    int j = 0;
-
-    for (j = 0; j < layout->cnt; j++) {
-        if (!strcmp(layout->list[j].xlator->name, xlator->name)) {
-            i = j;
-            break;
-        }
-    }
-
-    return i;
-}
-
 int
 dht_selfheal_dir_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                              int op_ret, int op_errno, struct iatt *statpre,
@@ -1033,18 +979,27 @@ dht_selfheal_dir_setattr(call_frame_t *frame, loc_t *loc, struct iatt *stbuf,
     int missing_attr = 0;
     int i = 0, ret = -1;
     dht_local_t *local = NULL;
+    dht_conf_t *conf = NULL;
     xlator_t *this = NULL;
     int cnt = 0;
 
     local = frame->local;
     this = frame->this;
+    conf = this->private;
+
+    /* We need to heal the attrs if:
+     * 1. Any directories were missing - the newly created dirs will need
+     *    to have the correct attrs set
+     * 2. An existing dir does not have the correct permissions -they may
+     *    have been changed when a brick was down.
+     */
 
     for (i = 0; i < layout->cnt; i++) {
         if (layout->list[i].err == -1)
             missing_attr++;
     }
 
-    if (missing_attr == 0) {
+    if ((missing_attr == 0) && (local->need_attrheal == 0)) {
         if (!local->heal_layout) {
             gf_msg_trace(this->name, 0, "Skip heal layout for %s gfid = %s ",
                          loc->path, uuid_utoa(loc->gfid));
@@ -1062,25 +1017,18 @@ dht_selfheal_dir_setattr(call_frame_t *frame, loc_t *loc, struct iatt *stbuf,
         return 0;
     }
 
-    local->call_cnt = missing_attr;
-    cnt = layout->cnt;
+    cnt = local->call_cnt = conf->subvolume_cnt;
 
     for (i = 0; i < cnt; i++) {
-        if (layout->list[i].err == -1) {
-            gf_msg_trace(this->name, 0, "%s: setattr on subvol %s, gfid = %s",
-                         loc->path, layout->list[i].xlator->name,
-                         uuid_utoa(loc->gfid));
-
-            STACK_WIND(
-                frame, dht_selfheal_dir_setattr_cbk, layout->list[i].xlator,
-                layout->list[i].xlator->fops->setattr, loc, stbuf, valid, NULL);
-        }
+        STACK_WIND(frame, dht_selfheal_dir_setattr_cbk, layout->list[i].xlator,
+                   layout->list[i].xlator->fops->setattr, loc, stbuf, valid,
+                   NULL);
     }
 
     return 0;
 }
 
-int
+static int
 dht_selfheal_dir_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                            int op_ret, int op_errno, inode_t *inode,
                            struct iatt *stbuf, struct iatt *preparent,
@@ -1110,11 +1058,10 @@ dht_selfheal_dir_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 
     if (op_ret) {
         gf_uuid_unparse(local->loc.gfid, gfid);
-        gf_msg(this->name,
-               ((op_errno == EEXIST) ? GF_LOG_DEBUG : GF_LOG_WARNING), op_errno,
-               DHT_MSG_DIR_SELFHEAL_FAILED,
-               "Directory selfheal failed: path = %s, gfid = %s",
-               local->loc.path, gfid);
+        gf_smsg(this->name,
+                ((op_errno == EEXIST) ? GF_LOG_DEBUG : GF_LOG_WARNING),
+                op_errno, DHT_MSG_DIR_SELFHEAL_FAILED, "path=%s",
+                local->loc.path, "gfid=%s", gfid, NULL);
         goto out;
     }
     dht_iatt_merge(this, &local->preparent, preparent);
@@ -1133,89 +1080,7 @@ out:
     return 0;
 }
 
-void
-dht_selfheal_dir_mkdir_setacl(dict_t *xattr, dict_t *dict)
-{
-    data_t *acl_default = NULL;
-    data_t *acl_access = NULL;
-    xlator_t *this = NULL;
-    int ret = -1;
-
-    GF_ASSERT(xattr);
-    GF_ASSERT(dict);
-
-    this = THIS;
-    GF_ASSERT(this);
-
-    acl_default = dict_get(xattr, POSIX_ACL_DEFAULT_XATTR);
-
-    if (!acl_default) {
-        gf_msg_debug(this->name, 0, "ACL_DEFAULT xattr not present");
-        goto cont;
-    }
-    ret = dict_set(dict, POSIX_ACL_DEFAULT_XATTR, acl_default);
-    if (ret)
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value.key = %s",
-               POSIX_ACL_DEFAULT_XATTR);
-cont:
-    acl_access = dict_get(xattr, POSIX_ACL_ACCESS_XATTR);
-    if (!acl_access) {
-        gf_msg_debug(this->name, 0, "ACL_ACCESS xattr not present");
-        goto out;
-    }
-    ret = dict_set(dict, POSIX_ACL_ACCESS_XATTR, acl_access);
-    if (ret)
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value.key = %s",
-               POSIX_ACL_ACCESS_XATTR);
-
-out:
-    return;
-}
-
-void
-dht_selfheal_dir_mkdir_setquota(dict_t *src, dict_t *dst)
-{
-    data_t *quota_limit_key = NULL;
-    data_t *quota_limit_obj_key = NULL;
-    xlator_t *this = NULL;
-    int ret = -1;
-
-    GF_ASSERT(src);
-    GF_ASSERT(dst);
-
-    this = THIS;
-    GF_ASSERT(this);
-
-    quota_limit_key = dict_get(src, QUOTA_LIMIT_KEY);
-    if (!quota_limit_key) {
-        gf_msg_debug(this->name, 0, "QUOTA_LIMIT_KEY xattr not present");
-        goto cont;
-    }
-    ret = dict_set(dst, QUOTA_LIMIT_KEY, quota_limit_key);
-    if (ret)
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value.key = %s", QUOTA_LIMIT_KEY);
-
-cont:
-    quota_limit_obj_key = dict_get(src, QUOTA_LIMIT_OBJECTS_KEY);
-    if (!quota_limit_obj_key) {
-        gf_msg_debug(this->name, 0,
-                     "QUOTA_LIMIT_OBJECTS_KEY xattr not present");
-        goto out;
-    }
-    ret = dict_set(dst, QUOTA_LIMIT_OBJECTS_KEY, quota_limit_obj_key);
-    if (ret)
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value.key = %s",
-               QUOTA_LIMIT_OBJECTS_KEY);
-
-out:
-    return;
-}
-
-int
+static int
 dht_selfheal_dir_mkdir_lookup_done(call_frame_t *frame, xlator_t *this)
 {
     dht_local_t *local = NULL;
@@ -1239,10 +1104,8 @@ dht_selfheal_dir_mkdir_lookup_done(call_frame_t *frame, xlator_t *this)
 
         ret = dict_set_gfuuid(dict, "gfid-req", local->gfid, true);
         if (ret)
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-                   "%s: Failed to set dictionary value:"
-                   " key = gfid-req",
-                   loc->path);
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                    "path=%s", loc->path, "key=gfid-req", NULL);
     } else if (local->params) {
         /* Send the dictionary from higher layers directly */
 
@@ -1254,18 +1117,15 @@ dht_selfheal_dir_mkdir_lookup_done(call_frame_t *frame, xlator_t *this)
     dht_dir_set_heal_xattr(this, local, dict, local->xattr, NULL, NULL);
 
     if (!dict) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "dict is NULL, need to make sure gfids are same");
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_IS_NULL, NULL);
         dict = dict_new();
         if (!dict)
             return -1;
     }
     ret = dict_set_flag(dict, GF_INTERNAL_CTX_KEY, GF_DHT_HEAL_DIR);
     if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value for"
-               " key = %s at path: %s",
-               GF_INTERNAL_CTX_KEY, loc->path);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED, "key=%s",
+                GF_INTERNAL_CTX_KEY, "path=%s", loc->path, NULL);
         /* We can still continue. As heal can still happen
          * unless quota limits have reached for the dir.
          */
@@ -1297,7 +1157,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_selfheal_dir_mkdir_lookup_cbk(call_frame_t *frame, void *cookie,
                                   xlator_t *this, int op_ret, int op_errno,
                                   inode_t *inode, struct iatt *stbuf,
@@ -1308,45 +1168,54 @@ dht_selfheal_dir_mkdir_lookup_cbk(call_frame_t *frame, void *cookie,
     int this_call_cnt = 0;
     int missing_dirs = 0;
     dht_layout_t *layout = NULL;
-    dht_conf_t *conf = 0;
+    xlator_t *prev = 0;
     loc_t *loc = NULL;
-    int check_mds = 0;
-    int errst = 0;
-    int32_t mds_xattr_val[1] = {0};
     char gfid_local[GF_UUID_BUF_SIZE] = {0};
+    int index = -1;
 
     VALIDATE_OR_GOTO(this->private, err);
 
     local = frame->local;
     layout = local->layout;
     loc = &local->loc;
-    conf = this->private;
+    prev = cookie;
 
-    if (local->gfid)
+    if (!gf_uuid_is_null(local->gfid))
         gf_uuid_unparse(local->gfid, gfid_local);
 
-    this_call_cnt = dht_frame_return(frame);
-
     LOCK(&frame->lock);
     {
+        index = dht_layout_index_for_subvol(layout, prev);
         if ((op_ret < 0) && (op_errno == ENOENT || op_errno == ESTALE)) {
             local->selfheal.hole_cnt = !local->selfheal.hole_cnt
                                            ? 1
                                            : local->selfheal.hole_cnt + 1;
+            /* the status might have changed. Update the layout with the
+             * new status
+             */
+            if (index >= 0) {
+                layout->list[index].err = op_errno;
+            }
         }
 
         if (!op_ret) {
             dht_iatt_merge(this, &local->stbuf, stbuf);
-        }
-        check_mds = dht_dict_get_array(xattr, conf->mds_xattr_key,
-                                       mds_xattr_val, 1, &errst);
-        if (dict_get(xattr, conf->mds_xattr_key) && check_mds && !errst) {
-            dict_unref(local->xattr);
-            local->xattr = dict_ref(xattr);
+            if (prev == local->mds_subvol) {
+                dict_unref(local->xattr);
+                local->xattr = dict_ref(xattr);
+            }
+            /* the status might have changed. Update the layout with the
+             * new status
+             */
+            if (index >= 0) {
+                layout->list[index].err = -1;
+            }
         }
     }
     UNLOCK(&frame->lock);
 
+    this_call_cnt = dht_frame_return(frame);
+
     if (is_last_call(this_call_cnt)) {
         if (local->selfheal.hole_cnt == layout->cnt) {
             gf_msg_debug(this->name, op_errno,
@@ -1382,7 +1251,7 @@ err:
     return 0;
 }
 
-int
+static int
 dht_selfheal_dir_mkdir_lock_cbk(call_frame_t *frame, void *cookie,
                                 xlator_t *this, int32_t op_ret,
                                 int32_t op_errno, dict_t *xdata)
@@ -1402,19 +1271,14 @@ dht_selfheal_dir_mkdir_lock_cbk(call_frame_t *frame, void *cookie,
     local->call_cnt = conf->subvolume_cnt;
 
     if (op_ret < 0) {
-        /* We get this error when the directory entry was not created
-         * on a newky attached tier subvol. Hence proceed and do mkdir
-         * on the tier subvol.
-         */
         if (op_errno == EINVAL) {
             local->call_cnt = 1;
             dht_selfheal_dir_mkdir_lookup_done(frame, this);
             return 0;
         }
 
-        gf_msg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_ENTRYLK_ERROR,
-               "acquiring entrylk after inodelk failed for %s",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_WARNING, op_errno, DHT_MSG_ENTRYLK_ERROR,
+                "path=%s", local->loc.path, NULL);
 
         local->op_errno = op_errno;
         goto err;
@@ -1428,10 +1292,8 @@ dht_selfheal_dir_mkdir_lock_cbk(call_frame_t *frame, void *cookie,
 
     ret = dict_set_int32(local->xattr_req, "list-xattr", 1);
     if (ret)
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary key list-xattr value "
-               " for path %s ",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED, "path=%s",
+                local->loc.path, NULL);
 
     for (i = 0; i < conf->subvolume_cnt; i++) {
         if (mds_subvol && conf->subvolumes[i] == mds_subvol) {
@@ -1454,18 +1316,21 @@ err:
     return 0;
 }
 
-int
+static int
 dht_selfheal_dir_mkdir(call_frame_t *frame, loc_t *loc, dht_layout_t *layout,
                        int force)
 {
     int missing_dirs = 0;
     int i = 0;
+    int op_errno = 0;
     int ret = -1;
     dht_local_t *local = NULL;
     xlator_t *this = NULL;
+    dht_conf_t *conf = NULL;
 
     local = frame->local;
     this = frame->this;
+    conf = this->private;
 
     local->selfheal.force_mkdir = force;
     local->selfheal.hole_cnt = 0;
@@ -1476,16 +1341,18 @@ dht_selfheal_dir_mkdir(call_frame_t *frame, loc_t *loc, dht_layout_t *layout,
     }
 
     if (missing_dirs == 0) {
+        /* We don't need to create any directories. Proceed to heal the
+         * attrs and xattrs
+         */
         if (!__is_root_gfid(local->stbuf.ia_gfid)) {
             if (local->need_xattr_heal) {
                 local->need_xattr_heal = 0;
-                ret = dht_dir_xattr_heal(this, local);
-                if (ret)
-                    gf_msg(this->name, GF_LOG_ERROR, ret,
-                           DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                           "xattr heal failed for "
-                           "directory  %s gfid %s ",
-                           local->loc.path, local->gfid);
+                ret = dht_dir_xattr_heal(this, local, &op_errno);
+                if (ret) {
+                    gf_smsg(this->name, GF_LOG_ERROR, op_errno,
+                            DHT_MSG_DIR_XATTR_HEAL_FAILED, "path=%s",
+                            local->loc.path, "gfid=%s", local->gfid, NULL);
+                }
             } else {
                 if (!gf_uuid_is_null(local->gfid))
                     gf_uuid_copy(loc->gfid, local->gfid);
@@ -1494,28 +1361,53 @@ dht_selfheal_dir_mkdir(call_frame_t *frame, loc_t *loc, dht_layout_t *layout,
                 if (!ret)
                     return 0;
 
-                gf_msg(this->name, GF_LOG_INFO, 0,
-                       DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                       "Failed to set mds xattr "
-                       "for directory  %s gfid %s ",
-                       local->loc.path, local->gfid);
+                gf_smsg(this->name, GF_LOG_INFO, 0, DHT_MSG_SET_XATTR_FAILED,
+                        "path=%s", local->loc.path, "gfid=%s", local->gfid,
+                        NULL);
             }
         }
         dht_selfheal_dir_setattr(frame, loc, &local->stbuf, 0xffffffff, layout);
         return 0;
     }
 
-    if (local->hashed_subvol == NULL)
-        local->hashed_subvol = dht_subvol_get_hashed(this, loc);
+    /* MDS xattr is populated only while DHT is having more than one
+     subvol.In case of graph switch while adding more dht subvols need to
+     consider hash subvol as a MDS to avoid MDS check failure at the time
+     of running fop on directory
+    */
+    if (!dict_get(local->xattr, conf->mds_xattr_key) &&
+        (conf->subvolume_cnt > 1)) {
+        if (local->hashed_subvol == NULL) {
+            local->hashed_subvol = dht_subvol_get_hashed(this, loc);
+            if (local->hashed_subvol == NULL) {
+                local->op_errno = EINVAL;
+                gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                        DHT_MSG_HASHED_SUBVOL_GET_FAILED, "gfid=%s",
+                        loc->pargfid, "name=%s", loc->name, "path=%s",
+                        loc->path, NULL);
+                goto err;
+            }
+        }
+        ret = dht_inode_ctx_mdsvol_set(local->inode, this,
+                                       local->hashed_subvol);
+        if (ret) {
+            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_SET_INODE_CTX_FAILED,
+                   "Failed to set hashed subvol for %s on inode vol is %s",
+                   local->loc.path,
+                   local->hashed_subvol ? local->hashed_subvol->name : "NULL");
+            goto err;
+        }
+    }
 
     if (local->hashed_subvol == NULL) {
-        local->op_errno = EINVAL;
-        gf_msg(this->name, GF_LOG_WARNING, local->op_errno,
-               DHT_MSG_HASHED_SUBVOL_GET_FAILED,
-               "(%s/%s) (path: %s): "
-               "hashed subvolume not found",
-               loc->pargfid, loc->name, loc->path);
-        goto err;
+        local->hashed_subvol = dht_subvol_get_hashed(this, loc);
+        if (local->hashed_subvol == NULL) {
+            local->op_errno = EINVAL;
+            gf_smsg(this->name, GF_LOG_WARNING, local->op_errno,
+                    DHT_MSG_HASHED_SUBVOL_GET_FAILED, "gfid=%s", loc->pargfid,
+                    "name=%s", loc->name, "path=%s", loc->path, NULL);
+            goto err;
+        }
     }
 
     local->current = &local->lock[0];
@@ -1531,7 +1423,7 @@ err:
     return -1;
 }
 
-int
+static int
 dht_selfheal_layout_alloc_start(xlator_t *this, loc_t *loc,
                                 dht_layout_t *layout)
 {
@@ -1627,7 +1519,7 @@ dht_get_layout_count(xlator_t *this, dht_layout_t *layout, int new_layout)
 
     /* if layout->spread_cnt is set, check if it is <= available
      * subvolumes (down brick and decommissioned bricks are considered
-     * un-availbale). Else return count (available up bricks) */
+     * un-available). Else return count (available up bricks) */
     count = ((layout->spread_cnt && (layout->spread_cnt <= count))
                  ? layout->spread_cnt
                  : ((count) ? count : 1));
@@ -1640,8 +1532,6 @@ dht_selfheal_layout_new_directory(call_frame_t *frame, loc_t *loc,
                                   dht_layout_t *new_layout);
 
 void
-dht_layout_entry_swap(dht_layout_t *layout, int i, int j);
-void
 dht_layout_range_swap(dht_layout_t *layout, int i, int j);
 
 /*
@@ -1650,7 +1540,7 @@ dht_layout_range_swap(dht_layout_t *layout, int i, int j);
  */
 #define OV_ENTRY(x, y) table[x * new->cnt + y]
 
-void
+static void
 dht_selfheal_layout_maximize_overlap(call_frame_t *frame, loc_t *loc,
                                      dht_layout_t *new, dht_layout_t *old)
 {
@@ -1727,7 +1617,7 @@ dht_selfheal_layout_maximize_overlap(call_frame_t *frame, loc_t *loc,
     }
 }
 
-dht_layout_t *
+static dht_layout_t *
 dht_fix_layout_of_directory(call_frame_t *frame, loc_t *loc,
                             dht_layout_t *layout)
 {
@@ -1752,9 +1642,8 @@ dht_fix_layout_of_directory(call_frame_t *frame, loc_t *loc,
     new_layout = dht_layout_new(this, priv->subvolume_cnt);
     if (!new_layout) {
         gf_uuid_unparse(loc->gfid, gfid);
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-               "mem allocation failed for new_layout, path:%s gfid:%s",
-               loc->path, gfid);
+        gf_smsg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_MEM_ALLOC_FAILED,
+                "new_layout, path=%s", loc->path, "gfid=%s", gfid, NULL);
         goto done;
     }
 
@@ -1764,10 +1653,9 @@ dht_fix_layout_of_directory(call_frame_t *frame, loc_t *loc,
 
     if (subvol_down) {
         gf_uuid_unparse(loc->gfid, gfid);
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_FIX_FAILED,
-               "Layout fix failed: %u subvolume(s) are down"
-               ". Skipping fix layout. path:%s gfid:%s",
-               subvol_down, loc->path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LAYOUT_FIX_FAILED,
+                "subvol-down=%u", subvol_down, "Skipping-fix-layout", "path=%s",
+                loc->path, "gfid=%s", gfid, NULL);
         GF_FREE(new_layout);
         return NULL;
     }
@@ -1785,10 +1673,10 @@ dht_fix_layout_of_directory(call_frame_t *frame, loc_t *loc,
 
     if (priv->du_stats) {
         for (i = 0; i < priv->subvolume_cnt; ++i) {
-            gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_SUBVOL_INFO,
-                   "subvolume %d (%s): %u chunks, path:%s", i,
-                   priv->subvolumes[i]->name, priv->du_stats[i].chunks,
-                   loc->path);
+            gf_smsg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_SUBVOL_INFO,
+                    "index=%d", i, "name=%s", priv->subvolumes[i]->name,
+                    "chunks=%u", priv->du_stats[i].chunks, "path=%s", loc->path,
+                    NULL);
 
             /* Maximize overlap if the bricks are all the same
              *  size.
@@ -1800,8 +1688,8 @@ dht_fix_layout_of_directory(call_frame_t *frame, loc_t *loc,
             }
         }
     } else {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NO_DISK_USAGE_STATUS,
-               "no du stats ?!?");
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NO_DISK_USAGE_STATUS,
+                NULL);
     }
 
     /* First give it a layout as though it is a new directory. This
@@ -1832,7 +1720,7 @@ done:
  * Having to call this 2x for each entry in the layout is pretty horrible, but
  * that's what all of this layout-sorting nonsense gets us.
  */
-uint32_t
+static uint32_t
 dht_get_chunks_from_xl(xlator_t *parent, xlator_t *child)
 {
     dht_conf_t *priv = parent->private;
@@ -1950,7 +1838,7 @@ done:
     return;
 }
 
-int
+static int
 dht_selfheal_dir_getafix(call_frame_t *frame, loc_t *loc, dht_layout_t *layout)
 {
     dht_local_t *local = NULL;
@@ -2009,9 +1897,8 @@ dht_selfheal_new_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
     linked_inode = inode_link(loc->inode, loc->parent, loc->name,
                               &local->stbuf);
     if (!linked_inode) {
-        gf_msg(frame->this->name, GF_LOG_WARNING, 0,
-               DHT_MSG_DIR_SELFHEAL_FAILED,
-               "linking inode failed (%s/%s) => %s", pgfid, loc->name, gfid);
+        gf_smsg(frame->this->name, GF_LOG_WARNING, 0, DHT_MSG_LINK_INODE_FAILED,
+                "pgfid=%s", pgfid, "name=%s", loc->name, "gfid=%s", gfid, NULL);
         ret = -1;
         goto out;
     }
@@ -2069,10 +1956,10 @@ dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
                        loc_t *loc, dht_layout_t *layout)
 {
     dht_local_t *local = NULL;
+    xlator_t *this = NULL;
     uint32_t down = 0;
     uint32_t misc = 0;
     int ret = 0;
-    xlator_t *this = NULL;
     char pgfid[GF_UUID_BUF_SIZE] = {0};
     char gfid[GF_UUID_BUF_SIZE] = {0};
     inode_t *linked_inode = NULL, *inode = NULL;
@@ -2083,6 +1970,20 @@ dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
     local->selfheal.dir_cbk = dir_cbk;
     local->selfheal.layout = dht_layout_ref(this, layout);
 
+    if (local->need_attrheal) {
+        if (__is_root_gfid(local->stbuf.ia_gfid)) {
+            local->stbuf.ia_gid = local->prebuf.ia_gid;
+            local->stbuf.ia_uid = local->prebuf.ia_uid;
+
+            local->stbuf.ia_ctime = local->prebuf.ia_ctime;
+            local->stbuf.ia_ctime_nsec = local->prebuf.ia_ctime_nsec;
+            local->stbuf.ia_prot = local->prebuf.ia_prot;
+
+        } else if (!IA_ISINVAL(local->mds_stbuf.ia_type)) {
+            local->stbuf = local->mds_stbuf;
+        }
+    }
+
     if (!__is_root_gfid(local->stbuf.ia_gfid)) {
         gf_uuid_unparse(local->stbuf.ia_gfid, gfid);
         gf_uuid_unparse(loc->parent->gfid, pgfid);
@@ -2090,9 +1991,9 @@ dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
         linked_inode = inode_link(loc->inode, loc->parent, loc->name,
                                   &local->stbuf);
         if (!linked_inode) {
-            gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_SELFHEAL_FAILED,
-                   "linking inode failed (%s/%s) => %s", pgfid, loc->name,
-                   gfid);
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_LINK_INODE_FAILED,
+                    "pgfid=%s", pgfid, "name=%s", loc->name, "gfid=%s", gfid,
+                    NULL);
             ret = 0;
             goto sorry_no_fix;
         }
@@ -2102,6 +2003,13 @@ dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
         inode_unref(inode);
     }
 
+    if (local->need_xattr_heal && (local->mds_xattr)) {
+        dht_dir_set_heal_xattr(this, local, local->xattr, local->mds_xattr,
+                               NULL, NULL);
+        dict_unref(local->mds_xattr);
+        local->mds_xattr = NULL;
+    }
+
     dht_layout_anomalies(this, loc, layout, &local->selfheal.hole_cnt,
                          &local->selfheal.overlaps_cnt,
                          &local->selfheal.missing_cnt, &local->selfheal.down,
@@ -2111,19 +2019,17 @@ dht_selfheal_directory(call_frame_t *frame, dht_selfheal_dir_cbk_t dir_cbk,
     misc = local->selfheal.misc;
 
     if (down) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_SELFHEAL_FAILED,
-               "Directory selfheal failed: %d subvolumes down."
-               "Not fixing. path = %s, gfid = %s",
-               down, loc->path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_SELFHEAL_FAILED,
+                "path=%s", loc->path, "subvol-down=%d", down, "Not-fixing",
+                "gfid=%s", gfid, NULL);
         ret = 0;
         goto sorry_no_fix;
     }
 
     if (misc) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_SELFHEAL_FAILED,
-               "Directory selfheal failed : %d subvolumes "
-               "have unrecoverable errors. path = %s, gfid = %s",
-               misc, loc->path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_SELFHEAL_FAILED,
+                "path=%s", loc->path, "misc=%d", misc, "unrecoverable-errors",
+                "gfid=%s", gfid, NULL);
 
         ret = 0;
         goto sorry_no_fix;
@@ -2209,29 +2115,28 @@ dht_dir_heal_xattrs(void *data)
     gf_uuid_unparse(local->loc.gfid, gfid);
 
     if (!mds_subvol) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_XATTR_HEAL_FAILED,
-               "No mds subvol for %s gfid = %s", local->loc.path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NO_MDS_SUBVOL, "path=%s",
+                local->loc.path, "gfid=%s", gfid, NULL);
         goto out;
     }
 
     if ((local->loc.inode && gf_uuid_is_null(local->loc.inode->gfid)) ||
         gf_uuid_is_null(local->loc.gfid)) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_XATTR_HEAL_FAILED,
-               "No gfid present so skip heal for path %s gfid = %s",
-               local->loc.path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_GFID_NOT_PRESENT,
+                "skip-heal path=%s", local->loc.path, "gfid=%s", gfid, NULL);
         goto out;
     }
 
     internal_xattr = dict_new();
     if (!internal_xattr) {
-        gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, 0,
-               "dictionary creation failed");
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_CREATE_FAILED,
+                "dictionary", NULL);
         goto out;
     }
     xdata = dict_new();
     if (!xdata) {
-        gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, 0,
-               "dictionary creation failed");
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_CREATE_FAILED,
+                "dictionary", NULL);
         goto out;
     }
 
@@ -2239,18 +2144,17 @@ dht_dir_heal_xattrs(void *data)
 
     user_xattr = dict_new();
     if (!user_xattr) {
-        gf_msg(this->name, GF_LOG_ERROR, DHT_MSG_NO_MEMORY, 0,
-               "dictionary creation failed");
+        gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_CREATE_FAILED,
+                "dictionary", NULL);
         goto out;
     }
 
     ret = syncop_listxattr(local->mds_subvol, &local->loc, &mds_xattr, NULL,
                            NULL);
     if (ret < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_DIR_XATTR_HEAL_FAILED,
-               "failed to list xattrs for "
-               "%s: on %s ",
-               local->loc.path, local->mds_subvol->name);
+        gf_smsg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LIST_XATTRS_FAILED,
+                "path=%s", local->loc.path, "name=%s", local->mds_subvol->name,
+                NULL);
     }
 
     if (!mds_xattr)
@@ -2265,10 +2169,9 @@ dht_dir_heal_xattrs(void *data)
         dict_get(user_xattr, QUOTA_LIMIT_OBJECTS_KEY)) {
         ret = dict_set_int32(xdata, GLUSTERFS_INTERNAL_FOP_KEY, 1);
         if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-                   "Failed to set dictionary value: key = %s,"
-                   " path = %s",
-                   GLUSTERFS_INTERNAL_FOP_KEY, local->loc.path);
+            gf_smsg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
+                    "key=%s", GLUSTERFS_INTERNAL_FOP_KEY, "path=%s",
+                    local->loc.path, NULL);
             goto out;
         }
     }
@@ -2280,16 +2183,25 @@ dht_dir_heal_xattrs(void *data)
         if (subvol == mds_subvol)
             continue;
         if (uret || uflag) {
+            /* Custom xattr heal is required - let posix handle it */
+            ret = dict_set_int8(xdata, "sync_backend_xattrs", _gf_true);
+            if (ret) {
+                gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
+                        "path=%s", local->loc.path, "key=%s",
+                        "sync_backend_xattrs", NULL);
+                goto out;
+            }
+
             ret = syncop_setxattr(subvol, &local->loc, user_xattr, 0, xdata,
                                   NULL);
             if (ret) {
                 xattr_hashed = 1;
-                gf_msg(this->name, GF_LOG_ERROR, -ret,
-                       DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                       "Directory xattr heal failed. Failed to set"
-                       "user xattr on path %s on "
-                       "subvol %s, gfid = %s ",
-                       local->loc.path, subvol->name, gfid);
+                gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                        DHT_MSG_DIR_XATTR_HEAL_FAILED,
+                        "set-user-xattr-failed path=%s", local->loc.path,
+                        "subvol=%s", subvol->name, "gfid=%s", gfid, NULL);
+            } else {
+                dict_del(xdata, "sync_backend_xattrs");
             }
         }
     }
@@ -2298,21 +2210,17 @@ dht_dir_heal_xattrs(void *data)
         ret = dht_dict_set_array(internal_xattr, conf->mds_xattr_key, allzero,
                                  1);
         if (ret) {
-            gf_msg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
-                   "Failed to set dictionary value:key = %s for "
-                   "path %s",
-                   conf->mds_xattr_key, local->loc.path);
+            gf_smsg(this->name, GF_LOG_WARNING, ENOMEM, DHT_MSG_DICT_SET_FAILED,
+                    "key=%s", conf->mds_xattr_key, "path=%s", local->loc.path,
+                    NULL);
             goto out;
         }
         ret = syncop_setxattr(mds_subvol, &local->loc, internal_xattr, 0, NULL,
                               NULL);
         if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, -ret,
-                   DHT_MSG_DIR_XATTR_HEAL_FAILED,
-                   "Failed to reset internal xattr "
-                   "on path %s on subvol %s"
-                   "gfid = %s ",
-                   local->loc.path, mds_subvol->name, gfid);
+            gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                    DHT_MSG_DIR_XATTR_HEAL_FAILED, "path=%s", local->loc.path,
+                    "subvol=%s", mds_subvol->name, "gfid=%s", gfid, NULL);
         }
     }
 
@@ -2353,18 +2261,18 @@ dht_dir_attr_heal(void *data)
 
     frame = data;
     local = frame->local;
-    mds_subvol = local->mds_subvol;
     this = frame->this;
     GF_VALIDATE_OR_GOTO("dht", this, out);
     GF_VALIDATE_OR_GOTO("dht", local, out);
     conf = this->private;
     GF_VALIDATE_OR_GOTO("dht", conf, out);
 
+    mds_subvol = local->mds_subvol;
     call_cnt = conf->subvolume_cnt;
 
     if (!__is_root_gfid(local->stbuf.ia_gfid) && (!mds_subvol)) {
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DIR_ATTR_HEAL_FAILED,
-               "No mds subvol for %s gfid = %s", local->loc.path, gfid);
+        gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_NO_MDS_SUBVOL, "path=%s",
+                local->loc.path, "gfid=%s", gfid, NULL);
         goto out;
     }
 
@@ -2372,11 +2280,9 @@ dht_dir_attr_heal(void *data)
         for (i = 0; i < conf->subvolume_cnt; i++) {
             if (conf->subvolumes[i] == mds_subvol) {
                 if (!conf->subvolume_status[i]) {
-                    gf_msg(this->name, GF_LOG_ERROR, 0,
-                           DHT_MSG_HASHED_SUBVOL_DOWN,
-                           "mds subvol is down for path "
-                           " %s gfid is %s Unable to set xattr ",
-                           local->loc.path, gfid);
+                    gf_smsg(this->name, GF_LOG_ERROR, 0,
+                            DHT_MSG_MDS_DOWN_UNABLE_TO_SET, "path=%s",
+                            local->loc.path, "gfid=%s", gfid, NULL);
                     goto out;
                 }
             }
@@ -2402,10 +2308,9 @@ dht_dir_attr_heal(void *data)
         if (ret) {
             gf_uuid_unparse(local->loc.gfid, gfid);
 
-            gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_DIR_ATTR_HEAL_FAILED,
-                   "Directory attr heal failed. Failed to set"
-                   " uid/gid on path %s on subvol %s, gfid = %s ",
-                   local->loc.path, subvol->name, gfid);
+            gf_smsg(this->name, GF_LOG_ERROR, -ret,
+                    DHT_MSG_DIR_ATTR_HEAL_FAILED, "path=%s", local->loc.path,
+                    "subvol=%s", subvol->name, "gfid=%s", gfid, NULL);
         }
     }
 out:
@@ -2420,7 +2325,7 @@ dht_dir_attr_heal_done(int ret, call_frame_t *sync_frame, void *data)
 }
 
 /* EXIT: dht_update_commit_hash_for_layout */
-int
+static int
 dht_update_commit_hash_for_layout_done(call_frame_t *frame, void *cookie,
                                        xlator_t *this, int32_t op_ret,
                                        int32_t op_errno, dict_t *xdata)
@@ -2440,7 +2345,7 @@ dht_update_commit_hash_for_layout_done(call_frame_t *frame, void *cookie,
     return 0;
 }
 
-int
+static int
 dht_update_commit_hash_for_layout_unlock(call_frame_t *frame, xlator_t *this)
 {
     dht_local_t *local = NULL;
@@ -2458,11 +2363,8 @@ dht_update_commit_hash_for_layout_unlock(call_frame_t *frame, xlator_t *this)
             local->op_ret = -1;
         }
 
-        gf_msg(this->name, GF_LOG_WARNING, errno,
-               DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-               "Winding unlock failed: stale locks left on brick"
-               " %s",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_WARNING, errno, DHT_MSG_WIND_UNLOCK_FAILED,
+                "path=%s", local->loc.path, NULL);
 
         dht_update_commit_hash_for_layout_done(frame, NULL, this, 0, 0, NULL);
     }
@@ -2470,7 +2372,7 @@ dht_update_commit_hash_for_layout_unlock(call_frame_t *frame, xlator_t *this)
     return 0;
 }
 
-int
+static int
 dht_update_commit_hash_for_layout_cbk(call_frame_t *frame, void *cookie,
                                       xlator_t *this, int op_ret, int op_errno,
                                       dict_t *xdata)
@@ -2497,7 +2399,7 @@ dht_update_commit_hash_for_layout_cbk(call_frame_t *frame, void *cookie,
     return 0;
 }
 
-int
+static int
 dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
                                          xlator_t *this, int32_t op_ret,
                                          int32_t op_errno, dict_t *xdata)
@@ -2525,11 +2427,8 @@ dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
     if (!xattr) {
         local->op_errno = errno;
 
-        gf_msg(this->name, GF_LOG_WARNING, errno,
-               DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-               "Directory commit hash update failed:"
-               " %s: Allocation failed",
-               local->loc.path);
+        gf_smsg(this->name, GF_LOG_WARNING, errno, DHT_MSG_COMMIT_HASH_FAILED,
+                "allocation-failed path=%s", local->loc.path, NULL);
 
         goto err;
     }
@@ -2540,11 +2439,10 @@ dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
         if (ret < 0) {
             local->op_errno = ENOENT;
 
-            gf_msg(this->name, GF_LOG_WARNING, 0,
-                   DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-                   "Directory commit hash update failed:"
-                   " %s: (subvol %s) Failed to find disk layout",
-                   local->loc.path, conf->local_subvols[i]->name);
+            gf_smsg(this->name, GF_LOG_WARNING, 0, DHT_MSG_COMMIT_HASH_FAILED,
+                    "path=%s", local->loc.path, "subvol=%s",
+                    conf->local_subvols[i]->name, "find-disk-layout-failed",
+                    NULL);
 
             goto err;
         }
@@ -2558,12 +2456,10 @@ dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
         if (ret == -1) {
             local->op_errno = errno;
 
-            gf_msg(this->name, GF_LOG_WARNING, errno,
-                   DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-                   "Directory commit hash update failed:"
-                   " %s: (subvol %s) Failed to extract disk"
-                   " layout",
-                   local->loc.path, conf->local_subvols[i]->name);
+            gf_smsg(this->name, GF_LOG_WARNING, errno,
+                    DHT_MSG_COMMIT_HASH_FAILED, "path=%s", local->loc.path,
+                    "subvol=%s", conf->local_subvols[i]->name,
+                    "extract-disk-layout-failed", NULL);
 
             goto err;
         }
@@ -2572,11 +2468,9 @@ dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
         if (!xattr[i]) {
             local->op_errno = errno;
 
-            gf_msg(this->name, GF_LOG_WARNING, errno,
-                   DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-                   "Directory commit hash update failed:"
-                   " %s: Allocation failed",
-                   local->loc.path);
+            gf_smsg(this->name, GF_LOG_WARNING, errno,
+                    DHT_MSG_COMMIT_HASH_FAILED, "path=%s Allocation-failed",
+                    local->loc.path, NULL);
 
             goto err;
         }
@@ -2585,12 +2479,10 @@ dht_update_commit_hash_for_layout_resume(call_frame_t *frame, void *cookie,
         if (ret != 0) {
             local->op_errno = ENOMEM;
 
-            gf_msg(this->name, GF_LOG_WARNING, 0,
-                   DHT_MSG_DIR_SELFHEAL_XATTR_FAILED,
-                   "Directory self heal xattr failed:"
-                   "%s: (subvol %s) Failed to set xattr"
-                   " dictionary,",
-                   local->loc.path, conf->local_subvols[i]->name);
+            gf_smsg(this->name, GF_LOG_WARNING, 0,
+                    DHT_MSG_DIR_SELFHEAL_XATTR_FAILED, "path=%s",
+                    local->loc.path, "subvol=%s", conf->local_subvols[i]->name,
+                    "set-xattr-failed", NULL);
 
             goto err;
         }
diff --git a/xlators/cluster/dht/src/dht-shared.c b/xlators/cluster/dht/src/dht-shared.c
index 0a7aa15e242..bb72b0ffbb5 100644
--- a/xlators/cluster/dht/src/dht-shared.c
+++ b/xlators/cluster/dht/src/dht-shared.c
@@ -9,7 +9,7 @@
 */
 
 /* TODO: add NS locking */
-#include "statedump.h"
+#include <glusterfs/statedump.h>
 #include "dht-common.h"
 #include "dht-messages.h"
 
@@ -17,24 +17,6 @@
 #define MAX(a, b) (((a) > (b)) ? (a) : (b))
 #endif
 
-#define GF_DECIDE_DEFRAG_THROTTLE_COUNT(throttle_count, conf)                  \
-    {                                                                          \
-        pthread_mutex_lock(&conf->defrag->dfq_mutex);                          \
-                                                                               \
-        if (!strcasecmp(conf->dthrottle, "lazy"))                              \
-            conf->defrag->recon_thread_count = 1;                              \
-                                                                               \
-        throttle_count = MAX((sysconf(_SC_NPROCESSORS_ONLN) - 4), 4);          \
-                                                                               \
-        if (!strcasecmp(conf->dthrottle, "normal"))                            \
-            conf->defrag->recon_thread_count = (throttle_count / 2);           \
-                                                                               \
-        if (!strcasecmp(conf->dthrottle, "aggressive"))                        \
-            conf->defrag->recon_thread_count = throttle_count;                 \
-                                                                               \
-        pthread_mutex_unlock(&conf->defrag->dfq_mutex);                        \
-    }
-
 /* TODO:
    - use volumename in xattr instead of "dht"
    - use NS locks
@@ -42,9 +24,7 @@
    - complete linkfile selfheal
 */
 
-extern dht_methods_t dht_methods;
-
-void
+static void
 dht_layout_dump(dht_layout_t *layout, const char *prefix)
 {
     char key[GF_DUMP_MAX_BUF_LEN];
@@ -52,8 +32,6 @@ dht_layout_dump(dht_layout_t *layout, const char *prefix)
 
     if (!layout)
         goto out;
-    if (!prefix)
-        goto out;
 
     gf_proc_dump_build_key(key, prefix, "cnt");
     gf_proc_dump_write(key, "%d", layout->cnt);
@@ -73,9 +51,9 @@ dht_layout_dump(dht_layout_t *layout, const char *prefix)
         gf_proc_dump_build_key(key, prefix, "list[%d].err", i);
         gf_proc_dump_write(key, "%d", layout->list[i].err);
         gf_proc_dump_build_key(key, prefix, "list[%d].start", i);
-        gf_proc_dump_write(key, "%u", layout->list[i].start);
+        gf_proc_dump_write(key, "0x%x", layout->list[i].start);
         gf_proc_dump_build_key(key, prefix, "list[%d].stop", i);
-        gf_proc_dump_write(key, "%u", layout->list[i].stop);
+        gf_proc_dump_write(key, "0x%x", layout->list[i].stop);
         if (layout->list[i].xlator) {
             gf_proc_dump_build_key(key, prefix, "list[%d].xlator.type", i);
             gf_proc_dump_write(key, "%s", layout->list[i].xlator->type);
@@ -152,7 +130,7 @@ dht_priv_dump(xlator_t *this)
             gf_proc_dump_write(key, "%lf", conf->du_stats[i].avail_percent);
 
             snprintf(key, sizeof(key), "du_stats[%d].avail_space", i);
-            gf_proc_dump_write(key, "%lu", conf->du_stats[i].avail_space);
+            gf_proc_dump_write(key, "%" PRIu64, conf->du_stats[i].avail_space);
 
             snprintf(key, sizeof(key), "du_stats[%d].avail_inodes", i);
             gf_proc_dump_write(key, "%lf", conf->du_stats[i].avail_inodes);
@@ -162,9 +140,9 @@ dht_priv_dump(xlator_t *this)
         }
     }
 
-    if (conf->last_stat_fetch.tv_sec)
+    if (conf->last_stat_fetch)
         gf_proc_dump_write("last_stat_fetch", "%s",
-                           ctime(&conf->last_stat_fetch.tv_sec));
+                           ctime(&conf->last_stat_fetch));
 
     UNLOCK(&conf->subvolume_lock);
 
@@ -264,7 +242,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_parse_decommissioned_bricks(xlator_t *this, dht_conf_t *conf,
                                 const char *bricks)
 {
@@ -278,6 +256,10 @@ dht_parse_decommissioned_bricks(xlator_t *this, dht_conf_t *conf,
         goto out;
 
     dup_brick = gf_strdup(bricks);
+    if (dup_brick == NULL) {
+        goto out;
+    }
+
     node = strtok_r(dup_brick, ",", &tmpstr);
     while (node) {
         for (i = 0; i < conf->subvolume_cnt; i++) {
@@ -306,14 +288,10 @@ out:
     return ret;
 }
 
-int
+static void
 dht_decommissioned_remove(xlator_t *this, dht_conf_t *conf)
 {
     int i = 0;
-    int ret = -1;
-
-    if (!conf)
-        goto out;
 
     for (i = 0; i < conf->subvolume_cnt; i++) {
         if (conf->decommissioned_bricks[i]) {
@@ -321,13 +299,9 @@ dht_decommissioned_remove(xlator_t *this, dht_conf_t *conf)
             conf->decommission_subvols_cnt--;
         }
     }
-
-    ret = 0;
-out:
-
-    return ret;
 }
-void
+
+static void
 dht_init_regex(xlator_t *this, dict_t *odict, char *name, regex_t *re,
                gf_boolean_t *re_valid, dht_conf_t *conf)
 {
@@ -384,7 +358,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_configure_throttle(xlator_t *this, dht_conf_t *conf, char *temp_str)
 {
     int rebal_thread_count = 0;
@@ -401,18 +375,20 @@ dht_configure_throttle(xlator_t *this, dht_conf_t *conf, char *temp_str)
         } else if ((gf_string2int(temp_str, &rebal_thread_count) == 0)) {
             if ((rebal_thread_count > 0) &&
                 (rebal_thread_count <= MAX_REBAL_THREADS)) {
+                conf->defrag->recon_thread_count = rebal_thread_count;
+                pthread_mutex_unlock(&conf->defrag->dfq_mutex);
                 gf_msg(this->name, GF_LOG_INFO, 0, 0,
                        "rebal thread count configured to %d",
                        rebal_thread_count);
-                conf->defrag->recon_thread_count = rebal_thread_count;
+                goto out;
             } else {
+                pthread_mutex_unlock(&conf->defrag->dfq_mutex);
                 gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_INVALID_OPTION,
                        "Invalid option: Reconfigure: "
                        "rebal-throttle should be "
                        "within range of 0 and maximum number of"
                        " cores available");
                 ret = -1;
-                pthread_mutex_unlock(&conf->defrag->dfq_mutex);
                 goto out;
             }
         } else {
@@ -521,9 +497,7 @@ dht_reconfigure(xlator_t *this, dict_t *options)
         if (ret == -1)
             goto out;
     } else {
-        ret = dht_decommissioned_remove(this, conf);
-        if (ret == -1)
-            goto out;
+        dht_decommissioned_remove(this, conf);
     }
 
     dht_init_regex(this, options, "rsync-hash-regex", &conf->rsync_regex,
@@ -563,6 +537,8 @@ gf_defrag_pattern_list_fill(xlator_t *this, gf_defrag_info_t *defrag,
     pattern_str = strtok_r(data, ",", &tmp_str);
     while (pattern_str) {
         dup_str = gf_strdup(pattern_str);
+        if (!dup_str)
+            goto out;
         pattern_list = GF_CALLOC(1, sizeof(gf_defrag_pattern_list_t), 1);
         if (!pattern_list) {
             goto out;
@@ -609,7 +585,7 @@ out:
     return ret;
 }
 
-int
+static int
 dht_init_methods(xlator_t *this)
 {
     int ret = -1;
@@ -622,7 +598,6 @@ dht_init_methods(xlator_t *this)
     methods = &(conf->methods);
 
     methods->migration_get_dst_subvol = dht_migration_get_dst_subvol;
-    methods->migration_needed = dht_migration_needed;
     methods->migration_other = NULL;
     methods->layout_search = dht_layout_search;
 
@@ -1071,84 +1046,6 @@ struct volume_options dht_options[] = {
     /* NUFA option */
     {.key = {"local-volume-name"}, .type = GF_OPTION_TYPE_XLATOR},
 
-    /* tier options */
-    {
-        .key = {"tier-pause"},
-        .type = GF_OPTION_TYPE_BOOL,
-        .default_value = "off",
-    },
-
-    {
-        .key = {"tier-promote-frequency"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "120",
-    },
-
-    {
-        .key = {"tier-demote-frequency"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "3600",
-    },
-
-    {
-        .key = {"write-freq-threshold"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "0",
-    },
-
-    {
-        .key = {"read-freq-threshold"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "0",
-    },
-    {
-        .key = {"watermark-hi"},
-        .type = GF_OPTION_TYPE_PERCENT,
-        .default_value = "90",
-    },
-    {
-        .key = {"watermark-low"},
-        .type = GF_OPTION_TYPE_PERCENT,
-        .default_value = "75",
-    },
-    {
-        .key = {"tier-mode"},
-        .type = GF_OPTION_TYPE_STR,
-        .default_value = "test",
-    },
-    {
-        .key = {"tier-compact"},
-        .type = GF_OPTION_TYPE_BOOL,
-        .default_value = "off",
-    },
-    {.key = {"tier-hot-compact-frequency"},
-     .type = GF_OPTION_TYPE_INT,
-     .default_value = "604800",
-     .description = "Frequency to compact DBs on hot tier in system"},
-    {.key = {"tier-cold-compact-frequency"},
-     .type = GF_OPTION_TYPE_INT,
-     .default_value = "604800",
-     .description = "Frequency to compact DBs on cold tier in system"},
-    {
-        .key = {"tier-max-mb"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "4000",
-    },
-    {
-        .key = {"tier-max-promote-file-size"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "0",
-    },
-    {
-        .key = {"tier-max-files"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "10000",
-    },
-    {
-        .key = {"tier-query-limit"},
-        .type = GF_OPTION_TYPE_INT,
-        .default_value = "100",
-    },
     /* switch option */
     {.key = {"pattern.switch.case"}, .type = GF_OPTION_TYPE_ANY},
 
diff --git a/xlators/cluster/dht/src/dht.c b/xlators/cluster/dht/src/dht.c
index 63ab926ca58..53de8292704 100644
--- a/xlators/cluster/dht/src/dht.c
+++ b/xlators/cluster/dht/src/dht.c
@@ -8,7 +8,6 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "statedump.h"
 #include "dht-common.h"
 
 struct xlator_fops dht_pt_fops = {
diff --git a/xlators/cluster/dht/src/dht.sym b/xlators/cluster/dht/src/dht.sym
deleted file mode 100644
index 24241a91baf..00000000000
--- a/xlators/cluster/dht/src/dht.sym
+++ /dev/null
@@ -1,9 +0,0 @@
-xlator_api
-fops
-cbks
-class_methods
-dht_methods
-options
-mem_acct_init
-reconfigure
-dumpops
diff --git a/xlators/cluster/dht/src/nufa.c b/xlators/cluster/dht/src/nufa.c
index 558611384fe..3648a564840 100644
--- a/xlators/cluster/dht/src/nufa.c
+++ b/xlators/cluster/dht/src/nufa.c
@@ -595,15 +595,9 @@ nufa_init(xlator_t *this)
 
 dht_methods_t dht_methods = {
     .migration_get_dst_subvol = dht_migration_get_dst_subvol,
-    .migration_needed = dht_migration_needed,
     .layout_search = dht_layout_search,
 };
 
-class_methods_t class_methods = {.init = nufa_init,
-                                 .fini = dht_fini,
-                                 .reconfigure = dht_reconfigure,
-                                 .notify = dht_notify};
-
 struct xlator_fops fops = {
     .lookup = nufa_lookup,
     .create = nufa_create,
@@ -645,3 +639,19 @@ struct xlator_fops fops = {
 };
 
 struct xlator_cbks cbks = {.forget = dht_forget};
+extern int32_t
+mem_acct_init(xlator_t *this);
+
+xlator_api_t xlator_api = {
+    .init = nufa_init,
+    .fini = dht_fini,
+    .notify = dht_notify,
+    .reconfigure = dht_reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1}, /* Present from the initial version */
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = dht_options,
+    .identifier = "nufa",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/cluster/dht/src/nufa.sym b/xlators/cluster/dht/src/nufa.sym
deleted file mode 100644
index 780b5fc0387..00000000000
--- a/xlators/cluster/dht/src/nufa.sym
+++ /dev/null
@@ -1,8 +0,0 @@
-fops
-cbks
-class_methods
-dht_methods
-options
-mem_acct_init
-reconfigure
-dumpops
diff --git a/xlators/cluster/dht/src/switch.c b/xlators/cluster/dht/src/switch.c
index a3c384b0f5c..207d109a025 100644
--- a/xlators/cluster/dht/src/switch.c
+++ b/xlators/cluster/dht/src/switch.c
@@ -610,9 +610,15 @@ set_switch_pattern(xlator_t *this, dht_conf_t *conf, const char *pattern_str)
     /* Get the pattern for considering switch case.
        "option block-size *avi:10MB" etc */
     option_string = gf_strdup(pattern_str);
+    if (option_string == NULL) {
+        goto err;
+    }
     switch_str = strtok_r(option_string, ";", &tmp_str);
     while (switch_str) {
         dup_str = gf_strdup(switch_str);
+        if (dup_str == NULL) {
+            goto err;
+        }
         switch_opt = GF_CALLOC(1, sizeof(struct switch_struct),
                                gf_switch_mt_switch_struct);
         if (!switch_opt) {
@@ -647,6 +653,9 @@ set_switch_pattern(xlator_t *this, dht_conf_t *conf, const char *pattern_str)
 
         if (childs) {
             dup_childs = gf_strdup(childs);
+            if (dup_childs == NULL) {
+                goto err;
+            }
             child = strtok_r(dup_childs, ",", &tmp);
             while (child) {
                 if (gf_switch_valid_child(this, child)) {
@@ -823,11 +832,6 @@ err:
     return -1;
 }
 
-class_methods_t class_methods = {.init = switch_init,
-                                 .fini = switch_fini,
-                                 .reconfigure = dht_reconfigure,
-                                 .notify = dht_notify};
-
 struct xlator_fops fops = {
     .lookup = switch_lookup,
     .create = switch_create,
@@ -869,3 +873,19 @@ struct xlator_fops fops = {
 };
 
 struct xlator_cbks cbks = {.forget = dht_forget};
+extern int32_t
+mem_acct_init(xlator_t *this);
+
+xlator_api_t xlator_api = {
+    .init = switch_init,
+    .fini = switch_fini,
+    .notify = dht_notify,
+    .reconfigure = dht_reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1}, /* Present from the initial version */
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = dht_options,
+    .identifier = "switch",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/cluster/dht/src/switch.sym b/xlators/cluster/dht/src/switch.sym
deleted file mode 100644
index 780b5fc0387..00000000000
--- a/xlators/cluster/dht/src/switch.sym
+++ /dev/null
@@ -1,8 +0,0 @@
-fops
-cbks
-class_methods
-dht_methods
-options
-mem_acct_init
-reconfigure
-dumpops
diff --git a/xlators/cluster/dht/src/tier-common.c b/xlators/cluster/dht/src/tier-common.c
deleted file mode 100644
index b86ed673042..00000000000
--- a/xlators/cluster/dht/src/tier-common.c
+++ /dev/null
@@ -1,1199 +0,0 @@
-/*
-  Copyright (c) 2015 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#include "glusterfs.h"
-#include "xlator.h"
-#include "libxlator.h"
-#include "dht-common.h"
-#include "defaults.h"
-#include "tier-common.h"
-#include "tier.h"
-
-int
-dht_link_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-             int op_errno, inode_t *inode, struct iatt *stbuf,
-             struct iatt *preparent, struct iatt *postparent, dict_t *xdata);
-
-int
-tier_link_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-              int op_errno, inode_t *inode, struct iatt *stbuf,
-              struct iatt *preparent, struct iatt *postparent, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    loc_t *oldloc = NULL;
-    loc_t *newloc = NULL;
-
-    local = frame->local;
-
-    oldloc = &local->loc;
-    newloc = &local->loc2;
-
-    if (op_ret == -1) {
-        /* No continuation on DHT inode missing errors, as we should
-         * then have a good stbuf that states P2 happened. We would
-         * get inode missing if, the file completed migrated between
-         * the lookup and the link call */
-        goto out;
-    }
-
-    if (local->call_cnt != 1) {
-        goto out;
-    }
-
-    local->call_cnt = 2;
-
-    /* Do this on the hot tier now */
-
-    STACK_WIND(frame, tier_link_cbk, local->cached_subvol,
-               local->cached_subvol->fops->link, oldloc, newloc, xdata);
-
-    return 0;
-
-out:
-    DHT_STRIP_PHASE1_FLAGS(stbuf);
-
-    DHT_STACK_UNWIND(link, frame, op_ret, op_errno, inode, stbuf, preparent,
-                     postparent, NULL);
-
-    return 0;
-}
-
-int
-tier_link(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
-          dict_t *xdata)
-{
-    xlator_t *cached_subvol = NULL;
-    xlator_t *hashed_subvol = NULL;
-    int op_errno = -1;
-    int ret = -1;
-    dht_local_t *local = NULL;
-    dht_conf_t *conf = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(oldloc, err);
-    VALIDATE_OR_GOTO(newloc, err);
-
-    conf = this->private;
-
-    local = dht_local_init(frame, oldloc, NULL, GF_FOP_LINK);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->call_cnt = 1;
-
-    cached_subvol = local->cached_subvol;
-
-    if (!cached_subvol) {
-        gf_msg_debug(this->name, 0, "no cached subvolume for path=%s",
-                     oldloc->path);
-        op_errno = ENOENT;
-        goto err;
-    }
-
-    hashed_subvol = TIER_HASHED_SUBVOL;
-
-    ret = loc_copy(&local->loc2, newloc);
-    if (ret == -1) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    if (hashed_subvol == cached_subvol) {
-        STACK_WIND(frame, dht_link_cbk, cached_subvol,
-                   cached_subvol->fops->link, oldloc, newloc, xdata);
-        return 0;
-    }
-
-    /* Create hardlinks to both the data file on the hot tier
-       and the linkto file on the cold tier */
-
-    gf_uuid_copy(local->gfid, oldloc->inode->gfid);
-
-    STACK_WIND(frame, tier_link_cbk, hashed_subvol, hashed_subvol->fops->link,
-               oldloc, newloc, xdata);
-
-    return 0;
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(link, frame, -1, op_errno, NULL, NULL, NULL, NULL, NULL);
-    return 0;
-}
-
-int
-tier_create_unlink_stale_linkto_cbk(call_frame_t *frame, void *cookie,
-                                    xlator_t *this, int op_ret, int op_errno,
-                                    struct iatt *preparent,
-                                    struct iatt *postparent, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-
-    local = frame->local;
-
-    if (local->params) {
-        dict_del(local->params, GLUSTERFS_INTERNAL_FOP_KEY);
-    }
-
-    DHT_STACK_UNWIND(create, frame, -1, local->op_errno, NULL, NULL, NULL, NULL,
-                     NULL, NULL);
-
-    return 0;
-}
-
-int
-tier_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                int op_errno, fd_t *fd, inode_t *inode, struct iatt *stbuf,
-                struct iatt *preparent, struct iatt *postparent, dict_t *xdata)
-{
-    xlator_t *prev = NULL;
-    int ret = -1;
-    dht_local_t *local = NULL;
-    xlator_t *hashed_subvol = NULL;
-    dht_conf_t *conf = NULL;
-
-    local = frame->local;
-    conf = this->private;
-
-    hashed_subvol = TIER_HASHED_SUBVOL;
-
-    if (!local) {
-        op_ret = -1;
-        op_errno = EINVAL;
-        goto out;
-    }
-
-    if (op_ret == -1) {
-        if (local->linked == _gf_true && local->xattr_req) {
-            local->op_errno = op_errno;
-            local->op_ret = op_ret;
-            ret = dht_fill_dict_to_avoid_unlink_of_migrating_file(
-                local->xattr_req);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-                       "Failed to set dictionary value to "
-                       "unlink of migrating file");
-                goto out;
-            }
-
-            STACK_WIND(frame, tier_create_unlink_stale_linkto_cbk,
-                       hashed_subvol, hashed_subvol->fops->unlink, &local->loc,
-                       0, local->xattr_req);
-            return 0;
-        }
-        goto out;
-    }
-
-    prev = cookie;
-
-    if (local->loc.parent) {
-        dht_inode_ctx_time_update(local->loc.parent, this, preparent, 0);
-
-        dht_inode_ctx_time_update(local->loc.parent, this, postparent, 1);
-    }
-
-    ret = dht_layout_preset(this, prev, inode);
-    if (ret != 0) {
-        gf_msg_debug(this->name, 0, "could not set preset layout for subvol %s",
-                     prev->name);
-        op_ret = -1;
-        op_errno = EINVAL;
-        goto out;
-    }
-
-    local->op_errno = op_errno;
-
-    if (local->linked == _gf_true) {
-        local->stbuf = *stbuf;
-        dht_linkfile_attr_heal(frame, this);
-    }
-out:
-    if (local) {
-        if (local->xattr_req) {
-            dict_del(local->xattr_req, TIER_LINKFILE_GFID);
-        }
-    }
-
-    DHT_STRIP_PHASE1_FLAGS(stbuf);
-
-    DHT_STACK_UNWIND(create, frame, op_ret, op_errno, fd, inode, stbuf,
-                     preparent, postparent, xdata);
-
-    return 0;
-}
-
-int
-tier_create_linkfile_create_cbk(call_frame_t *frame, void *cookie,
-                                xlator_t *this, int32_t op_ret,
-                                int32_t op_errno, inode_t *inode,
-                                struct iatt *stbuf, struct iatt *preparent,
-                                struct iatt *postparent, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    xlator_t *cached_subvol = NULL;
-    dht_conf_t *conf = NULL;
-    int ret = -1;
-    unsigned char *gfid = NULL;
-
-    local = frame->local;
-    if (!local) {
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    if (op_ret == -1) {
-        local->op_errno = op_errno;
-        goto err;
-    }
-
-    conf = this->private;
-    if (!conf) {
-        local->op_errno = EINVAL;
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    cached_subvol = TIER_UNHASHED_SUBVOL;
-
-    if (local->params) {
-        dict_del(local->params, conf->link_xattr_name);
-        dict_del(local->params, GLUSTERFS_INTERNAL_FOP_KEY);
-    }
-
-    /*
-     * We will delete the linkfile if data file creation fails.
-     * When deleting this stale linkfile, there is a possibility
-     * for a race between this linkfile deletion and a stale
-     * linkfile deletion triggered by another lookup from different
-     * client.
-     *
-     * For eg:
-     *
-     *     Client 1                        Client 2
-     *
-     * 1   linkfile created for foo
-     *
-     * 2   data file creation failed
-     *
-     * 3                                   creating a file with same name
-     *
-     * 4                                   lookup before creation deleted
-     *                                     the linkfile created by client1
-     *                                     considering as a stale linkfile.
-     *
-     * 5                                   New linkfile created for foo
-     *                                     with different gfid.
-     *
-     * 6 Trigger linkfile deletion as
-     *   data file creation failed.
-     *
-     * 7 Linkfile deleted which is
-     *   created by client2.
-     *
-     * 8                                   Data file created.
-     *
-     * With this race, we will end up having a file in a non-hashed subvol
-     * without a linkfile in hashed subvol.
-     *
-     * To avoid this, we store the gfid of linkfile created by client, So
-     * If we delete the linkfile , we validate gfid of existing file with
-     * stored value from posix layer.
-     *
-     * Storing this value in local->xattr_req as local->params was also used
-     * to create the data file. During the linkfile deletion we will use
-     * local->xattr_req dictionary.
-     */
-    if (!local->xattr_req) {
-        local->xattr_req = dict_new();
-        if (!local->xattr_req) {
-            local->op_errno = ENOMEM;
-            op_errno = ENOMEM;
-            goto err;
-        }
-    }
-
-    gfid = GF_MALLOC(sizeof(uuid_t), gf_common_mt_char);
-    if (!gfid) {
-        local->op_errno = ENOMEM;
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    gf_uuid_copy(gfid, stbuf->ia_gfid);
-    ret = dict_set_dynptr(local->xattr_req, TIER_LINKFILE_GFID, gfid,
-                          sizeof(uuid_t));
-    if (ret) {
-        GF_FREE(gfid);
-        gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set dictionary value"
-               " : key = %s",
-               TIER_LINKFILE_GFID);
-    }
-
-    STACK_WIND_COOKIE(frame, tier_create_cbk, cached_subvol, cached_subvol,
-                      cached_subvol->fops->create, &local->loc, local->flags,
-                      local->mode, local->umask, local->fd, local->params);
-
-    return 0;
-err:
-    DHT_STACK_UNWIND(create, frame, -1, op_errno, NULL, NULL, NULL, NULL, NULL,
-                     NULL);
-    return 0;
-}
-
-gf_boolean_t
-tier_is_hot_tier_decommissioned(xlator_t *this)
-{
-    dht_conf_t *conf = NULL;
-    xlator_t *hot_tier = NULL;
-    int i = 0;
-
-    conf = this->private;
-    hot_tier = conf->subvolumes[1];
-
-    if (conf->decommission_subvols_cnt) {
-        for (i = 0; i < conf->subvolume_cnt; i++) {
-            if (conf->decommissioned_bricks[i] &&
-                conf->decommissioned_bricks[i] == hot_tier)
-                return _gf_true;
-        }
-    }
-
-    return _gf_false;
-}
-
-int
-tier_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
-            mode_t mode, mode_t umask, fd_t *fd, dict_t *params)
-{
-    int op_errno = -1;
-    dht_local_t *local = NULL;
-    dht_conf_t *conf = NULL;
-    xlator_t *hot_subvol = NULL;
-    xlator_t *cold_subvol = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-
-    conf = this->private;
-
-    dht_get_du_info(frame, this, loc);
-
-    local = dht_local_init(frame, loc, fd, GF_FOP_CREATE);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    cold_subvol = TIER_HASHED_SUBVOL;
-    hot_subvol = TIER_UNHASHED_SUBVOL;
-
-    if (conf->subvolumes[0] != cold_subvol) {
-        hot_subvol = conf->subvolumes[0];
-    }
-    /*
-     * if hot tier full, write to cold.
-     * Also if hot tier is full, create in cold
-     */
-    if (dht_is_subvol_filled(this, hot_subvol) ||
-        tier_is_hot_tier_decommissioned(this)) {
-        gf_msg_debug(this->name, 0, "creating %s on %s", loc->path,
-                     cold_subvol->name);
-
-        STACK_WIND_COOKIE(frame, tier_create_cbk, cold_subvol, cold_subvol,
-                          cold_subvol->fops->create, loc, flags, mode, umask,
-                          fd, params);
-    } else {
-        local->params = dict_ref(params);
-        local->flags = flags;
-        local->mode = mode;
-        local->umask = umask;
-        local->cached_subvol = hot_subvol;
-        local->hashed_subvol = cold_subvol;
-
-        gf_msg_debug(this->name, 0, "creating %s on %s (link at %s)", loc->path,
-                     hot_subvol->name, cold_subvol->name);
-
-        dht_linkfile_create(frame, tier_create_linkfile_create_cbk, this,
-                            hot_subvol, cold_subvol, loc);
-
-        goto out;
-    }
-out:
-    return 0;
-
-err:
-
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(create, frame, -1, op_errno, NULL, NULL, NULL, NULL, NULL,
-                     NULL);
-
-    return 0;
-}
-
-int
-tier_unlink_nonhashed_linkfile_cbk(call_frame_t *frame, void *cookie,
-                                   xlator_t *this, int op_ret, int op_errno,
-                                   struct iatt *preparent,
-                                   struct iatt *postparent, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    xlator_t *prev = NULL;
-
-    local = frame->local;
-    prev = cookie;
-
-    LOCK(&frame->lock);
-    {
-        if ((op_ret == -1) && (op_errno != ENOENT)) {
-            local->op_errno = op_errno;
-            local->op_ret = op_ret;
-            gf_msg_debug(this->name, op_errno,
-                         "Unlink link: subvolume %s"
-                         " returned -1",
-                         prev->name);
-            goto unlock;
-        }
-
-        local->op_ret = 0;
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    if (local->op_ret == -1)
-        goto err;
-    DHT_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
-                     &local->preparent, &local->postparent, NULL);
-
-    return 0;
-
-err:
-    DHT_STACK_UNWIND(unlink, frame, -1, local->op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int
-tier_unlink_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int op_ret, int op_errno, inode_t *inode,
-                       struct iatt *preparent, dict_t *xdata,
-                       struct iatt *postparent)
-{
-    dht_local_t *local = NULL;
-    xlator_t *prev = NULL;
-    dht_conf_t *conf = NULL;
-    xlator_t *hot_subvol = NULL;
-
-    local = frame->local;
-    prev = cookie;
-    conf = this->private;
-    hot_subvol = TIER_UNHASHED_SUBVOL;
-
-    if (!op_ret) {
-        /*
-         * linkfile present on hot tier. unlinking the linkfile
-         */
-        STACK_WIND_COOKIE(frame, tier_unlink_nonhashed_linkfile_cbk, hot_subvol,
-                          hot_subvol, hot_subvol->fops->unlink, &local->loc,
-                          local->flags, NULL);
-        return 0;
-    }
-
-    LOCK(&frame->lock);
-    {
-        if (op_errno == ENOENT) {
-            local->op_ret = 0;
-            local->op_errno = op_errno;
-        } else {
-            local->op_ret = op_ret;
-            local->op_errno = op_errno;
-        }
-        gf_msg_debug(this->name, op_errno, "Lookup : subvolume %s returned -1",
-                     prev->name);
-    }
-
-    UNLOCK(&frame->lock);
-
-    DHT_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
-                     &local->preparent, &local->postparent, xdata);
-
-    return 0;
-}
-
-int
-tier_unlink_linkfile_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                         int op_ret, int op_errno, struct iatt *preparent,
-                         struct iatt *postparent, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    xlator_t *prev = NULL;
-
-    local = frame->local;
-    prev = cookie;
-
-    LOCK(&frame->lock);
-    {
-        /* Ignore EINVAL for tier to ignore error when the file
-                does not exist on the other tier  */
-        if ((op_ret == -1) && !((op_errno == ENOENT) || (op_errno == EINVAL))) {
-            local->op_errno = op_errno;
-            local->op_ret = op_ret;
-            gf_msg_debug(this->name, op_errno,
-                         "Unlink link: subvolume %s"
-                         " returned -1",
-                         prev->name);
-            goto unlock;
-        }
-
-        local->op_ret = 0;
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    if (local->op_ret == -1)
-        goto err;
-
-    DHT_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
-                     &local->preparent, &local->postparent, xdata);
-
-    return 0;
-
-err:
-    DHT_STACK_UNWIND(unlink, frame, -1, local->op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-tier_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                int op_errno, struct iatt *preparent, struct iatt *postparent,
-                dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    xlator_t *prev = NULL;
-    struct iatt *stbuf = NULL;
-    dht_conf_t *conf = NULL;
-    int ret = -1;
-    xlator_t *hot_tier = NULL;
-    xlator_t *cold_tier = NULL;
-
-    local = frame->local;
-    prev = cookie;
-    conf = this->private;
-
-    cold_tier = TIER_HASHED_SUBVOL;
-    hot_tier = TIER_UNHASHED_SUBVOL;
-
-    LOCK(&frame->lock);
-    {
-        if (op_ret == -1) {
-            if (op_errno == ENOENT) {
-                local->op_ret = 0;
-            } else {
-                local->op_ret = -1;
-                local->op_errno = op_errno;
-            }
-            gf_msg_debug(this->name, op_errno,
-                         "Unlink: subvolume %s returned -1"
-                         " with errno = %d",
-                         prev->name, op_errno);
-            goto unlock;
-        }
-
-        local->op_ret = 0;
-
-        local->postparent = *postparent;
-        local->preparent = *preparent;
-
-        if (local->loc.parent) {
-            dht_inode_ctx_time_update(local->loc.parent, this,
-                                      &local->preparent, 0);
-            dht_inode_ctx_time_update(local->loc.parent, this,
-                                      &local->postparent, 1);
-        }
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    if (local->op_ret)
-        goto out;
-
-    if (cold_tier != local->cached_subvol) {
-        /*
-         * File is present in hot tier, so there will be
-         * a link file on cold tier, deleting the linkfile
-         * from cold tier
-         */
-        STACK_WIND_COOKIE(frame, tier_unlink_linkfile_cbk, cold_tier, cold_tier,
-                          cold_tier->fops->unlink, &local->loc, local->flags,
-                          xdata);
-        return 0;
-    }
-
-    ret = dict_get_bin(xdata, DHT_IATT_IN_XDATA_KEY, (void **)&stbuf);
-    if (!ret && stbuf &&
-        ((IS_DHT_MIGRATION_PHASE2(stbuf)) || IS_DHT_MIGRATION_PHASE1(stbuf))) {
-        /*
-         * File is migrating from cold to hot tier.
-         * Delete the destination linkfile.
-         */
-        STACK_WIND_COOKIE(frame, tier_unlink_lookup_cbk, hot_tier, hot_tier,
-                          hot_tier->fops->lookup, &local->loc, NULL);
-        return 0;
-    }
-
-out:
-    DHT_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
-                     &local->preparent, &local->postparent, xdata);
-
-    return 0;
-}
-
-int
-tier_unlink(call_frame_t *frame, xlator_t *this, loc_t *loc, int xflag,
-            dict_t *xdata)
-{
-    xlator_t *cached_subvol = NULL;
-    xlator_t *hashed_subvol = NULL;
-    dht_conf_t *conf = NULL;
-    int op_errno = -1;
-    dht_local_t *local = NULL;
-    int ret = -1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-
-    conf = this->private;
-
-    local = dht_local_init(frame, loc, NULL, GF_FOP_UNLINK);
-    if (!local) {
-        op_errno = ENOMEM;
-
-        goto err;
-    }
-
-    hashed_subvol = TIER_HASHED_SUBVOL;
-
-    cached_subvol = local->cached_subvol;
-    if (!cached_subvol) {
-        gf_msg_debug(this->name, 0, "no cached subvolume for path=%s",
-                     loc->path);
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    local->flags = xflag;
-    if (IA_ISREG(loc->inode->ia_type) && (hashed_subvol == cached_subvol)) {
-        /*
-         * File resides in cold tier. We need to stat
-         * the file to see if it is being promoted.
-         * If yes we need to delete the destination
-         * file as well.
-         *
-         * Currently we are doing this check only for
-         * regular files.
-         */
-        xdata = xdata ? dict_ref(xdata) : dict_new();
-        if (xdata) {
-            ret = dict_set_int8(xdata, DHT_IATT_IN_XDATA_KEY, 1);
-            if (ret) {
-                gf_msg_debug(this->name, 0, "Failed to set dictionary key %s",
-                             DHT_IATT_IN_XDATA_KEY);
-            }
-        }
-    }
-
-    /*
-     * File is on hot tier, delete the data file first, then
-     * linkfile from cold.
-     */
-    STACK_WIND_COOKIE(frame, tier_unlink_cbk, cached_subvol, cached_subvol,
-                      cached_subvol->fops->unlink, loc, xflag, xdata);
-    if (xdata)
-        dict_unref(xdata);
-    return 0;
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(unlink, frame, -1, op_errno, NULL, NULL, NULL);
-
-    return 0;
-}
-
-int
-tier_readdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                 int op_errno, gf_dirent_t *orig_entries, dict_t *xdata)
-{
-    gf_dirent_t entries;
-    gf_dirent_t *orig_entry = NULL;
-    gf_dirent_t *entry = NULL;
-    int count = 0;
-
-    INIT_LIST_HEAD(&entries.list);
-
-    if (op_ret < 0)
-        goto unwind;
-
-    list_for_each_entry(orig_entry, (&orig_entries->list), list)
-    {
-        entry = gf_dirent_for_name(orig_entry->d_name);
-        if (!entry) {
-            gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-                   "Memory allocation failed ");
-            goto unwind;
-        }
-
-        entry->d_off = orig_entry->d_off;
-        entry->d_ino = orig_entry->d_ino;
-        entry->d_type = orig_entry->d_type;
-        entry->d_len = orig_entry->d_len;
-
-        list_add_tail(&entry->list, &entries.list);
-        count++;
-    }
-    op_ret = count;
-
-unwind:
-    if (op_ret < 0)
-        op_ret = 0;
-
-    DHT_STACK_UNWIND(readdir, frame, op_ret, op_errno, &entries, NULL);
-
-    gf_dirent_free(&entries);
-
-    return 0;
-}
-
-int
-tier_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                  int op_errno, gf_dirent_t *orig_entries, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    gf_dirent_t entries;
-    gf_dirent_t *orig_entry = NULL;
-    gf_dirent_t *entry = NULL;
-    xlator_t *prev = NULL;
-    xlator_t *next_subvol = NULL;
-    off_t next_offset = 0;
-    int count = 0;
-    dht_conf_t *conf = NULL;
-    int ret = 0;
-    inode_table_t *itable = NULL;
-    inode_t *inode = NULL;
-
-    INIT_LIST_HEAD(&entries.list);
-    prev = cookie;
-    local = frame->local;
-    itable = local->fd ? local->fd->inode->table : NULL;
-
-    conf = this->private;
-    GF_VALIDATE_OR_GOTO(this->name, conf, unwind);
-
-    if (op_ret < 0)
-        goto done;
-
-    list_for_each_entry(orig_entry, (&orig_entries->list), list)
-    {
-        next_offset = orig_entry->d_off;
-
-        if (IA_ISINVAL(orig_entry->d_stat.ia_type)) {
-            /*stat failed somewhere- ignore this entry*/
-            continue;
-        }
-
-        entry = gf_dirent_for_name(orig_entry->d_name);
-        if (!entry) {
-            goto unwind;
-        }
-
-        entry->d_off = orig_entry->d_off;
-        entry->d_stat = orig_entry->d_stat;
-        entry->d_ino = orig_entry->d_ino;
-        entry->d_type = orig_entry->d_type;
-        entry->d_len = orig_entry->d_len;
-
-        if (orig_entry->dict)
-            entry->dict = dict_ref(orig_entry->dict);
-
-        if (check_is_linkfile(NULL, (&orig_entry->d_stat), orig_entry->dict,
-                              conf->link_xattr_name)) {
-            goto entries;
-
-        } else if (IA_ISDIR(entry->d_stat.ia_type)) {
-            if (orig_entry->inode) {
-                dht_inode_ctx_time_update(orig_entry->inode, this,
-                                          &entry->d_stat, 1);
-            }
-        } else {
-            if (orig_entry->inode) {
-                ret = dht_layout_preset(this, prev, orig_entry->inode);
-                if (ret)
-                    gf_msg(this->name, GF_LOG_WARNING, 0,
-                           DHT_MSG_LAYOUT_SET_FAILED,
-                           "failed to link the layout "
-                           "in inode");
-
-                entry->inode = inode_ref(orig_entry->inode);
-            } else if (itable) {
-                /*
-                 * orig_entry->inode might be null if any upper
-                 * layer xlators below client set to null, to
-                 * force a lookup on the inode even if the inode
-                 * is present in the inode table. In that case
-                 * we just update the ctx to make sure we didn't
-                 * missed anything.
-                 */
-                inode = inode_find(itable, orig_entry->d_stat.ia_gfid);
-                if (inode) {
-                    ret = dht_layout_preset(this, TIER_HASHED_SUBVOL, inode);
-                    if (ret)
-                        gf_msg(this->name, GF_LOG_WARNING, 0,
-                               DHT_MSG_LAYOUT_SET_FAILED,
-                               "failed to link the layout"
-                               " in inode");
-                    inode_unref(inode);
-                    inode = NULL;
-                }
-            }
-        }
-
-    entries:
-        list_add_tail(&entry->list, &entries.list);
-        count++;
-    }
-    op_ret = count;
-
-done:
-    if (count == 0) {
-        /* non-zero next_offset means that
-           EOF is not yet hit on the current subvol
-        */
-        if (next_offset != 0) {
-            next_subvol = prev;
-        } else {
-            goto unwind;
-        }
-
-        STACK_WIND_COOKIE(frame, tier_readdirp_cbk, next_subvol, next_subvol,
-                          next_subvol->fops->readdirp, local->fd, local->size,
-                          next_offset, local->xattr);
-        return 0;
-    }
-
-unwind:
-    if (op_ret < 0)
-        op_ret = 0;
-
-    DHT_STACK_UNWIND(readdirp, frame, op_ret, op_errno, &entries, NULL);
-
-    gf_dirent_free(&entries);
-
-    return 0;
-}
-
-int
-tier_do_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-                off_t yoff, int whichop, dict_t *dict)
-{
-    dht_local_t *local = NULL;
-    int op_errno = -1;
-    xlator_t *hashed_subvol = NULL;
-    int ret = 0;
-    dht_conf_t *conf = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(this->private, err);
-
-    conf = this->private;
-
-    local = dht_local_init(frame, NULL, NULL, whichop);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    local->fd = fd_ref(fd);
-    local->size = size;
-    local->xattr_req = (dict) ? dict_ref(dict) : NULL;
-
-    hashed_subvol = TIER_HASHED_SUBVOL;
-
-    /* TODO: do proper readdir */
-    if (whichop == GF_FOP_READDIRP) {
-        if (dict)
-            local->xattr = dict_ref(dict);
-        else
-            local->xattr = dict_new();
-
-        if (local->xattr) {
-            ret = dict_set_uint32(local->xattr, conf->link_xattr_name, 256);
-            if (ret)
-                gf_msg(this->name, GF_LOG_WARNING, 0, DHT_MSG_DICT_SET_FAILED,
-                       "Failed to set dictionary value"
-                       " : key = %s",
-                       conf->link_xattr_name);
-        }
-
-        STACK_WIND_COOKIE(frame, tier_readdirp_cbk, hashed_subvol,
-                          hashed_subvol, hashed_subvol->fops->readdirp, fd,
-                          size, yoff, local->xattr);
-
-    } else {
-        STACK_WIND_COOKIE(frame, tier_readdir_cbk, hashed_subvol, hashed_subvol,
-                          hashed_subvol->fops->readdir, fd, size, yoff,
-                          local->xattr);
-    }
-
-    return 0;
-
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(readdir, frame, -1, op_errno, NULL, NULL);
-
-    return 0;
-}
-
-int
-tier_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-             off_t yoff, dict_t *xdata)
-{
-    int op = GF_FOP_READDIR;
-    dht_conf_t *conf = NULL;
-    int i = 0;
-
-    conf = this->private;
-    if (!conf)
-        goto out;
-
-    for (i = 0; i < conf->subvolume_cnt; i++) {
-        if (!conf->subvolume_status[i]) {
-            op = GF_FOP_READDIRP;
-            break;
-        }
-    }
-
-    if (conf->use_readdirp)
-        op = GF_FOP_READDIRP;
-
-out:
-    tier_do_readdir(frame, this, fd, size, yoff, op, 0);
-    return 0;
-}
-
-int
-tier_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-              off_t yoff, dict_t *dict)
-{
-    tier_do_readdir(frame, this, fd, size, yoff, GF_FOP_READDIRP, dict);
-    return 0;
-}
-
-int
-tier_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                int op_errno, struct statvfs *statvfs, dict_t *xdata)
-{
-    gf_boolean_t event = _gf_false;
-    qdstatfs_action_t action = qdstatfs_action_OFF;
-    dht_local_t *local = NULL;
-    int this_call_cnt = 0;
-    int bsize = 0;
-    int frsize = 0;
-    GF_UNUSED int ret = 0;
-    unsigned long new_usage = 0;
-    unsigned long cur_usage = 0;
-    xlator_t *prev = NULL;
-    dht_conf_t *conf = NULL;
-    tier_statvfs_t *tier_stat = NULL;
-
-    prev = cookie;
-    local = frame->local;
-    GF_ASSERT(local);
-
-    conf = this->private;
-
-    if (xdata)
-        ret = dict_get_int8(xdata, "quota-deem-statfs", (int8_t *)&event);
-
-    tier_stat = &local->tier_statvfs;
-
-    LOCK(&frame->lock);
-    {
-        if (op_ret == -1) {
-            local->op_errno = op_errno;
-            goto unlock;
-        }
-        if (!statvfs) {
-            op_errno = EINVAL;
-            local->op_ret = -1;
-            goto unlock;
-        }
-        local->op_ret = 0;
-
-        if (local->quota_deem_statfs) {
-            if (event == _gf_true) {
-                action = qdstatfs_action_COMPARE;
-            } else {
-                action = qdstatfs_action_NEGLECT;
-            }
-        } else {
-            if (event == _gf_true) {
-                action = qdstatfs_action_REPLACE;
-                local->quota_deem_statfs = _gf_true;
-            }
-        }
-
-        if (local->quota_deem_statfs) {
-            switch (action) {
-                case qdstatfs_action_NEGLECT:
-                    goto unlock;
-
-                case qdstatfs_action_REPLACE:
-                    local->statvfs = *statvfs;
-                    goto unlock;
-
-                case qdstatfs_action_COMPARE:
-                    new_usage = statvfs->f_blocks - statvfs->f_bfree;
-                    cur_usage = local->statvfs.f_blocks -
-                                local->statvfs.f_bfree;
-
-                    /* Take the max of the usage from subvols */
-                    if (new_usage >= cur_usage)
-                        local->statvfs = *statvfs;
-                    goto unlock;
-
-                default:
-                    break;
-            }
-        }
-
-        if (local->statvfs.f_bsize != 0) {
-            bsize = max(local->statvfs.f_bsize, statvfs->f_bsize);
-            frsize = max(local->statvfs.f_frsize, statvfs->f_frsize);
-            dht_normalize_stats(&local->statvfs, bsize, frsize);
-            dht_normalize_stats(statvfs, bsize, frsize);
-        } else {
-            local->statvfs.f_bsize = statvfs->f_bsize;
-            local->statvfs.f_frsize = statvfs->f_frsize;
-        }
-
-        if (prev == TIER_HASHED_SUBVOL) {
-            local->statvfs.f_blocks = statvfs->f_blocks;
-            local->statvfs.f_files = statvfs->f_files;
-            local->statvfs.f_fsid = statvfs->f_fsid;
-            local->statvfs.f_flag = statvfs->f_flag;
-            local->statvfs.f_namemax = statvfs->f_namemax;
-            tier_stat->blocks_used = (statvfs->f_blocks - statvfs->f_bfree);
-            tier_stat->pblocks_used = (statvfs->f_blocks - statvfs->f_bavail);
-            tier_stat->files_used = (statvfs->f_files - statvfs->f_ffree);
-            tier_stat->pfiles_used = (statvfs->f_files - statvfs->f_favail);
-            tier_stat->hashed_fsid = statvfs->f_fsid;
-        } else {
-            tier_stat->unhashed_fsid = statvfs->f_fsid;
-            tier_stat->unhashed_blocks_used = (statvfs->f_blocks -
-                                               statvfs->f_bfree);
-            tier_stat->unhashed_pblocks_used = (statvfs->f_blocks -
-                                                statvfs->f_bavail);
-            tier_stat->unhashed_files_used = (statvfs->f_files -
-                                              statvfs->f_ffree);
-            tier_stat->unhashed_pfiles_used = (statvfs->f_files -
-                                               statvfs->f_favail);
-        }
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    this_call_cnt = dht_frame_return(frame);
-    if (is_last_call(this_call_cnt)) {
-        if (tier_stat->unhashed_fsid != tier_stat->hashed_fsid) {
-            tier_stat->blocks_used += tier_stat->unhashed_blocks_used;
-            tier_stat->pblocks_used += tier_stat->unhashed_pblocks_used;
-            tier_stat->files_used += tier_stat->unhashed_files_used;
-            tier_stat->pfiles_used += tier_stat->unhashed_pfiles_used;
-        }
-        local->statvfs.f_bfree = local->statvfs.f_blocks -
-                                 tier_stat->blocks_used;
-        local->statvfs.f_bavail = local->statvfs.f_blocks -
-                                  tier_stat->pblocks_used;
-        local->statvfs.f_ffree = local->statvfs.f_files - tier_stat->files_used;
-        local->statvfs.f_favail = local->statvfs.f_files -
-                                  tier_stat->pfiles_used;
-        DHT_STACK_UNWIND(statfs, frame, local->op_ret, local->op_errno,
-                         &local->statvfs, xdata);
-    }
-
-    return 0;
-}
-
-int
-tier_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
-{
-    dht_local_t *local = NULL;
-    dht_conf_t *conf = NULL;
-    int op_errno = -1;
-    int i = -1;
-    inode_t *inode = NULL;
-    inode_table_t *itable = NULL;
-    uuid_t root_gfid = {
-        0,
-    };
-    loc_t newloc = {
-        0,
-    };
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(this->private, err);
-
-    conf = this->private;
-
-    local = dht_local_init(frame, NULL, NULL, GF_FOP_STATFS);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    if (loc->inode && !IA_ISDIR(loc->inode->ia_type)) {
-        itable = loc->inode->table;
-        if (!itable) {
-            op_errno = EINVAL;
-            goto err;
-        }
-
-        loc = &local->loc2;
-        root_gfid[15] = 1;
-
-        inode = inode_find(itable, root_gfid);
-        if (!inode) {
-            op_errno = EINVAL;
-            goto err;
-        }
-
-        dht_build_root_loc(inode, &newloc);
-        loc = &newloc;
-    }
-
-    local->call_cnt = conf->subvolume_cnt;
-
-    for (i = 0; i < conf->subvolume_cnt; i++) {
-        STACK_WIND_COOKIE(frame, tier_statfs_cbk, conf->subvolumes[i],
-                          conf->subvolumes[i],
-                          conf->subvolumes[i]->fops->statfs, loc, xdata);
-    }
-
-    return 0;
-
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    DHT_STACK_UNWIND(statfs, frame, -1, op_errno, NULL, NULL);
-
-    return 0;
-}
diff --git a/xlators/cluster/dht/src/tier-common.h b/xlators/cluster/dht/src/tier-common.h
deleted file mode 100644
index b1ebaa8004d..00000000000
--- a/xlators/cluster/dht/src/tier-common.h
+++ /dev/null
@@ -1,55 +0,0 @@
-/*
-  Copyright (c) 2015 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#ifndef _TIER_COMMON_H_
-#define _TIER_COMMON_H_
-/* Function definitions */
-int
-tier_create_unlink_stale_linkto_cbk(call_frame_t *frame, void *cookie,
-                                    xlator_t *this, int op_ret, int op_errno,
-                                    struct iatt *preparent,
-                                    struct iatt *postparent, dict_t *xdata);
-
-int
-tier_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int op_ret,
-                int op_errno, fd_t *fd, inode_t *inode, struct iatt *stbuf,
-                struct iatt *preparent, struct iatt *postparent, dict_t *xdata);
-
-int
-tier_create_linkfile_create_cbk(call_frame_t *frame, void *cookie,
-                                xlator_t *this, int32_t op_ret,
-                                int32_t op_errno, inode_t *inode,
-                                struct iatt *stbuf, struct iatt *preparent,
-                                struct iatt *postparent, dict_t *xdata);
-
-int
-tier_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
-            mode_t mode, mode_t umask, fd_t *fd, dict_t *params);
-
-int32_t
-tier_unlink(call_frame_t *frame, xlator_t *this, loc_t *loc, int xflag,
-            dict_t *xdata);
-
-int32_t
-tier_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-              off_t off, dict_t *dict);
-
-int
-tier_readdir(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-             off_t yoff, dict_t *xdata);
-
-int
-tier_link(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
-          dict_t *xdata);
-
-int
-tier_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata);
-
-#endif
diff --git a/xlators/cluster/dht/src/tier.c b/xlators/cluster/dht/src/tier.c
deleted file mode 100644
index 584f1dd76ba..00000000000
--- a/xlators/cluster/dht/src/tier.c
+++ /dev/null
@@ -1,3090 +0,0 @@
-/*
-  Copyright (c) 2015 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#include <dlfcn.h>
-
-#include "dht-common.h"
-#include "tier.h"
-#include "tier-common.h"
-#include "syscall.h"
-#include "events.h"
-#include "tier-ctr-interface.h"
-
-/*Hard coded DB info*/
-static gfdb_db_type_t dht_tier_db_type = GFDB_SQLITE3;
-/*Hard coded DB info*/
-
-/*Mutex for updating the data movement stats*/
-static pthread_mutex_t dm_stat_mutex = PTHREAD_MUTEX_INITIALIZER;
-
-/* Stores the path location of promotion query files */
-static char *promotion_qfile;
-/* Stores the path location of demotion query files */
-static char *demotion_qfile;
-
-static void *libhandle;
-static gfdb_methods_t gfdb_methods;
-
-#define DB_QUERY_RECORD_SIZE 4096
-
-/*
- * Closes all the fds and frees the qfile_array
- * */
-static void
-qfile_array_free(tier_qfile_array_t *qfile_array)
-{
-    ssize_t i = 0;
-
-    if (qfile_array) {
-        if (qfile_array->fd_array) {
-            for (i = 0; i < qfile_array->array_size; i++) {
-                if (qfile_array->fd_array[i] != -1) {
-                    sys_close(qfile_array->fd_array[i]);
-                }
-            }
-        }
-        GF_FREE(qfile_array->fd_array);
-    }
-    GF_FREE(qfile_array);
-}
-
-/* Create a new query file list with given size */
-static tier_qfile_array_t *
-qfile_array_new(ssize_t array_size)
-{
-    int ret = -1;
-    tier_qfile_array_t *qfile_array = NULL;
-    ssize_t i = 0;
-
-    GF_VALIDATE_OR_GOTO("tier", (array_size > 0), out);
-
-    qfile_array = GF_CALLOC(1, sizeof(tier_qfile_array_t),
-                            gf_tier_mt_qfile_array_t);
-    if (!qfile_array) {
-        gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to allocate memory for tier_qfile_array_t");
-        goto out;
-    }
-
-    qfile_array->fd_array = GF_MALLOC(array_size * sizeof(int),
-                                      gf_dht_mt_int32_t);
-    if (!qfile_array->fd_array) {
-        gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to allocate memory for "
-               "tier_qfile_array_t->fd_array");
-        goto out;
-    }
-
-    /* Init all the fds to -1 */
-    for (i = 0; i < array_size; i++) {
-        qfile_array->fd_array[i] = -1;
-    }
-
-    qfile_array->array_size = array_size;
-    qfile_array->next_index = 0;
-
-    /* Set exhausted count to list size as the list is empty */
-    qfile_array->exhausted_count = qfile_array->array_size;
-
-    ret = 0;
-out:
-    if (ret) {
-        qfile_array_free(qfile_array);
-        qfile_array = NULL;
-    }
-    return qfile_array;
-}
-
-/* Checks if the query file list is empty or totally exhausted. */
-static gf_boolean_t
-is_qfile_array_empty(tier_qfile_array_t *qfile_array)
-{
-    return (qfile_array->exhausted_count == qfile_array->array_size)
-               ? _gf_true
-               : _gf_false;
-}
-
-/* Shifts the next_fd pointer to the next available fd in the list */
-static void
-shift_next_index(tier_qfile_array_t *qfile_array)
-{
-    int qfile_fd = 0;
-    int spin_count = 0;
-
-    if (is_qfile_array_empty(qfile_array)) {
-        return;
-    }
-
-    do {
-        /* change next_index in a rotional manner */
-        (qfile_array->next_index == (qfile_array->array_size - 1))
-            ? qfile_array->next_index = 0
-            : qfile_array->next_index++;
-
-        qfile_fd = (qfile_array->fd_array[qfile_array->next_index]);
-
-        spin_count++;
-
-    } while ((qfile_fd == -1) && (spin_count < qfile_array->array_size));
-}
-
-/*
- * This is a non-thread safe function to read query records
- * from a list of query files in a Round-Robin manner.
- * As in when the query files get exhuasted they are closed.
- * Returns:
- * 0   if all the query records in all the query files of the list are
- *     exhausted.
- * > 0 if a query record is successfully read. Indicates the size of the query
- *     record read.
- * < 0  if there was failure
- * */
-static int
-read_query_record_list(tier_qfile_array_t *qfile_array,
-                       gfdb_query_record_t **query_record)
-{
-    int ret = -1;
-    int qfile_fd = 0;
-
-    GF_VALIDATE_OR_GOTO("tier", qfile_array, out);
-    GF_VALIDATE_OR_GOTO("tier", qfile_array->fd_array, out);
-
-    do {
-        if (is_qfile_array_empty(qfile_array)) {
-            ret = 0;
-            break;
-        }
-
-        qfile_fd = qfile_array->fd_array[qfile_array->next_index];
-        ret = gfdb_methods.gfdb_read_query_record(qfile_fd, query_record);
-        if (ret <= 0) {
-            /*The qfile_fd has reached EOF or
-             * there was an error.
-             * 1. Close the exhausted fd
-             * 2. increment the exhausted count
-             * 3. shift next_qfile to next qfile
-             **/
-            sys_close(qfile_fd);
-            qfile_array->fd_array[qfile_array->next_index] = -1;
-            qfile_array->exhausted_count++;
-            /* shift next_qfile to next qfile */
-            shift_next_index(qfile_array);
-            continue;
-        } else {
-            /* shift next_qfile to next qfile */
-            shift_next_index(qfile_array);
-            break;
-        }
-    } while (1);
-out:
-    return ret;
-}
-
-/* Check and update the watermark every WM_INTERVAL seconds */
-#define WM_INTERVAL 5
-#define WM_INTERVAL_EMERG 1
-
-static int
-tier_check_same_node(xlator_t *this, loc_t *loc, gf_defrag_info_t *defrag)
-{
-    int ret = -1;
-    dict_t *dict = NULL;
-    char *uuid_str = NULL;
-    uuid_t node_uuid = {
-        0,
-    };
-
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-    GF_VALIDATE_OR_GOTO(this->name, loc, out);
-    GF_VALIDATE_OR_GOTO(this->name, defrag, out);
-
-    if (syncop_getxattr(this, loc, &dict, GF_XATTR_NODE_UUID_KEY, NULL, NULL)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Unable to get NODE_UUID_KEY %s %s\n", loc->name, loc->path);
-        goto out;
-    }
-
-    if (dict_get_str(dict, GF_XATTR_NODE_UUID_KEY, &uuid_str) < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to get node-uuids for %s", loc->path);
-        goto out;
-    }
-
-    if (gf_uuid_parse(uuid_str, node_uuid)) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "uuid_parse failed for %s", loc->path);
-        goto out;
-    }
-
-    if (gf_uuid_compare(node_uuid, defrag->node_uuid)) {
-        gf_msg_debug(this->name, 0, "%s does not belong to this node",
-                     loc->path);
-        ret = 1;
-        goto out;
-    }
-
-    ret = 0;
-out:
-    if (dict)
-        dict_unref(dict);
-
-    return ret;
-}
-
-int
-tier_get_fs_stat(xlator_t *this, loc_t *root_loc)
-{
-    int ret = 0;
-    gf_defrag_info_t *defrag = NULL;
-    dht_conf_t *conf = NULL;
-    dict_t *xdata = NULL;
-    struct statvfs statfs = {
-        0,
-    };
-    gf_tier_conf_t *tier_conf = NULL;
-
-    conf = this->private;
-    if (!conf) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_STATUS,
-               "conf is NULL");
-        ret = -1;
-        goto exit;
-    }
-
-    defrag = conf->defrag;
-    if (!defrag) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_STATUS,
-               "defrag is NULL");
-        ret = -1;
-        goto exit;
-    }
-
-    tier_conf = &defrag->tier_conf;
-
-    xdata = dict_new();
-    if (!xdata) {
-        gf_msg(this->name, GF_LOG_ERROR, ENOMEM, DHT_MSG_NO_MEMORY,
-               "failed to allocate dictionary");
-        ret = -1;
-        goto exit;
-    }
-
-    ret = dict_set_int8(xdata, GF_INTERNAL_IGNORE_DEEM_STATFS, 1);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_DICT_SET_FAILED,
-               "Failed to set " GF_INTERNAL_IGNORE_DEEM_STATFS " in dict");
-        ret = -1;
-        goto exit;
-    }
-
-    /* Find how much free space is on the hot subvolume.
-     * Then see if that value */
-    /* is less than or greater than user defined watermarks.
-     * Stash results in */
-    /* the tier_conf data structure. */
-
-    ret = syncop_statfs(conf->subvolumes[1], root_loc, &statfs, xdata, NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LOG_TIER_STATUS,
-               "Unable to obtain statfs.");
-        goto exit;
-    }
-
-    pthread_mutex_lock(&dm_stat_mutex);
-
-    tier_conf->block_size = statfs.f_bsize;
-    tier_conf->blocks_total = statfs.f_blocks;
-    tier_conf->blocks_used = statfs.f_blocks - statfs.f_bfree;
-
-    tier_conf->percent_full = GF_PERCENTAGE(tier_conf->blocks_used,
-                                            statfs.f_blocks);
-    pthread_mutex_unlock(&dm_stat_mutex);
-
-exit:
-    if (xdata)
-        dict_unref(xdata);
-    return ret;
-}
-
-static void
-tier_send_watermark_event(const char *volname, tier_watermark_op_t old_wm,
-                          tier_watermark_op_t new_wm)
-{
-    if (old_wm == TIER_WM_LOW || old_wm == TIER_WM_NONE) {
-        if (new_wm == TIER_WM_MID) {
-            gf_event(EVENT_TIER_WATERMARK_RAISED_TO_MID, "vol=%s", volname);
-        } else if (new_wm == TIER_WM_HI) {
-            gf_event(EVENT_TIER_WATERMARK_HI, "vol=%s", volname);
-        }
-    } else if (old_wm == TIER_WM_MID) {
-        if (new_wm == TIER_WM_LOW) {
-            gf_event(EVENT_TIER_WATERMARK_DROPPED_TO_LOW, "vol=%s", volname);
-        } else if (new_wm == TIER_WM_HI) {
-            gf_event(EVENT_TIER_WATERMARK_HI, "vol=%s", volname);
-        }
-    } else if (old_wm == TIER_WM_HI) {
-        if (new_wm == TIER_WM_MID) {
-            gf_event(EVENT_TIER_WATERMARK_DROPPED_TO_MID, "vol=%s", volname);
-        } else if (new_wm == TIER_WM_LOW) {
-            gf_event(EVENT_TIER_WATERMARK_DROPPED_TO_LOW, "vol=%s", volname);
-        }
-    }
-}
-
-int
-tier_check_watermark(xlator_t *this)
-{
-    int ret = -1;
-    gf_defrag_info_t *defrag = NULL;
-    dht_conf_t *conf = NULL;
-    gf_tier_conf_t *tier_conf = NULL;
-    tier_watermark_op_t wm = TIER_WM_NONE;
-
-    conf = this->private;
-    if (!conf)
-        goto exit;
-
-    defrag = conf->defrag;
-    if (!defrag)
-        goto exit;
-
-    tier_conf = &defrag->tier_conf;
-
-    if (tier_conf->percent_full < tier_conf->watermark_low) {
-        wm = TIER_WM_LOW;
-
-    } else if (tier_conf->percent_full < tier_conf->watermark_hi) {
-        wm = TIER_WM_MID;
-
-    } else {
-        wm = TIER_WM_HI;
-    }
-
-    if (wm != tier_conf->watermark_last) {
-        tier_send_watermark_event(tier_conf->volname, tier_conf->watermark_last,
-                                  wm);
-
-        tier_conf->watermark_last = wm;
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Tier watermark now %d", wm);
-    }
-
-    ret = 0;
-
-exit:
-    return ret;
-}
-
-static gf_boolean_t
-is_hot_tier_full(gf_tier_conf_t *tier_conf)
-{
-    if (tier_conf && (tier_conf->mode == TIER_MODE_WM) &&
-        (tier_conf->watermark_last == TIER_WM_HI))
-        return _gf_true;
-
-    return _gf_false;
-}
-
-int
-tier_do_migration(xlator_t *this, int promote)
-{
-    gf_defrag_info_t *defrag = NULL;
-    dht_conf_t *conf = NULL;
-    long rand = 0;
-    int migrate = 0;
-    gf_tier_conf_t *tier_conf = NULL;
-
-    conf = this->private;
-    if (!conf)
-        goto exit;
-
-    defrag = conf->defrag;
-    if (!defrag)
-        goto exit;
-
-    if (tier_check_watermark(this) != 0) {
-        gf_msg(this->name, GF_LOG_CRITICAL, errno, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to get watermark");
-        goto exit;
-    }
-
-    tier_conf = &defrag->tier_conf;
-
-    switch (tier_conf->watermark_last) {
-        case TIER_WM_LOW:
-            migrate = promote ? 1 : 0;
-            break;
-        case TIER_WM_HI:
-            migrate = promote ? 0 : 1;
-            break;
-        case TIER_WM_MID:
-            /* coverity[DC.WEAK_CRYPTO] */
-            rand = random() % 100;
-            if (promote) {
-                migrate = (rand > tier_conf->percent_full);
-            } else {
-                migrate = (rand <= tier_conf->percent_full);
-            }
-            break;
-    }
-
-exit:
-    return migrate;
-}
-
-int
-tier_migrate(xlator_t *this, int is_promotion, dict_t *migrate_data, loc_t *loc,
-             gf_tier_conf_t *tier_conf)
-{
-    int ret = -1;
-
-    pthread_mutex_lock(&tier_conf->pause_mutex);
-    if (is_promotion)
-        tier_conf->promote_in_progress = 1;
-    else
-        tier_conf->demote_in_progress = 1;
-    pthread_mutex_unlock(&tier_conf->pause_mutex);
-
-    /* Data migration */
-    ret = syncop_setxattr(this, loc, migrate_data, 0, NULL, NULL);
-
-    pthread_mutex_lock(&tier_conf->pause_mutex);
-    if (is_promotion)
-        tier_conf->promote_in_progress = 0;
-    else
-        tier_conf->demote_in_progress = 0;
-    pthread_mutex_unlock(&tier_conf->pause_mutex);
-
-    return ret;
-}
-
-/* returns  _gf_true: if file can be promoted
- * returns _gf_false: if file cannot be promoted
- */
-static gf_boolean_t
-tier_can_promote_file(xlator_t *this, char const *file_name,
-                      struct iatt *current, gf_defrag_info_t *defrag)
-{
-    gf_boolean_t ret = _gf_false;
-    fsblkcnt_t estimated_usage = 0;
-
-    if (defrag->tier_conf.tier_max_promote_size &&
-        (current->ia_size > defrag->tier_conf.tier_max_promote_size)) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "File %s (gfid:%s) with size (%" PRIu64
-               ") exceeds maxsize "
-               "(%d) for promotion. File will not be promoted.",
-               file_name, uuid_utoa(current->ia_gfid), current->ia_size,
-               defrag->tier_conf.tier_max_promote_size);
-        goto err;
-    }
-
-    /* bypass further validations for TEST mode */
-    if (defrag->tier_conf.mode != TIER_MODE_WM) {
-        ret = _gf_true;
-        goto err;
-    }
-
-    /* convert the file size to blocks as per the block size of the
-     * destination tier
-     * NOTE: add (block_size - 1) to get the correct block size when
-     *       there is a remainder after a modulo
-     */
-    estimated_usage = ((current->ia_size + defrag->tier_conf.block_size - 1) /
-                       defrag->tier_conf.block_size) +
-                      defrag->tier_conf.blocks_used;
-
-    /* test if the estimated block usage goes above HI watermark */
-    if (GF_PERCENTAGE(estimated_usage, defrag->tier_conf.blocks_total) >=
-        defrag->tier_conf.watermark_hi) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Estimated block count consumption on "
-               "hot tier (%" PRIu64
-               ") exceeds hi watermark (%d%%). "
-               "File will not be promoted.",
-               estimated_usage, defrag->tier_conf.watermark_hi);
-        goto err;
-    }
-    ret = _gf_true;
-err:
-    return ret;
-}
-
-static int
-tier_set_migrate_data(dict_t *migrate_data)
-{
-    int failed = 1;
-
-    failed = dict_set_str(migrate_data, GF_XATTR_FILE_MIGRATE_KEY, "force");
-    if (failed) {
-        goto bail_out;
-    }
-
-    /* Flag to suggest the xattr call is from migrator */
-    failed = dict_set_str(migrate_data, "from.migrator", "yes");
-    if (failed) {
-        goto bail_out;
-    }
-
-    /* Flag to suggest its a tiering migration
-     * The reason for this dic key-value is that
-     * promotions and demotions are multithreaded
-     * so the original frame from gf_defrag_start()
-     * is not carried. A new frame will be created when
-     * we do syncop_setxattr(). This does not have the
-     * frame->root->pid of the original frame. So we pass
-     * this dic key-value when we do syncop_setxattr() to do
-     * data migration and set the frame->root->pid to
-     * GF_CLIENT_PID_TIER_DEFRAG in dht_setxattr() just before
-     * calling dht_start_rebalance_task() */
-    failed = dict_set_str(migrate_data, TIERING_MIGRATION_KEY, "yes");
-    if (failed) {
-        goto bail_out;
-    }
-
-    failed = 0;
-
-bail_out:
-    return failed;
-}
-
-static char *
-tier_get_parent_path(xlator_t *this, loc_t *p_loc, struct iatt *par_stbuf,
-                     int *per_link_status)
-{
-    int ret = -1;
-    char *parent_path = NULL;
-    dict_t *xdata_request = NULL;
-    dict_t *xdata_response = NULL;
-
-    xdata_request = dict_new();
-    if (!xdata_request) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to create xdata_request dict");
-        goto err;
-    }
-    ret = dict_set_int32(xdata_request, GET_ANCESTRY_PATH_KEY, 42);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to set value to dict : key %s \n",
-               GET_ANCESTRY_PATH_KEY);
-        goto err;
-    }
-
-    ret = syncop_lookup(this, p_loc, par_stbuf, NULL, xdata_request,
-                        &xdata_response);
-    /* When the parent gfid is a stale entry, the lookup
-     * will fail and stop the demotion process.
-     * The parent gfid can be stale when a huge folder is
-     * deleted while the files within it are being migrated
-     */
-    if (ret == -ESTALE) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_STALE_LOOKUP,
-               "Stale entry in parent lookup for %s", uuid_utoa(p_loc->gfid));
-        *per_link_status = 1;
-        goto err;
-    } else if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LOG_TIER_ERROR,
-               "Error in parent lookup for %s", uuid_utoa(p_loc->gfid));
-        *per_link_status = -1;
-        goto err;
-    }
-    ret = dict_get_str(xdata_response, GET_ANCESTRY_PATH_KEY, &parent_path);
-    if (ret || !parent_path) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to get parent path for %s", uuid_utoa(p_loc->gfid));
-        *per_link_status = -1;
-        goto err;
-    }
-
-err:
-    if (xdata_request) {
-        dict_unref(xdata_request);
-    }
-
-    if (xdata_response) {
-        dict_unref(xdata_response);
-        xdata_response = NULL;
-    }
-
-    return parent_path;
-}
-
-static int
-tier_get_file_name_and_path(xlator_t *this, uuid_t gfid,
-                            gfdb_link_info_t *link_info,
-                            char const *parent_path, loc_t *loc,
-                            int *per_link_status)
-{
-    int ret = -1;
-
-    loc->name = gf_strdup(link_info->file_name);
-    if (!loc->name) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Memory "
-               "allocation failed for %s",
-               uuid_utoa(gfid));
-        *per_link_status = -1;
-        goto err;
-    }
-    ret = gf_asprintf((char **)&(loc->path), "%s/%s", parent_path, loc->name);
-    if (ret < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to "
-               "construct file path for %s %s\n",
-               parent_path, loc->name);
-        *per_link_status = -1;
-        goto err;
-    }
-
-    ret = 0;
-
-err:
-    return ret;
-}
-
-static int
-tier_lookup_file(xlator_t *this, loc_t *p_loc, loc_t *loc, struct iatt *current,
-                 int *per_link_status)
-{
-    int ret = -1;
-
-    ret = syncop_lookup(this, loc, current, NULL, NULL, NULL);
-
-    /* The file may be deleted even when the parent
-     * is available and the lookup will
-     * return a stale entry which would stop the
-     * migration. so if its a stale entry, then skip
-     * the file and keep migrating.
-     */
-    if (ret == -ESTALE) {
-        gf_msg(this->name, GF_LOG_WARNING, -ret, DHT_MSG_STALE_LOOKUP,
-               "Stale lookup for %s", uuid_utoa(p_loc->gfid));
-        *per_link_status = 1;
-        goto err;
-    } else if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to "
-               "lookup file %s\n",
-               loc->name);
-        *per_link_status = -1;
-        goto err;
-    }
-    ret = 0;
-
-err:
-    return ret;
-}
-
-static gf_boolean_t
-tier_is_file_already_at_destination(xlator_t *src_subvol,
-                                    query_cbk_args_t *query_cbk_args,
-                                    dht_conf_t *conf, int *per_link_status)
-{
-    gf_boolean_t at_destination = _gf_true;
-
-    if (src_subvol == NULL) {
-        *per_link_status = 1;
-        goto err;
-    }
-    if (query_cbk_args->is_promotion && src_subvol == conf->subvolumes[1]) {
-        *per_link_status = 1;
-        goto err;
-    }
-
-    if (!query_cbk_args->is_promotion && src_subvol == conf->subvolumes[0]) {
-        *per_link_status = 1;
-        goto err;
-    }
-    at_destination = _gf_false;
-
-err:
-    return at_destination;
-}
-
-static void
-tier_update_migration_counters(query_cbk_args_t *query_cbk_args,
-                               gf_defrag_info_t *defrag,
-                               uint64_t *total_migrated_bytes, int *total_files)
-{
-    if (query_cbk_args->is_promotion) {
-        defrag->total_files_promoted++;
-        *total_migrated_bytes += defrag->tier_conf.st_last_promoted_size;
-        pthread_mutex_lock(&dm_stat_mutex);
-        defrag->tier_conf.blocks_used += defrag->tier_conf
-                                             .st_last_promoted_size;
-        pthread_mutex_unlock(&dm_stat_mutex);
-    } else {
-        defrag->total_files_demoted++;
-        *total_migrated_bytes += defrag->tier_conf.st_last_demoted_size;
-        pthread_mutex_lock(&dm_stat_mutex);
-        defrag->tier_conf.blocks_used -= defrag->tier_conf.st_last_demoted_size;
-        pthread_mutex_unlock(&dm_stat_mutex);
-    }
-    if (defrag->tier_conf.blocks_total) {
-        pthread_mutex_lock(&dm_stat_mutex);
-        defrag->tier_conf.percent_full = GF_PERCENTAGE(
-            defrag->tier_conf.blocks_used, defrag->tier_conf.blocks_total);
-        pthread_mutex_unlock(&dm_stat_mutex);
-    }
-
-    (*total_files)++;
-}
-
-static int
-tier_migrate_link(xlator_t *this, dht_conf_t *conf, uuid_t gfid,
-                  gfdb_link_info_t *link_info, gf_defrag_info_t *defrag,
-                  query_cbk_args_t *query_cbk_args, dict_t *migrate_data,
-                  int *per_link_status, int *total_files,
-                  uint64_t *total_migrated_bytes)
-{
-    int ret = -1;
-    struct iatt current = {
-        0,
-    };
-    struct iatt par_stbuf = {
-        0,
-    };
-    loc_t p_loc = {
-        0,
-    };
-    loc_t loc = {
-        0,
-    };
-    xlator_t *src_subvol = NULL;
-    inode_t *linked_inode = NULL;
-    char *parent_path = NULL;
-
-    /* Lookup for parent and get the path of parent */
-    gf_uuid_copy(p_loc.gfid, link_info->pargfid);
-    p_loc.inode = inode_new(defrag->root_inode->table);
-    if (!p_loc.inode) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to create reference to inode"
-               " for %s",
-               uuid_utoa(p_loc.gfid));
-
-        *per_link_status = -1;
-        goto err;
-    }
-
-    parent_path = tier_get_parent_path(this, &p_loc, &par_stbuf,
-                                       per_link_status);
-    if (!parent_path) {
-        goto err;
-    }
-
-    linked_inode = inode_link(p_loc.inode, NULL, NULL, &par_stbuf);
-    inode_unref(p_loc.inode);
-    p_loc.inode = linked_inode;
-
-    /* Preparing File Inode */
-    gf_uuid_copy(loc.gfid, gfid);
-    loc.inode = inode_new(defrag->root_inode->table);
-    gf_uuid_copy(loc.pargfid, link_info->pargfid);
-    loc.parent = inode_ref(p_loc.inode);
-
-    /* Get filename and Construct file path */
-    if (tier_get_file_name_and_path(this, gfid, link_info, parent_path, &loc,
-                                    per_link_status) != 0) {
-        goto err;
-    }
-    gf_uuid_copy(loc.parent->gfid, link_info->pargfid);
-
-    /* lookup file inode */
-    if (tier_lookup_file(this, &p_loc, &loc, &current, per_link_status) != 0) {
-        goto err;
-    }
-
-    if (query_cbk_args->is_promotion) {
-        if (!tier_can_promote_file(this, link_info->file_name, &current,
-                                   defrag)) {
-            *per_link_status = 1;
-            goto err;
-        }
-    }
-
-    linked_inode = inode_link(loc.inode, NULL, NULL, &current);
-    inode_unref(loc.inode);
-    loc.inode = linked_inode;
-
-    /*
-     * Do not promote/demote if file already is where it
-     * should be. It means another brick moved the file
-     * so is not an error. So we set per_link_status = 1
-     * so that we ignore counting this.
-     */
-    src_subvol = dht_subvol_get_cached(this, loc.inode);
-
-    if (tier_is_file_already_at_destination(src_subvol, query_cbk_args, conf,
-                                            per_link_status)) {
-        goto err;
-    }
-
-    gf_msg_debug(this->name, 0, "Tier %s: src_subvol %s file %s",
-                 (query_cbk_args->is_promotion ? "promote" : "demote"),
-                 src_subvol->name, loc.path);
-
-    ret = tier_check_same_node(this, &loc, defrag);
-    if (ret != 0) {
-        if (ret < 0) {
-            *per_link_status = -1;
-            goto err;
-        }
-        ret = 0;
-        /* By setting per_link_status to 1 we are
-         * ignoring this status and will not be counting
-         * this file for migration */
-        *per_link_status = 1;
-        goto err;
-    }
-
-    gf_uuid_copy(loc.gfid, loc.inode->gfid);
-
-    if (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Tiering paused. "
-               "Exiting tier_migrate_link");
-        goto err;
-    }
-
-    ret = tier_migrate(this, query_cbk_args->is_promotion, migrate_data, &loc,
-                       &defrag->tier_conf);
-
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, -ret, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to "
-               "migrate %s ",
-               loc.path);
-        *per_link_status = -1;
-        goto err;
-    }
-
-    tier_update_migration_counters(query_cbk_args, defrag, total_migrated_bytes,
-                                   total_files);
-
-    ret = 0;
-
-err:
-    GF_FREE((char *)loc.name);
-    loc.name = NULL;
-    loc_wipe(&loc);
-    loc_wipe(&p_loc);
-
-    if ((*total_files >= defrag->tier_conf.max_migrate_files) ||
-        (*total_migrated_bytes > defrag->tier_conf.max_migrate_bytes)) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Reached cycle migration limit."
-               "migrated bytes %" PRId64 " files %d",
-               *total_migrated_bytes, *total_files);
-        ret = -1;
-    }
-
-    return ret;
-}
-
-static int
-tier_migrate_using_query_file(void *_args)
-{
-    int ret = -1;
-    query_cbk_args_t *query_cbk_args = (query_cbk_args_t *)_args;
-    xlator_t *this = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    gfdb_query_record_t *query_record = NULL;
-    gfdb_link_info_t *link_info = NULL;
-    dict_t *migrate_data = NULL;
-    /*
-     * per_file_status and per_link_status
-     *  0  : success
-     * -1 : failure
-     *  1  : ignore the status and don't count for migration
-     * */
-    int per_file_status = 0;
-    int per_link_status = 0;
-    int total_status = 0;
-    dht_conf_t *conf = NULL;
-    uint64_t total_migrated_bytes = 0;
-    int total_files = 0;
-    loc_t root_loc = {0};
-    gfdb_time_t start_time = {0};
-    gfdb_time_t current_time = {0};
-    int total_time = 0;
-    int max_time = 0;
-    gf_boolean_t emergency_demote_mode = _gf_false;
-
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args, out);
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args->this, out);
-    this = query_cbk_args->this;
-    GF_VALIDATE_OR_GOTO(this->name, query_cbk_args->defrag, out);
-    GF_VALIDATE_OR_GOTO(this->name, query_cbk_args->qfile_array, out);
-    GF_VALIDATE_OR_GOTO(this->name, this->private, out);
-
-    conf = this->private;
-
-    defrag = query_cbk_args->defrag;
-    migrate_data = dict_new();
-    if (!migrate_data)
-        goto out;
-
-    emergency_demote_mode = (!query_cbk_args->is_promotion &&
-                             is_hot_tier_full(&defrag->tier_conf));
-
-    if (tier_set_migrate_data(migrate_data) != 0) {
-        goto out;
-    }
-
-    dht_build_root_loc(defrag->root_inode, &root_loc);
-
-    ret = gettimeofday(&start_time, NULL);
-    if (query_cbk_args->is_promotion) {
-        max_time = defrag->tier_conf.tier_promote_frequency;
-    } else {
-        max_time = defrag->tier_conf.tier_demote_frequency;
-    }
-
-    /* Per file */
-    while ((ret = read_query_record_list(query_cbk_args->qfile_array,
-                                         &query_record)) != 0) {
-        if (ret < 0) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "Failed to fetch query record "
-                   "from query file");
-            goto out;
-        }
-
-        if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
-            ret = -1;
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "Exiting tier migration as"
-                   "defrag status is not started");
-            goto out;
-        }
-
-        ret = gettimeofday(&current_time, NULL);
-        if (ret < 0) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "Could not get current time.");
-            goto out;
-        }
-
-        total_time = current_time.tv_sec - start_time.tv_sec;
-        if (total_time > max_time) {
-            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                   "Max cycle time reached. Exiting migration.");
-            goto out;
-        }
-
-        per_file_status = 0;
-        per_link_status = 0;
-
-        if (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING) {
-            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                   "Tiering paused. "
-                   "Exiting tier_migrate_using_query_file");
-            break;
-        }
-
-        if (defrag->tier_conf.mode == TIER_MODE_WM) {
-            ret = tier_get_fs_stat(this, &root_loc);
-            if (ret != 0) {
-                gfdb_methods.gfdb_query_record_free(query_record);
-                query_record = NULL;
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_STATUS,
-                       "tier_get_fs_stat() FAILED ... "
-                       "skipping file migrations until next cycle");
-                break;
-            }
-
-            if (!tier_do_migration(this, query_cbk_args->is_promotion)) {
-                gfdb_methods.gfdb_query_record_free(query_record);
-                query_record = NULL;
-
-                /* We have crossed the high watermark. Stop processing
-                 * files if this is a promotion cycle so demotion gets
-                 * a chance to start if not already running*/
-
-                if (query_cbk_args->is_promotion &&
-                    is_hot_tier_full(&defrag->tier_conf)) {
-                    gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                           "High watermark crossed during "
-                           "promotion. Exiting "
-                           "tier_migrate_using_query_file");
-                    break;
-                }
-                continue;
-            }
-        }
-
-        per_link_status = 0;
-
-        /* For now we only support single link migration. And we will
-         * ignore other hard links in the link info list of query record
-         * TODO: Multiple hard links migration */
-        if (!list_empty(&query_record->link_list)) {
-            link_info = list_first_entry(&query_record->link_list,
-                                         gfdb_link_info_t, list);
-        }
-        if (link_info != NULL) {
-            if (tier_migrate_link(this, conf, query_record->gfid, link_info,
-                                  defrag, query_cbk_args, migrate_data,
-                                  &per_link_status, &total_files,
-                                  &total_migrated_bytes) != 0) {
-                gf_msg(
-                    this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                    "%s failed for %s(gfid:%s)",
-                    (query_cbk_args->is_promotion ? "Promotion" : "Demotion"),
-                    link_info->file_name, uuid_utoa(query_record->gfid));
-            }
-        }
-        per_file_status = per_link_status;
-
-        if (per_file_status < 0) { /* Failure */
-            pthread_mutex_lock(&dm_stat_mutex);
-            defrag->total_failures++;
-            pthread_mutex_unlock(&dm_stat_mutex);
-        } else if (per_file_status == 0) { /* Success */
-            pthread_mutex_lock(&dm_stat_mutex);
-            defrag->total_files++;
-            pthread_mutex_unlock(&dm_stat_mutex);
-        } else if (per_file_status == 1) { /* Ignore */
-            per_file_status = 0;
-            /* Since this attempt was ignored we
-             * decrement the lookup count*/
-            pthread_mutex_lock(&dm_stat_mutex);
-            defrag->num_files_lookedup--;
-            pthread_mutex_unlock(&dm_stat_mutex);
-        }
-        total_status = total_status + per_file_status;
-        per_link_status = 0;
-        per_file_status = 0;
-
-        gfdb_methods.gfdb_query_record_free(query_record);
-        query_record = NULL;
-
-        /* If we are demoting and the entry watermark was HI, then
-         * we are done with emergency demotions if the current
-         * watermark has fallen below hi-watermark level
-         */
-        if (emergency_demote_mode) {
-            if (tier_check_watermark(this) == 0) {
-                if (!is_hot_tier_full(&defrag->tier_conf)) {
-                    break;
-                }
-            }
-        }
-    }
-
-out:
-    if (migrate_data)
-        dict_unref(migrate_data);
-
-    gfdb_methods.gfdb_query_record_free(query_record);
-    query_record = NULL;
-
-    return total_status;
-}
-
-/* This is the call back function per record/file from data base */
-static int
-tier_gf_query_callback(gfdb_query_record_t *gfdb_query_record, void *_args)
-{
-    int ret = -1;
-    query_cbk_args_t *query_cbk_args = _args;
-
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args, out);
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args->defrag, out);
-    GF_VALIDATE_OR_GOTO("tier", (query_cbk_args->query_fd > 0), out);
-
-    ret = gfdb_methods.gfdb_write_query_record(query_cbk_args->query_fd,
-                                               gfdb_query_record);
-    if (ret) {
-        gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed writing query record to query file");
-        goto out;
-    }
-
-    pthread_mutex_lock(&dm_stat_mutex);
-    query_cbk_args->defrag->num_files_lookedup++;
-    pthread_mutex_unlock(&dm_stat_mutex);
-
-    ret = 0;
-out:
-    return ret;
-}
-
-/* Create query file in tier process */
-static int
-tier_process_self_query(tier_brick_list_t *local_brick, void *args)
-{
-    int ret = -1;
-    char *db_path = NULL;
-    query_cbk_args_t *query_cbk_args = NULL;
-    xlator_t *this = NULL;
-    gfdb_conn_node_t *conn_node = NULL;
-    dict_t *params_dict = NULL;
-    dict_t *ctr_ipc_dict = NULL;
-    gfdb_brick_info_t *gfdb_brick_info = args;
-
-    /*Init of all the essentials*/
-    GF_VALIDATE_OR_GOTO("tier", gfdb_brick_info, out);
-    query_cbk_args = gfdb_brick_info->_query_cbk_args;
-
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args->this, out);
-    this = query_cbk_args->this;
-
-    GF_VALIDATE_OR_GOTO(this->name, gfdb_brick_info->_query_cbk_args, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->xlator, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->brick_db_path, out);
-
-    db_path = local_brick->brick_db_path;
-
-    /*Preparing DB parameters before init_db i.e getting db connection*/
-    params_dict = dict_new();
-    if (!params_dict) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "DB Params cannot initialized");
-        goto out;
-    }
-    SET_DB_PARAM_TO_DICT(this->name, params_dict,
-                         (char *)gfdb_methods.get_db_path_key(), db_path, ret,
-                         out);
-
-    /*Get the db connection*/
-    conn_node = gfdb_methods.init_db((void *)params_dict, dht_tier_db_type);
-    if (!conn_node) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "FATAL: Failed initializing db operations");
-        goto out;
-    }
-
-    /* Query for eligible files from db */
-    query_cbk_args->query_fd = open(local_brick->qfile_path,
-                                    O_WRONLY | O_CREAT | O_APPEND,
-                                    S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
-    if (query_cbk_args->query_fd < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, errno, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to open query file %s", local_brick->qfile_path);
-        goto out;
-    }
-    if (!gfdb_brick_info->_gfdb_promote) {
-        if (query_cbk_args->defrag->tier_conf.watermark_last == TIER_WM_HI) {
-            /* emergency demotion mode */
-            ret = gfdb_methods.find_all(
-                conn_node, tier_gf_query_callback, (void *)query_cbk_args,
-                query_cbk_args->defrag->tier_conf.query_limit);
-        } else {
-            if (query_cbk_args->defrag->write_freq_threshold == 0 &&
-                query_cbk_args->defrag->read_freq_threshold == 0) {
-                ret = gfdb_methods.find_unchanged_for_time(
-                    conn_node, tier_gf_query_callback, (void *)query_cbk_args,
-                    gfdb_brick_info->time_stamp);
-            } else {
-                ret = gfdb_methods.find_unchanged_for_time_freq(
-                    conn_node, tier_gf_query_callback, (void *)query_cbk_args,
-                    gfdb_brick_info->time_stamp,
-                    query_cbk_args->defrag->write_freq_threshold,
-                    query_cbk_args->defrag->read_freq_threshold, _gf_false);
-            }
-        }
-    } else {
-        if (query_cbk_args->defrag->write_freq_threshold == 0 &&
-            query_cbk_args->defrag->read_freq_threshold == 0) {
-            ret = gfdb_methods.find_recently_changed_files(
-                conn_node, tier_gf_query_callback, (void *)query_cbk_args,
-                gfdb_brick_info->time_stamp);
-        } else {
-            ret = gfdb_methods.find_recently_changed_files_freq(
-                conn_node, tier_gf_query_callback, (void *)query_cbk_args,
-                gfdb_brick_info->time_stamp,
-                query_cbk_args->defrag->write_freq_threshold,
-                query_cbk_args->defrag->read_freq_threshold, _gf_false);
-        }
-    }
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "FATAL: query from db failed");
-        goto out;
-    }
-
-    /*Clear the heat on the DB entries*/
-    /*Preparing ctr_ipc_dict*/
-    ctr_ipc_dict = dict_new();
-    if (!ctr_ipc_dict) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "ctr_ipc_dict cannot initialized");
-        goto out;
-    }
-
-    SET_DB_PARAM_TO_DICT(this->name, ctr_ipc_dict, GFDB_IPC_CTR_KEY,
-                         GFDB_IPC_CTR_CLEAR_OPS, ret, out);
-
-    ret = syncop_ipc(local_brick->xlator, GF_IPC_TARGET_CTR, ctr_ipc_dict,
-                     NULL);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed clearing the heat "
-               "on db %s error %d",
-               local_brick->brick_db_path, ret);
-        goto out;
-    }
-
-    ret = 0;
-out:
-    if (params_dict) {
-        dict_unref(params_dict);
-        params_dict = NULL;
-    }
-
-    if (ctr_ipc_dict) {
-        dict_unref(ctr_ipc_dict);
-        ctr_ipc_dict = NULL;
-    }
-
-    if (query_cbk_args && query_cbk_args->query_fd >= 0) {
-        sys_close(query_cbk_args->query_fd);
-        query_cbk_args->query_fd = -1;
-    }
-    gfdb_methods.fini_db(conn_node);
-
-    return ret;
-}
-
-/*Ask CTR to create the query file*/
-static int
-tier_process_ctr_query(tier_brick_list_t *local_brick, void *args)
-{
-    int ret = -1;
-    query_cbk_args_t *query_cbk_args = NULL;
-    xlator_t *this = NULL;
-    dict_t *ctr_ipc_in_dict = NULL;
-    dict_t *ctr_ipc_out_dict = NULL;
-    gfdb_brick_info_t *gfdb_brick_info = args;
-    gfdb_ipc_ctr_params_t *ipc_ctr_params = NULL;
-    int count = 0;
-
-    /*Init of all the essentials*/
-    GF_VALIDATE_OR_GOTO("tier", gfdb_brick_info, out);
-    query_cbk_args = gfdb_brick_info->_query_cbk_args;
-
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args->this, out);
-    this = query_cbk_args->this;
-
-    GF_VALIDATE_OR_GOTO(this->name, gfdb_brick_info->_query_cbk_args, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->xlator, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->brick_db_path, out);
-
-    /*Preparing ctr_ipc_in_dict*/
-    ctr_ipc_in_dict = dict_new();
-    if (!ctr_ipc_in_dict) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "ctr_ipc_in_dict cannot initialized");
-        goto out;
-    }
-
-    ipc_ctr_params = GF_CALLOC(1, sizeof(gfdb_ipc_ctr_params_t),
-                               gf_tier_mt_ipc_ctr_params_t);
-    if (!ipc_ctr_params) {
-        goto out;
-    }
-
-    /* set all the query params*/
-    ipc_ctr_params->is_promote = gfdb_brick_info->_gfdb_promote;
-
-    ipc_ctr_params->write_freq_threshold = query_cbk_args->defrag
-                                               ->write_freq_threshold;
-
-    ipc_ctr_params->read_freq_threshold = query_cbk_args->defrag
-                                              ->read_freq_threshold;
-
-    ipc_ctr_params->query_limit = query_cbk_args->defrag->tier_conf.query_limit;
-
-    ipc_ctr_params->emergency_demote = (!gfdb_brick_info->_gfdb_promote &&
-                                        query_cbk_args->defrag->tier_conf
-                                                .watermark_last == TIER_WM_HI);
-
-    memcpy(&ipc_ctr_params->time_stamp, gfdb_brick_info->time_stamp,
-           sizeof(gfdb_time_t));
-
-    SET_DB_PARAM_TO_DICT(this->name, ctr_ipc_in_dict, GFDB_IPC_CTR_KEY,
-                         GFDB_IPC_CTR_QUERY_OPS, ret, out);
-
-    SET_DB_PARAM_TO_DICT(this->name, ctr_ipc_in_dict,
-                         GFDB_IPC_CTR_GET_QFILE_PATH, local_brick->qfile_path,
-                         ret, out);
-
-    ret = dict_set_bin(ctr_ipc_in_dict, GFDB_IPC_CTR_GET_QUERY_PARAMS,
-                       ipc_ctr_params, sizeof(*ipc_ctr_params));
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-               "Failed setting %s to params dictionary",
-               GFDB_IPC_CTR_GET_QUERY_PARAMS);
-        GF_FREE(ipc_ctr_params);
-        goto out;
-    }
-    ipc_ctr_params = NULL;
-
-    ret = syncop_ipc(local_brick->xlator, GF_IPC_TARGET_CTR, ctr_ipc_in_dict,
-                     &ctr_ipc_out_dict);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_IPC_TIER_ERROR,
-               "Failed query on %s ret %d", local_brick->brick_db_path, ret);
-        goto out;
-    }
-
-    ret = dict_get_int32(ctr_ipc_out_dict, GFDB_IPC_CTR_RET_QUERY_COUNT,
-                         &count);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed getting count "
-               "of records on %s",
-               local_brick->brick_db_path);
-        goto out;
-    }
-
-    if (count < 0) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed query on %s", local_brick->brick_db_path);
-        ret = -1;
-        goto out;
-    }
-
-    pthread_mutex_lock(&dm_stat_mutex);
-    query_cbk_args->defrag->num_files_lookedup = count;
-    pthread_mutex_unlock(&dm_stat_mutex);
-
-    ret = 0;
-out:
-
-    if (ctr_ipc_in_dict) {
-        dict_unref(ctr_ipc_in_dict);
-        ctr_ipc_in_dict = NULL;
-    }
-
-    if (ctr_ipc_out_dict) {
-        dict_unref(ctr_ipc_out_dict);
-        ctr_ipc_out_dict = NULL;
-    }
-
-    GF_FREE(ipc_ctr_params);
-
-    return ret;
-}
-
-/* This is the call back function for each brick from hot/cold bricklist
- * It picks up each bricks db and queries for eligible files for migration.
- * The list of eligible files are populated in appropriate query files*/
-static int
-tier_process_brick(tier_brick_list_t *local_brick, void *args)
-{
-    int ret = -1;
-    dict_t *ctr_ipc_in_dict = NULL;
-    dict_t *ctr_ipc_out_dict = NULL;
-    char *strval = NULL;
-
-    GF_VALIDATE_OR_GOTO("tier", local_brick, out);
-
-    GF_VALIDATE_OR_GOTO("tier", local_brick->xlator, out);
-
-    if (dht_tier_db_type == GFDB_SQLITE3) {
-        /*Preparing ctr_ipc_in_dict*/
-        ctr_ipc_in_dict = dict_new();
-        if (!ctr_ipc_in_dict) {
-            gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "ctr_ipc_in_dict cannot initialized");
-            goto out;
-        }
-
-        ret = dict_set_str(ctr_ipc_in_dict, GFDB_IPC_CTR_KEY,
-                           GFDB_IPC_CTR_GET_DB_PARAM_OPS);
-        if (ret) {
-            gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-                   "Failed to set %s "
-                   "to params dictionary",
-                   GFDB_IPC_CTR_KEY);
-            goto out;
-        }
-
-        ret = dict_set_str(ctr_ipc_in_dict, GFDB_IPC_CTR_GET_DB_PARAM_OPS, "");
-        if (ret) {
-            gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-                   "Failed to set %s "
-                   "to params dictionary",
-                   GFDB_IPC_CTR_GET_DB_PARAM_OPS);
-            goto out;
-        }
-
-        ret = dict_set_str(ctr_ipc_in_dict, GFDB_IPC_CTR_GET_DB_KEY,
-                           "journal_mode");
-        if (ret) {
-            gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-                   "Failed to set %s "
-                   "to params dictionary",
-                   GFDB_IPC_CTR_GET_DB_KEY);
-            goto out;
-        }
-
-        ret = syncop_ipc(local_brick->xlator, GF_IPC_TARGET_CTR,
-                         ctr_ipc_in_dict, &ctr_ipc_out_dict);
-        if (ret || ctr_ipc_out_dict == NULL) {
-            gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "Failed to get "
-                   "journal_mode of sql db %s",
-                   local_brick->brick_db_path);
-            goto out;
-        }
-
-        ret = dict_get_str(ctr_ipc_out_dict, "journal_mode", &strval);
-        if (ret) {
-            gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_GET_PARAM_FAILED,
-                   "Failed to get %s "
-                   "from params dictionary"
-                   "journal_mode",
-                   strval);
-            goto out;
-        }
-
-        if (strval && (strncmp(strval, "wal", SLEN("wal")) == 0)) {
-            ret = tier_process_self_query(local_brick, args);
-            if (ret) {
-                goto out;
-            }
-        } else {
-            ret = tier_process_ctr_query(local_brick, args);
-            if (ret) {
-                goto out;
-            }
-        }
-        ret = 0;
-
-    } else {
-        ret = tier_process_self_query(local_brick, args);
-        if (ret) {
-            goto out;
-        }
-    }
-
-    ret = 0;
-out:
-    if (ctr_ipc_in_dict)
-        dict_unref(ctr_ipc_in_dict);
-
-    if (ctr_ipc_out_dict)
-        dict_unref(ctr_ipc_out_dict);
-
-    return ret;
-}
-
-static int
-tier_build_migration_qfile(migration_args_t *args,
-                           query_cbk_args_t *query_cbk_args,
-                           gf_boolean_t is_promotion)
-{
-    gfdb_time_t current_time;
-    gfdb_brick_info_t gfdb_brick_info;
-    gfdb_time_t time_in_past;
-    int ret = -1;
-    tier_brick_list_t *local_brick = NULL;
-    int i = 0;
-    time_in_past.tv_sec = args->freq_time;
-    time_in_past.tv_usec = 0;
-
-    ret = gettimeofday(&current_time, NULL);
-    if (ret == -1) {
-        gf_msg(args->this->name, GF_LOG_ERROR, errno,
-               DHT_MSG_SYS_CALL_GET_TIME_FAILED, "Failed to get current time");
-        goto out;
-    }
-    time_in_past.tv_sec = current_time.tv_sec - time_in_past.tv_sec;
-
-    /* The migration daemon may run a varying numberof usec after the */
-    /* sleep call triggers. A file may be registered in CTR some number */
-    /* of usec X after the daemon started and missed in the subsequent */
-    /* cycle if the daemon starts Y usec after the period in seconds */
-    /* where Y>X. Normalize away this problem by always setting usec */
-    /* to 0. */
-    time_in_past.tv_usec = 0;
-
-    gfdb_brick_info.time_stamp = &time_in_past;
-    gfdb_brick_info._gfdb_promote = is_promotion;
-    gfdb_brick_info._query_cbk_args = query_cbk_args;
-
-    list_for_each_entry(local_brick, args->brick_list, list)
-    {
-        /* Construct query file path for this brick
-         * i.e
-         * /var/run/gluster/xlator_name/
-         * {promote/demote}-brickname-indexinbricklist
-         * So that no two query files will have same path even
-         * bricks have the same name
-         * */
-        snprintf(local_brick->qfile_path, PATH_MAX, "%s-%s-%d",
-                 GET_QFILE_PATH(gfdb_brick_info._gfdb_promote),
-                 local_brick->brick_name, i);
-
-        /* Delete any old query files for this brick */
-        sys_unlink(local_brick->qfile_path);
-
-        ret = tier_process_brick(local_brick, &gfdb_brick_info);
-        if (ret) {
-            gf_msg(args->this->name, GF_LOG_ERROR, 0,
-                   DHT_MSG_BRICK_QUERY_FAILED, "Brick %s query failed\n",
-                   local_brick->brick_db_path);
-        }
-        i++;
-    }
-    ret = 0;
-out:
-    return ret;
-}
-
-static int
-tier_migrate_files_using_qfile(migration_args_t *comp,
-                               query_cbk_args_t *query_cbk_args)
-{
-    int ret = -1;
-    tier_brick_list_t *local_brick = NULL;
-    tier_brick_list_t *temp = NULL;
-    gfdb_time_t current_time = {
-        0,
-    };
-    ssize_t qfile_array_size = 0;
-    int count = 0;
-    int temp_fd = 0;
-    gf_tier_conf_t *tier_conf = NULL;
-
-    tier_conf = &(query_cbk_args->defrag->tier_conf);
-
-    /* Time for error query files */
-    gettimeofday(&current_time, NULL);
-
-    /* Build the qfile list */
-    list_for_each_entry_safe(local_brick, temp, comp->brick_list, list)
-    {
-        qfile_array_size++;
-    }
-    query_cbk_args->qfile_array = qfile_array_new(qfile_array_size);
-    if (!query_cbk_args->qfile_array) {
-        gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to create new "
-               "qfile_array");
-        goto out;
-    }
-
-    /*Open all qfiles*/
-    count = 0;
-    query_cbk_args->qfile_array->exhausted_count = 0;
-    list_for_each_entry_safe(local_brick, temp, comp->brick_list, list)
-    {
-        temp_fd = query_cbk_args->qfile_array->fd_array[count];
-        temp_fd = open(local_brick->qfile_path, O_RDONLY,
-                       S_IRUSR | S_IWUSR | S_IRGRP | S_IROTH);
-        if (temp_fd < 0) {
-            gf_msg("tier", GF_LOG_ERROR, errno, DHT_MSG_LOG_TIER_ERROR,
-                   "Failed to open "
-                   "%s to the query file",
-                   local_brick->qfile_path);
-            query_cbk_args->qfile_array->exhausted_count++;
-        }
-        query_cbk_args->qfile_array->fd_array[count] = temp_fd;
-        count++;
-    }
-
-    /* Moving the query file index to the next, so that we won't the same
-     * query file every cycle as the first one */
-    query_cbk_args->qfile_array
-        ->next_index = (query_cbk_args->is_promotion)
-                           ? tier_conf->last_promote_qfile_index
-                           : tier_conf->last_demote_qfile_index;
-    shift_next_index(query_cbk_args->qfile_array);
-    if (query_cbk_args->is_promotion) {
-        tier_conf->last_promote_qfile_index = query_cbk_args->qfile_array
-                                                  ->next_index;
-    } else {
-        tier_conf->last_demote_qfile_index = query_cbk_args->qfile_array
-                                                 ->next_index;
-    }
-
-    /* Migrate files using query file list */
-    ret = tier_migrate_using_query_file((void *)query_cbk_args);
-out:
-    qfile_array_free(query_cbk_args->qfile_array);
-
-    /* If there is an error rename all the query files to .err files
-     * with a timestamp for better debugging */
-    if (ret) {
-        struct tm tm = {
-            0,
-        };
-        char time_str[128] = {
-            0,
-        };
-        char query_file_path_err[PATH_MAX] = {
-            0,
-        };
-        int32_t len = 0;
-
-        /* Time format for error query files */
-        gmtime_r(&current_time.tv_sec, &tm);
-        strftime(time_str, sizeof(time_str), "%F-%T", &tm);
-
-        list_for_each_entry_safe(local_brick, temp, comp->brick_list, list)
-        {
-            /* rename error qfile*/
-            len = snprintf(query_file_path_err, sizeof(query_file_path_err),
-                           "%s-%s.err", local_brick->qfile_path, time_str);
-            if ((len >= 0) && (len < sizeof(query_file_path_err))) {
-                if (sys_rename(local_brick->qfile_path, query_file_path_err) ==
-                    -1)
-                    gf_msg_debug("tier", 0,
-                                 "rename "
-                                 "failed");
-            }
-        }
-    }
-
-    query_cbk_args->qfile_array = NULL;
-
-    return ret;
-}
-
-int
-tier_demote(migration_args_t *demotion_args)
-{
-    query_cbk_args_t query_cbk_args;
-    int ret = -1;
-
-    GF_VALIDATE_OR_GOTO("tier", demotion_args, out);
-    GF_VALIDATE_OR_GOTO("tier", demotion_args->this, out);
-    GF_VALIDATE_OR_GOTO(demotion_args->this->name, demotion_args->brick_list,
-                        out);
-    GF_VALIDATE_OR_GOTO(demotion_args->this->name, demotion_args->defrag, out);
-
-    THIS = demotion_args->this;
-
-    query_cbk_args.this = demotion_args->this;
-    query_cbk_args.defrag = demotion_args->defrag;
-    query_cbk_args.is_promotion = 0;
-
-    /*Build the query file using bricklist*/
-    ret = tier_build_migration_qfile(demotion_args, &query_cbk_args, _gf_false);
-    if (ret)
-        goto out;
-
-    /* Migrate files using the query file */
-    ret = tier_migrate_files_using_qfile(demotion_args, &query_cbk_args);
-    if (ret)
-        goto out;
-
-out:
-    demotion_args->return_value = ret;
-    return ret;
-}
-
-int
-tier_promote(migration_args_t *promotion_args)
-{
-    int ret = -1;
-    query_cbk_args_t query_cbk_args;
-
-    GF_VALIDATE_OR_GOTO("tier", promotion_args->this, out);
-    GF_VALIDATE_OR_GOTO(promotion_args->this->name, promotion_args->brick_list,
-                        out);
-    GF_VALIDATE_OR_GOTO(promotion_args->this->name, promotion_args->defrag,
-                        out);
-
-    THIS = promotion_args->this;
-
-    query_cbk_args.this = promotion_args->this;
-    query_cbk_args.defrag = promotion_args->defrag;
-    query_cbk_args.is_promotion = 1;
-
-    /*Build the query file using bricklist*/
-    ret = tier_build_migration_qfile(promotion_args, &query_cbk_args, _gf_true);
-    if (ret)
-        goto out;
-
-    /* Migrate files using the query file */
-    ret = tier_migrate_files_using_qfile(promotion_args, &query_cbk_args);
-    if (ret)
-        goto out;
-
-out:
-    promotion_args->return_value = ret;
-    return ret;
-}
-
-/*
- * Command the CTR on a brick to compact the local database using an IPC
- */
-static int
-tier_process_self_compact(tier_brick_list_t *local_brick, void *args)
-{
-    int ret = -1;
-    char *db_path = NULL;
-    query_cbk_args_t *query_cbk_args = NULL;
-    xlator_t *this = NULL;
-    gfdb_conn_node_t *conn_node = NULL;
-    dict_t *params_dict = NULL;
-    dict_t *ctr_ipc_dict = NULL;
-    gfdb_brick_info_t *gfdb_brick_info = args;
-
-    /*Init of all the essentials*/
-    GF_VALIDATE_OR_GOTO("tier", gfdb_brick_info, out);
-    query_cbk_args = gfdb_brick_info->_query_cbk_args;
-
-    GF_VALIDATE_OR_GOTO("tier", query_cbk_args->this, out);
-    this = query_cbk_args->this;
-
-    GF_VALIDATE_OR_GOTO(this->name, gfdb_brick_info->_query_cbk_args, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->xlator, out);
-
-    GF_VALIDATE_OR_GOTO(this->name, local_brick->brick_db_path, out);
-
-    db_path = local_brick->brick_db_path;
-
-    /*Preparing DB parameters before init_db i.e getting db connection*/
-    params_dict = dict_new();
-    if (!params_dict) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "DB Params cannot initialized");
-        goto out;
-    }
-    SET_DB_PARAM_TO_DICT(this->name, params_dict,
-                         (char *)gfdb_methods.get_db_path_key(), db_path, ret,
-                         out);
-
-    /*Get the db connection*/
-    conn_node = gfdb_methods.init_db((void *)params_dict, dht_tier_db_type);
-    if (!conn_node) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "FATAL: Failed initializing db operations");
-        goto out;
-    }
-
-    ret = 0;
-
-    /*Preparing ctr_ipc_dict*/
-    ctr_ipc_dict = dict_new();
-    if (!ctr_ipc_dict) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "ctr_ipc_dict cannot initialized");
-        goto out;
-    }
-
-    ret = dict_set_int32(ctr_ipc_dict, "compact_active",
-                         query_cbk_args->defrag->tier_conf.compact_active);
-
-    if (ret) {
-        gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-               "Failed to set %s "
-               "to params dictionary",
-               "compact_active");
-        goto out;
-    }
-
-    ret = dict_set_int32(
-        ctr_ipc_dict, "compact_mode_switched",
-        query_cbk_args->defrag->tier_conf.compact_mode_switched);
-
-    if (ret) {
-        gf_msg("tier", GF_LOG_ERROR, 0, LG_MSG_SET_PARAM_FAILED,
-               "Failed to set %s "
-               "to params dictionary",
-               "compact_mode_switched");
-        goto out;
-    }
-
-    SET_DB_PARAM_TO_DICT(this->name, ctr_ipc_dict, GFDB_IPC_CTR_KEY,
-                         GFDB_IPC_CTR_SET_COMPACT_PRAGMA, ret, out);
-
-    gf_msg(this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-           "Starting Compaction IPC");
-
-    ret = syncop_ipc(local_brick->xlator, GF_IPC_TARGET_CTR, ctr_ipc_dict,
-                     NULL);
-
-    gf_msg(this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-           "Ending Compaction IPC");
-
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed compaction "
-               "on db %s error %d",
-               local_brick->brick_db_path, ret);
-        goto out;
-    }
-
-    gf_msg(this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-           "SUCCESS: %s Compaction", local_brick->brick_name);
-
-    ret = 0;
-out:
-    if (params_dict) {
-        dict_unref(params_dict);
-        params_dict = NULL;
-    }
-
-    if (ctr_ipc_dict) {
-        dict_unref(ctr_ipc_dict);
-        ctr_ipc_dict = NULL;
-    }
-
-    gfdb_methods.fini_db(conn_node);
-
-    return ret;
-}
-
-/*
- * This is the call back function for each brick from hot/cold bricklist.
- * It determines the database type on each brick and calls the corresponding
- * function to prepare the compaction IPC.
- */
-static int
-tier_compact_db_brick(tier_brick_list_t *local_brick, void *args)
-{
-    int ret = -1;
-
-    GF_VALIDATE_OR_GOTO("tier", local_brick, out);
-
-    GF_VALIDATE_OR_GOTO("tier", local_brick->xlator, out);
-
-    ret = tier_process_self_compact(local_brick, args);
-    if (ret) {
-        gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Brick %s did not compact", local_brick->brick_name);
-        goto out;
-    }
-
-    ret = 0;
-
-out:
-
-    return ret;
-}
-
-static int
-tier_send_compact(migration_args_t *args, query_cbk_args_t *query_cbk_args)
-{
-    gfdb_time_t current_time;
-    gfdb_brick_info_t gfdb_brick_info;
-    gfdb_time_t time_in_past;
-    int ret = -1;
-    tier_brick_list_t *local_brick = NULL;
-
-    time_in_past.tv_sec = args->freq_time;
-    time_in_past.tv_usec = 0;
-
-    ret = gettimeofday(&current_time, NULL);
-    if (ret == -1) {
-        gf_msg(args->this->name, GF_LOG_ERROR, errno,
-               DHT_MSG_SYS_CALL_GET_TIME_FAILED, "Failed to get current time");
-        goto out;
-    }
-    time_in_past.tv_sec = current_time.tv_sec - time_in_past.tv_sec;
-
-    /* The migration daemon may run a varying numberof usec after the sleep
-       call triggers. A file may be registered in CTR some number of usec X
-       after the daemon started and missed in the subsequent cycle if the
-       daemon starts Y usec after the period in seconds where Y>X. Normalize
-       away this problem by always setting usec to 0. */
-    time_in_past.tv_usec = 0;
-
-    gfdb_brick_info.time_stamp = &time_in_past;
-
-    /* This is meant to say we are always compacting at this point */
-    /* We simply borrow the promotion flag to do this */
-    gfdb_brick_info._gfdb_promote = 1;
-
-    gfdb_brick_info._query_cbk_args = query_cbk_args;
-
-    list_for_each_entry(local_brick, args->brick_list, list)
-    {
-        gf_msg(args->this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Start compaction for %s", local_brick->brick_name);
-
-        ret = tier_compact_db_brick(local_brick, &gfdb_brick_info);
-        if (ret) {
-            gf_msg(args->this->name, GF_LOG_ERROR, 0,
-                   DHT_MSG_BRICK_QUERY_FAILED, "Brick %s compaction failed\n",
-                   local_brick->brick_db_path);
-        }
-
-        gf_msg(args->this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-               "End compaction for %s", local_brick->brick_name);
-    }
-    ret = 0;
-out:
-    return ret;
-}
-
-static int
-tier_compact(void *args)
-{
-    int ret = -1;
-    query_cbk_args_t query_cbk_args;
-    migration_args_t *compaction_args = args;
-
-    GF_VALIDATE_OR_GOTO("tier", compaction_args->this, out);
-    GF_VALIDATE_OR_GOTO(compaction_args->this->name,
-                        compaction_args->brick_list, out);
-    GF_VALIDATE_OR_GOTO(compaction_args->this->name, compaction_args->defrag,
-                        out);
-
-    THIS = compaction_args->this;
-
-    query_cbk_args.this = compaction_args->this;
-    query_cbk_args.defrag = compaction_args->defrag;
-    query_cbk_args.is_compaction = 1;
-
-    /* Send the compaction pragma out to all the bricks on the bricklist. */
-    /* tier_get_bricklist ensures all bricks on the list are local to */
-    /* this node. */
-    ret = tier_send_compact(compaction_args, &query_cbk_args);
-    if (ret)
-        goto out;
-
-    ret = 0;
-out:
-    compaction_args->return_value = ret;
-    return ret;
-}
-
-static int
-tier_get_bricklist(xlator_t *xl, struct list_head *local_bricklist_head)
-{
-    xlator_list_t *child = NULL;
-    char *rv = NULL;
-    char *rh = NULL;
-    char *brickname = NULL;
-    char db_name[PATH_MAX] = "";
-    int ret = 0;
-    tier_brick_list_t *local_brick = NULL;
-    int32_t len = 0;
-
-    GF_VALIDATE_OR_GOTO("tier", xl, out);
-    GF_VALIDATE_OR_GOTO("tier", local_bricklist_head, out);
-
-    /*
-     * This function obtains remote subvolumes and filters out only
-     * those running on the same node as the tier daemon.
-     */
-    if (strcmp(xl->type, "protocol/client") == 0) {
-        ret = dict_get_str(xl->options, "remote-host", &rh);
-        if (ret < 0)
-            goto out;
-
-        if (gf_is_local_addr(rh)) {
-            local_brick = GF_CALLOC(1, sizeof(tier_brick_list_t),
-                                    gf_tier_mt_bricklist_t);
-            if (!local_brick) {
-                goto out;
-            }
-
-            ret = dict_get_str(xl->options, "remote-subvolume", &rv);
-            if (ret < 0)
-                goto out;
-
-            brickname = strrchr(rv, '/') + 1;
-            snprintf(db_name, sizeof(db_name), "%s.db", brickname);
-
-            local_brick->brick_db_path = GF_MALLOC(PATH_MAX, gf_common_mt_char);
-            if (!local_brick->brick_db_path) {
-                gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_STATUS,
-                       "Failed to allocate memory for"
-                       " bricklist.");
-                ret = -1;
-                goto out;
-            }
-
-            len = snprintf(local_brick->brick_db_path, PATH_MAX, "%s/%s/%s", rv,
-                           GF_HIDDEN_PATH, db_name);
-            if ((len < 0) || (len >= PATH_MAX)) {
-                gf_msg("tier", GF_LOG_ERROR, EINVAL, DHT_MSG_LOG_TIER_STATUS,
-                       "DB path too long");
-                ret = -1;
-                goto out;
-            }
-
-            local_brick->xlator = xl;
-
-            snprintf(local_brick->brick_name, NAME_MAX, "%s", brickname);
-
-            list_add_tail(&(local_brick->list), local_bricklist_head);
-
-            ret = 0;
-            goto out;
-        }
-    }
-
-    for (child = xl->children; child; child = child->next) {
-        ret = tier_get_bricklist(child->xlator, local_bricklist_head);
-        if (ret) {
-            goto out;
-        }
-    }
-
-    ret = 0;
-out:
-
-    if (ret) {
-        if (local_brick) {
-            GF_FREE(local_brick->brick_db_path);
-        }
-        GF_FREE(local_brick);
-    }
-
-    return ret;
-}
-
-int
-tier_get_freq_demote(gf_tier_conf_t *tier_conf)
-{
-    if ((tier_conf->mode == TIER_MODE_WM) &&
-        (tier_conf->watermark_last == TIER_WM_HI))
-        return DEFAULT_DEMOTE_DEGRADED;
-    else
-        return tier_conf->tier_demote_frequency;
-}
-
-int
-tier_get_freq_promote(gf_tier_conf_t *tier_conf)
-{
-    return tier_conf->tier_promote_frequency;
-}
-
-int
-tier_get_freq_compact_hot(gf_tier_conf_t *tier_conf)
-{
-    return tier_conf->tier_compact_hot_frequency;
-}
-
-int
-tier_get_freq_compact_cold(gf_tier_conf_t *tier_conf)
-{
-    return tier_conf->tier_compact_cold_frequency;
-}
-
-static int
-tier_check_demote(gfdb_time_t current_time, int freq)
-{
-    return ((current_time.tv_sec % freq) == 0) ? _gf_true : _gf_false;
-}
-
-static gf_boolean_t
-tier_check_promote(gf_tier_conf_t *tier_conf, gfdb_time_t current_time,
-                   int freq)
-{
-    if ((tier_conf->mode == TIER_MODE_WM) &&
-        (tier_conf->watermark_last == TIER_WM_HI))
-        return _gf_false;
-
-    else
-        return ((current_time.tv_sec % freq) == 0) ? _gf_true : _gf_false;
-}
-
-static gf_boolean_t
-tier_check_compact(gf_tier_conf_t *tier_conf, gfdb_time_t current_time,
-                   int freq_compact)
-{
-    if (!(tier_conf->compact_active || tier_conf->compact_mode_switched))
-        return _gf_false;
-
-    return ((current_time.tv_sec % freq_compact) == 0) ? _gf_true : _gf_false;
-}
-
-void
-clear_bricklist(struct list_head *brick_list)
-{
-    tier_brick_list_t *local_brick = NULL;
-    tier_brick_list_t *temp = NULL;
-
-    if (list_empty(brick_list)) {
-        return;
-    }
-
-    list_for_each_entry_safe(local_brick, temp, brick_list, list)
-    {
-        list_del(&local_brick->list);
-        GF_FREE(local_brick->brick_db_path);
-        GF_FREE(local_brick);
-    }
-}
-
-static void
-set_brick_list_qpath(struct list_head *brick_list, gf_boolean_t is_cold)
-{
-    tier_brick_list_t *local_brick = NULL;
-    int i = 0;
-
-    GF_VALIDATE_OR_GOTO("tier", brick_list, out);
-
-    list_for_each_entry(local_brick, brick_list, list)
-    {
-        /* Construct query file path for this brick
-         * i.e
-         * /var/run/gluster/xlator_name/
-         * {promote/demote}-brickname-indexinbricklist
-         * So that no two query files will have same path even
-         * bricks have the same name
-         * */
-        snprintf(local_brick->qfile_path, PATH_MAX, "%s-%s-%d",
-                 GET_QFILE_PATH(is_cold), local_brick->brick_name, i);
-        i++;
-    }
-out:
-    return;
-}
-
-static int
-tier_prepare_compact(migration_args_t *args, gfdb_time_t current_time)
-{
-    xlator_t *this = NULL;
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    gf_tier_conf_t *tier_conf = NULL;
-    gf_boolean_t is_hot_tier = args->is_hot_tier;
-    int freq = 0;
-    int ret = -1;
-    const char *tier_type = is_hot_tier ? "hot" : "cold";
-
-    this = args->this;
-
-    conf = this->private;
-
-    defrag = conf->defrag;
-
-    tier_conf = &defrag->tier_conf;
-
-    freq = is_hot_tier ? tier_get_freq_compact_hot(tier_conf)
-                       : tier_get_freq_compact_cold(tier_conf);
-
-    defrag->tier_conf.compact_mode_switched =
-        is_hot_tier ? defrag->tier_conf.compact_mode_switched_hot
-                    : defrag->tier_conf.compact_mode_switched_cold;
-
-    gf_msg(this->name, GF_LOG_TRACE, 0, DHT_MSG_LOG_TIER_STATUS,
-           "Compact mode %i", defrag->tier_conf.compact_mode_switched);
-
-    if (tier_check_compact(tier_conf, current_time, freq)) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "Start compaction on %s tier", tier_type);
-
-        args->freq_time = freq;
-        ret = tier_compact(args);
-        if (ret) {
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "Compaction failed on "
-                   "%s tier",
-                   tier_type);
-            goto out;
-        }
-
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-               "End compaction on %s tier", tier_type);
-
-        if (is_hot_tier) {
-            defrag->tier_conf.compact_mode_switched_hot = _gf_false;
-        } else {
-            defrag->tier_conf.compact_mode_switched_cold = _gf_false;
-        }
-    }
-
-out:
-    return ret;
-}
-
-static int
-tier_get_wm_interval(tier_mode_t mode, tier_watermark_op_t wm)
-{
-    if (mode == TIER_MODE_WM && wm == TIER_WM_HI)
-        return WM_INTERVAL_EMERG;
-
-    return WM_INTERVAL;
-}
-
-/*
- * Main tiering loop. This is called from the promotion and the
- * demotion threads spawned in tier_start().
- *
- * Every second, wake from sleep to perform tasks.
- * 1. Check trigger to migrate data.
- * 2. Check for state changes (pause, unpause, stop).
- */
-static void *
-tier_run(void *in_args)
-{
-    dht_conf_t *conf = NULL;
-    gfdb_time_t current_time = {0};
-    int freq = 0;
-    int ret = 0;
-    xlator_t *any = NULL;
-    xlator_t *xlator = NULL;
-    gf_tier_conf_t *tier_conf = NULL;
-    loc_t root_loc = {0};
-    int check_watermark = 0;
-    gf_defrag_info_t *defrag = NULL;
-    xlator_t *this = NULL;
-    migration_args_t *args = in_args;
-    GF_VALIDATE_OR_GOTO("tier", args, out);
-    GF_VALIDATE_OR_GOTO("tier", args->brick_list, out);
-
-    this = args->this;
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-
-    conf = this->private;
-    GF_VALIDATE_OR_GOTO("tier", conf, out);
-
-    defrag = conf->defrag;
-    GF_VALIDATE_OR_GOTO("tier", defrag, out);
-
-    if (list_empty(args->brick_list)) {
-        gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Brick list for tier is empty. Exiting.");
-        goto out;
-    }
-
-    defrag->defrag_status = GF_DEFRAG_STATUS_STARTED;
-    tier_conf = &defrag->tier_conf;
-
-    dht_build_root_loc(defrag->root_inode, &root_loc);
-
-    while (1) {
-        /*
-         * Check if a graph switch occurred. If so, stop migration
-         * thread. It will need to be restarted manually.
-         */
-        any = THIS->ctx->active->first;
-        xlator = xlator_search_by_name(any, this->name);
-
-        if (xlator != this) {
-            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                   "Detected graph switch. Exiting migration "
-                   "daemon.");
-            goto out;
-        }
-
-        gf_defrag_check_pause_tier(tier_conf);
-
-        sleep(1);
-
-        if (defrag->defrag_status != GF_DEFRAG_STATUS_STARTED) {
-            ret = 1;
-            gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "defrag->defrag_status != "
-                   "GF_DEFRAG_STATUS_STARTED");
-            goto out;
-        }
-
-        if (defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER ||
-            defrag->cmd == GF_DEFRAG_CMD_DETACH_START) {
-            ret = 0;
-            defrag->defrag_status = GF_DEFRAG_STATUS_COMPLETE;
-            gf_msg(this->name, GF_LOG_DEBUG, 0, DHT_MSG_LOG_TIER_ERROR,
-                   "defrag->defrag_cmd == "
-                   "GF_DEFRAG_CMD_START_DETACH_TIER");
-            goto out;
-        }
-
-        if (gf_defrag_get_pause_state(&defrag->tier_conf) != TIER_RUNNING)
-            continue;
-
-        /* To have proper synchronization amongst all
-         * brick holding nodes, so that promotion and demotions
-         * start atomically w.r.t promotion/demotion frequency
-         * period, all nodes should have their system time
-         * in-sync with each other either manually set or
-         * using a NTP server*/
-        ret = gettimeofday(&current_time, NULL);
-        if (ret == -1) {
-            gf_msg(this->name, GF_LOG_ERROR, errno,
-                   DHT_MSG_SYS_CALL_GET_TIME_FAILED,
-                   "Failed to get current time");
-            goto out;
-        }
-
-        check_watermark++;
-
-        /* emergency demotion requires frequent watermark monitoring */
-        if (check_watermark >=
-            tier_get_wm_interval(tier_conf->mode, tier_conf->watermark_last)) {
-            check_watermark = 0;
-            if (tier_conf->mode == TIER_MODE_WM) {
-                ret = tier_get_fs_stat(this, &root_loc);
-                if (ret != 0) {
-                    continue;
-                }
-                ret = tier_check_watermark(this);
-                if (ret != 0) {
-                    gf_msg(this->name, GF_LOG_CRITICAL, errno,
-                           DHT_MSG_LOG_TIER_ERROR, "Failed to get watermark");
-                    continue;
-                }
-            }
-        }
-
-        if (args->is_promotion) {
-            freq = tier_get_freq_promote(tier_conf);
-
-            if (tier_check_promote(tier_conf, current_time, freq)) {
-                args->freq_time = freq;
-                ret = tier_promote(args);
-                if (ret) {
-                    gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                           "Promotion failed");
-                }
-            }
-        } else if (args->is_compaction) {
-            tier_prepare_compact(args, current_time);
-        } else {
-            freq = tier_get_freq_demote(tier_conf);
-
-            if (tier_check_demote(current_time, freq)) {
-                args->freq_time = freq;
-                ret = tier_demote(args);
-                if (ret) {
-                    gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                           "Demotion failed");
-                }
-            }
-        }
-
-        /* Check the statfs immediately after the processing threads
-           return */
-        check_watermark = WM_INTERVAL;
-    }
-
-    ret = 0;
-out:
-
-    args->return_value = ret;
-
-    return NULL;
-}
-
-int
-tier_start(xlator_t *this, gf_defrag_info_t *defrag)
-{
-    pthread_t promote_thread;
-    pthread_t demote_thread;
-    pthread_t hot_compact_thread;
-    pthread_t cold_compact_thread;
-    int ret = -1;
-    struct list_head bricklist_hot = {0};
-    struct list_head bricklist_cold = {0};
-    migration_args_t promotion_args = {0};
-    migration_args_t demotion_args = {0};
-    migration_args_t hot_compaction_args = {0};
-    migration_args_t cold_compaction_args = {0};
-    dht_conf_t *conf = NULL;
-
-    INIT_LIST_HEAD((&bricklist_hot));
-    INIT_LIST_HEAD((&bricklist_cold));
-
-    conf = this->private;
-
-    tier_get_bricklist(conf->subvolumes[1], &bricklist_hot);
-    set_brick_list_qpath(&bricklist_hot, _gf_false);
-
-    demotion_args.this = this;
-    demotion_args.brick_list = &bricklist_hot;
-    demotion_args.defrag = defrag;
-    demotion_args.is_promotion = _gf_false;
-    demotion_args.is_compaction = _gf_false;
-
-    ret = gf_thread_create(&demote_thread, NULL, &tier_run, &demotion_args,
-                           "tierdem");
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to start demotion thread.");
-        defrag->defrag_status = GF_DEFRAG_STATUS_FAILED;
-        goto cleanup;
-    }
-
-    tier_get_bricklist(conf->subvolumes[0], &bricklist_cold);
-    set_brick_list_qpath(&bricklist_cold, _gf_true);
-
-    promotion_args.this = this;
-    promotion_args.brick_list = &bricklist_cold;
-    promotion_args.defrag = defrag;
-    promotion_args.is_promotion = _gf_true;
-
-    ret = gf_thread_create(&promote_thread, NULL, &tier_run, &promotion_args,
-                           "tierpro");
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to start promotion thread.");
-        defrag->defrag_status = GF_DEFRAG_STATUS_FAILED;
-        goto waitforspawned;
-    }
-
-    hot_compaction_args.this = this;
-    hot_compaction_args.brick_list = &bricklist_hot;
-    hot_compaction_args.defrag = defrag;
-    hot_compaction_args.is_promotion = _gf_false;
-    hot_compaction_args.is_compaction = _gf_true;
-    hot_compaction_args.is_hot_tier = _gf_true;
-
-    ret = gf_thread_create(&hot_compact_thread, NULL, &tier_run,
-                           &hot_compaction_args, "tierhcom");
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to start compaction thread.");
-        defrag->defrag_status = GF_DEFRAG_STATUS_FAILED;
-        goto waitforspawnedpromote;
-    }
-
-    cold_compaction_args.this = this;
-    cold_compaction_args.brick_list = &bricklist_cold;
-    cold_compaction_args.defrag = defrag;
-    cold_compaction_args.is_promotion = _gf_false;
-    cold_compaction_args.is_compaction = _gf_true;
-    cold_compaction_args.is_hot_tier = _gf_false;
-
-    ret = gf_thread_create(&cold_compact_thread, NULL, &tier_run,
-                           &cold_compaction_args, "tierccom");
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Failed to start compaction thread.");
-        defrag->defrag_status = GF_DEFRAG_STATUS_FAILED;
-        goto waitforspawnedhotcompact;
-    }
-    pthread_join(cold_compact_thread, NULL);
-
-waitforspawnedhotcompact:
-    pthread_join(hot_compact_thread, NULL);
-
-waitforspawnedpromote:
-    pthread_join(promote_thread, NULL);
-
-waitforspawned:
-    pthread_join(demote_thread, NULL);
-
-cleanup:
-    clear_bricklist(&bricklist_cold);
-    clear_bricklist(&bricklist_hot);
-    return ret;
-}
-
-int32_t
-tier_migration_needed(xlator_t *this)
-{
-    gf_defrag_info_t *defrag = NULL;
-    dht_conf_t *conf = NULL;
-    int ret = 0;
-
-    conf = this->private;
-
-    GF_VALIDATE_OR_GOTO(this->name, conf, out);
-    GF_VALIDATE_OR_GOTO(this->name, conf->defrag, out);
-
-    defrag = conf->defrag;
-
-    if ((defrag->cmd == GF_DEFRAG_CMD_START_TIER) ||
-        (defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER))
-        ret = 1;
-out:
-    return ret;
-}
-
-int32_t
-tier_migration_get_dst(xlator_t *this, dht_local_t *local)
-{
-    dht_conf_t *conf = NULL;
-    int32_t ret = -1;
-    gf_defrag_info_t *defrag = NULL;
-
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-    GF_VALIDATE_OR_GOTO(this->name, this->private, out);
-
-    conf = this->private;
-
-    defrag = conf->defrag;
-
-    if (defrag && defrag->cmd == GF_DEFRAG_CMD_START_DETACH_TIER) {
-        local->rebalance.target_node = conf->subvolumes[0];
-
-    } else if (conf->subvolumes[0] == local->cached_subvol)
-        local->rebalance.target_node = conf->subvolumes[1];
-    else
-        local->rebalance.target_node = conf->subvolumes[0];
-
-    if (local->rebalance.target_node)
-        ret = 0;
-
-out:
-    return ret;
-}
-
-xlator_t *
-tier_search(xlator_t *this, dht_layout_t *layout, const char *name)
-{
-    xlator_t *subvol = NULL;
-    dht_conf_t *conf = NULL;
-
-    GF_VALIDATE_OR_GOTO("tier", this, out);
-    GF_VALIDATE_OR_GOTO(this->name, this->private, out);
-
-    conf = this->private;
-
-    subvol = TIER_HASHED_SUBVOL;
-
-out:
-    return subvol;
-}
-
-static int
-tier_load_externals(xlator_t *this)
-{
-    int ret = -1;
-    char *libpathfull = (LIBDIR "/libgfdb.so.0");
-    get_gfdb_methods_t get_gfdb_methods;
-
-    GF_VALIDATE_OR_GOTO("this", this, out);
-
-    libhandle = dlopen(libpathfull, RTLD_NOW);
-    if (!libhandle) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Error loading libgfdb.so %s\n", dlerror());
-        ret = -1;
-        goto out;
-    }
-
-    get_gfdb_methods = dlsym(libhandle, "get_gfdb_methods");
-    if (!get_gfdb_methods) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Error loading get_gfdb_methods()");
-        ret = -1;
-        goto out;
-    }
-
-    get_gfdb_methods(&gfdb_methods);
-
-    ret = 0;
-
-out:
-    if (ret && libhandle)
-        dlclose(libhandle);
-
-    return ret;
-}
-
-static tier_mode_t
-tier_validate_mode(char *mode)
-{
-    int ret = -1;
-
-    if (strcmp(mode, "test") == 0) {
-        ret = TIER_MODE_TEST;
-    } else {
-        ret = TIER_MODE_WM;
-    }
-
-    return ret;
-}
-
-static gf_boolean_t
-tier_validate_compact_mode(char *mode)
-{
-    gf_boolean_t ret = _gf_false;
-
-    gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-           "tier_validate_compact_mode: mode = %s", mode);
-
-    if (!strcmp(mode, "on")) {
-        ret = _gf_true;
-    } else {
-        ret = _gf_false;
-    }
-
-    gf_msg("tier", GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_STATUS,
-           "tier_validate_compact_mode: ret = %i", ret);
-
-    return ret;
-}
-
-int
-tier_init_methods(xlator_t *this)
-{
-    int ret = -1;
-    dht_conf_t *conf = NULL;
-    dht_methods_t *methods = NULL;
-
-    GF_VALIDATE_OR_GOTO("tier", this, err);
-
-    conf = this->private;
-
-    methods = &(conf->methods);
-
-    methods->migration_get_dst_subvol = tier_migration_get_dst;
-    methods->migration_other = tier_start;
-    methods->migration_needed = tier_migration_needed;
-    methods->layout_search = tier_search;
-
-    ret = 0;
-err:
-    return ret;
-}
-
-static void
-tier_save_vol_name(xlator_t *this)
-{
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    char *suffix = NULL;
-    int name_len = 0;
-
-    conf = this->private;
-    defrag = conf->defrag;
-
-    suffix = strstr(this->name, "-tier-dht");
-
-    if (suffix)
-        name_len = suffix - this->name;
-    else
-        name_len = strlen(this->name);
-
-    if (name_len > GD_VOLUME_NAME_MAX)
-        name_len = GD_VOLUME_NAME_MAX;
-
-    strncpy(defrag->tier_conf.volname, this->name, name_len);
-    defrag->tier_conf.volname[name_len] = 0;
-}
-
-int
-tier_init(xlator_t *this)
-{
-    int ret = -1;
-    int freq = 0;
-    int maxsize = 0;
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    char *voldir = NULL;
-    char *mode = NULL;
-    char *paused = NULL;
-    tier_mode_t tier_mode = DEFAULT_TIER_MODE;
-    gf_boolean_t compact_mode = _gf_false;
-
-    ret = dht_init(this);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "tier_init failed");
-        goto out;
-    }
-
-    conf = this->private;
-
-    ret = tier_init_methods(this);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "tier_init_methods failed");
-        goto out;
-    }
-
-    if (conf->subvolume_cnt != 2) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Invalid number of subvolumes %d", conf->subvolume_cnt);
-        goto out;
-    }
-
-    /* if instatiated from client side initialization is complete. */
-    if (!conf->defrag) {
-        ret = 0;
-        goto out;
-    }
-
-    /* if instatiated from server side, load db libraries */
-    ret = tier_load_externals(this);
-    if (ret) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "Could not load externals. Aborting");
-        goto out;
-    }
-
-    defrag = conf->defrag;
-
-    defrag->tier_conf.last_demote_qfile_index = 0;
-    defrag->tier_conf.last_promote_qfile_index = 0;
-
-    defrag->tier_conf.is_tier = 1;
-    defrag->this = this;
-
-    ret = dict_get_int32(this->options, "tier-max-promote-file-size", &maxsize);
-    if (ret) {
-        maxsize = 0;
-    }
-
-    defrag->tier_conf.tier_max_promote_size = maxsize;
-
-    ret = dict_get_int32(this->options, "tier-promote-frequency", &freq);
-    if (ret) {
-        freq = DEFAULT_PROMOTE_FREQ_SEC;
-    }
-
-    defrag->tier_conf.tier_promote_frequency = freq;
-
-    ret = dict_get_int32(this->options, "tier-demote-frequency", &freq);
-    if (ret) {
-        freq = DEFAULT_DEMOTE_FREQ_SEC;
-    }
-
-    defrag->tier_conf.tier_demote_frequency = freq;
-
-    ret = dict_get_int32(this->options, "tier-hot-compact-frequency", &freq);
-    if (ret) {
-        freq = DEFAULT_HOT_COMPACT_FREQ_SEC;
-    }
-
-    defrag->tier_conf.tier_compact_hot_frequency = freq;
-
-    ret = dict_get_int32(this->options, "tier-cold-compact-frequency", &freq);
-    if (ret) {
-        freq = DEFAULT_COLD_COMPACT_FREQ_SEC;
-    }
-
-    defrag->tier_conf.tier_compact_cold_frequency = freq;
-
-    ret = dict_get_int32(this->options, "watermark-hi", &freq);
-    if (ret) {
-        freq = DEFAULT_WM_HI;
-    }
-
-    defrag->tier_conf.watermark_hi = freq;
-
-    ret = dict_get_int32(this->options, "watermark-low", &freq);
-    if (ret) {
-        freq = DEFAULT_WM_LOW;
-    }
-
-    defrag->tier_conf.watermark_low = freq;
-
-    ret = dict_get_int32(this->options, "write-freq-threshold", &freq);
-    if (ret) {
-        freq = DEFAULT_WRITE_FREQ_SEC;
-    }
-
-    defrag->write_freq_threshold = freq;
-
-    ret = dict_get_int32(this->options, "read-freq-threshold", &freq);
-    if (ret) {
-        freq = DEFAULT_READ_FREQ_SEC;
-    }
-
-    defrag->read_freq_threshold = freq;
-
-    ret = dict_get_int32(this->options, "tier-max-mb", &freq);
-    if (ret) {
-        freq = DEFAULT_TIER_MAX_MIGRATE_MB;
-    }
-
-    defrag->tier_conf.max_migrate_bytes = (uint64_t)freq * 1024 * 1024;
-
-    ret = dict_get_int32(this->options, "tier-max-files", &freq);
-    if (ret) {
-        freq = DEFAULT_TIER_MAX_MIGRATE_FILES;
-    }
-
-    defrag->tier_conf.max_migrate_files = freq;
-
-    ret = dict_get_int32(this->options, "tier-query-limit",
-                         &(defrag->tier_conf.query_limit));
-    if (ret) {
-        defrag->tier_conf.query_limit = DEFAULT_TIER_QUERY_LIMIT;
-    }
-
-    ret = dict_get_str(this->options, "tier-compact", &mode);
-
-    if (ret) {
-        defrag->tier_conf.compact_active = DEFAULT_COMP_MODE;
-    } else {
-        compact_mode = tier_validate_compact_mode(mode);
-        /* If compaction is now active, we need to inform the bricks on
-           the hot and cold tier of this. See dht-common.h for more. */
-        defrag->tier_conf.compact_active = compact_mode;
-        if (compact_mode) {
-            defrag->tier_conf.compact_mode_switched_hot = _gf_true;
-            defrag->tier_conf.compact_mode_switched_cold = _gf_true;
-        }
-    }
-
-    ret = dict_get_str(this->options, "tier-mode", &mode);
-    if (ret) {
-        defrag->tier_conf.mode = DEFAULT_TIER_MODE;
-    } else {
-        tier_mode = tier_validate_mode(mode);
-        defrag->tier_conf.mode = tier_mode;
-    }
-
-    pthread_mutex_init(&defrag->tier_conf.pause_mutex, 0);
-
-    gf_defrag_set_pause_state(&defrag->tier_conf, TIER_RUNNING);
-
-    ret = dict_get_str(this->options, "tier-pause", &paused);
-
-    if (paused && strcmp(paused, "on") == 0)
-        gf_defrag_set_pause_state(&defrag->tier_conf, TIER_REQUEST_PAUSE);
-
-    ret = gf_asprintf(&voldir, "%s/%s", DEFAULT_VAR_RUN_DIRECTORY, this->name);
-    if (ret < 0)
-        goto out;
-
-    ret = mkdir_p(voldir, 0777, _gf_true);
-    if (ret == -1 && errno != EEXIST) {
-        gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-               "tier_init failed");
-
-        GF_FREE(voldir);
-        goto out;
-    }
-
-    GF_FREE(voldir);
-
-    ret = gf_asprintf(&promotion_qfile, "%s/%s/promote",
-                      DEFAULT_VAR_RUN_DIRECTORY, this->name);
-    if (ret < 0)
-        goto out;
-
-    ret = gf_asprintf(&demotion_qfile, "%s/%s/demote",
-                      DEFAULT_VAR_RUN_DIRECTORY, this->name);
-    if (ret < 0) {
-        GF_FREE(promotion_qfile);
-        goto out;
-    }
-
-    gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-           "Promote/demote frequency %d/%d "
-           "Write/Read freq thresholds %d/%d",
-           defrag->tier_conf.tier_promote_frequency,
-           defrag->tier_conf.tier_demote_frequency,
-           defrag->write_freq_threshold, defrag->read_freq_threshold);
-
-    tier_save_vol_name(this);
-
-    ret = 0;
-
-out:
-
-    return ret;
-}
-
-int
-tier_cli_pause_done(int op_ret, call_frame_t *sync_frame, void *data)
-{
-    gf_msg("tier", GF_LOG_INFO, 0, DHT_MSG_TIER_PAUSED,
-           "Migrate file paused with op_ret %d", op_ret);
-
-    return op_ret;
-}
-
-int
-tier_cli_pause(void *data)
-{
-    gf_defrag_info_t *defrag = NULL;
-    xlator_t *this = NULL;
-    dht_conf_t *conf = NULL;
-    int ret = -1;
-
-    this = data;
-
-    conf = this->private;
-    GF_VALIDATE_OR_GOTO(this->name, conf, exit);
-
-    defrag = conf->defrag;
-    GF_VALIDATE_OR_GOTO(this->name, defrag, exit);
-
-    gf_defrag_pause_tier(this, defrag);
-
-    ret = 0;
-exit:
-    return ret;
-}
-
-int
-tier_reconfigure(xlator_t *this, dict_t *options)
-{
-    dht_conf_t *conf = NULL;
-    gf_defrag_info_t *defrag = NULL;
-    char *mode = NULL;
-    int migrate_mb = 0;
-    gf_boolean_t req_pause = _gf_false;
-    int ret = 0;
-    call_frame_t *frame = NULL;
-    gf_boolean_t last_compact_setting = _gf_false;
-
-    conf = this->private;
-
-    if (conf->defrag) {
-        defrag = conf->defrag;
-        GF_OPTION_RECONF("tier-max-promote-file-size",
-                         defrag->tier_conf.tier_max_promote_size, options,
-                         int32, out);
-
-        GF_OPTION_RECONF("tier-promote-frequency",
-                         defrag->tier_conf.tier_promote_frequency, options,
-                         int32, out);
-
-        GF_OPTION_RECONF("tier-demote-frequency",
-                         defrag->tier_conf.tier_demote_frequency, options,
-                         int32, out);
-
-        GF_OPTION_RECONF("write-freq-threshold", defrag->write_freq_threshold,
-                         options, int32, out);
-
-        GF_OPTION_RECONF("read-freq-threshold", defrag->read_freq_threshold,
-                         options, int32, out);
-
-        GF_OPTION_RECONF("watermark-hi", defrag->tier_conf.watermark_hi,
-                         options, int32, out);
-
-        GF_OPTION_RECONF("watermark-low", defrag->tier_conf.watermark_low,
-                         options, int32, out);
-
-        last_compact_setting = defrag->tier_conf.compact_active;
-
-        GF_OPTION_RECONF("tier-compact", defrag->tier_conf.compact_active,
-                         options, bool, out);
-
-        if (last_compact_setting != defrag->tier_conf.compact_active) {
-            defrag->tier_conf.compact_mode_switched_hot = _gf_true;
-            defrag->tier_conf.compact_mode_switched_cold = _gf_true;
-            gf_msg(this->name, GF_LOG_INFO, 0, DHT_MSG_LOG_TIER_STATUS,
-                   "compact mode switched");
-        }
-
-        GF_OPTION_RECONF("tier-hot-compact-frequency",
-                         defrag->tier_conf.tier_compact_hot_frequency, options,
-                         int32, out);
-
-        GF_OPTION_RECONF("tier-cold-compact-frequency",
-                         defrag->tier_conf.tier_compact_cold_frequency, options,
-                         int32, out);
-
-        GF_OPTION_RECONF("tier-mode", mode, options, str, out);
-        defrag->tier_conf.mode = tier_validate_mode(mode);
-
-        GF_OPTION_RECONF("tier-max-mb", migrate_mb, options, int32, out);
-        defrag->tier_conf.max_migrate_bytes = (uint64_t)migrate_mb * 1024 *
-                                              1024;
-
-        GF_OPTION_RECONF("tier-max-files", defrag->tier_conf.max_migrate_files,
-                         options, int32, out);
-
-        GF_OPTION_RECONF("tier-query-limit", defrag->tier_conf.query_limit,
-                         options, int32, out);
-
-        GF_OPTION_RECONF("tier-pause", req_pause, options, bool, out);
-
-        if (req_pause == _gf_true) {
-            frame = create_frame(this, this->ctx->pool);
-            if (!frame)
-                goto out;
-
-            frame->root->pid = GF_CLIENT_PID_DEFRAG;
-
-            ret = synctask_new(this->ctx->env, tier_cli_pause,
-                               tier_cli_pause_done, frame, this);
-
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                       "pause tier failed on reconfigure");
-            }
-        } else {
-            ret = gf_defrag_resume_tier(this, defrag);
-            if (ret) {
-                gf_msg(this->name, GF_LOG_ERROR, 0, DHT_MSG_LOG_TIER_ERROR,
-                       "resume tier failed on reconfigure");
-            }
-        }
-    }
-
-out:
-    return dht_reconfigure(this, options);
-}
-
-void
-tier_fini(xlator_t *this)
-{
-    if (libhandle)
-        dlclose(libhandle);
-
-    GF_FREE(demotion_qfile);
-    GF_FREE(promotion_qfile);
-
-    dht_fini(this);
-}
-
-class_methods_t class_methods = {.init = tier_init,
-                                 .fini = tier_fini,
-                                 .reconfigure = tier_reconfigure,
-                                 .notify = dht_notify};
-
-struct xlator_fops fops = {
-
-    .lookup = dht_lookup,
-    .create = tier_create,
-    .mknod = dht_mknod,
-
-    .open = dht_open,
-    .statfs = tier_statfs,
-    .opendir = dht_opendir,
-    .readdir = tier_readdir,
-    .readdirp = tier_readdirp,
-    .fsyncdir = dht_fsyncdir,
-    .symlink = dht_symlink,
-    .unlink = tier_unlink,
-    .link = tier_link,
-    .mkdir = dht_mkdir,
-    .rmdir = dht_rmdir,
-    .rename = dht_rename,
-    .entrylk = dht_entrylk,
-    .fentrylk = dht_fentrylk,
-
-    /* Inode read operations */
-    .stat = dht_stat,
-    .fstat = dht_fstat,
-    .access = dht_access,
-    .readlink = dht_readlink,
-    .getxattr = dht_getxattr,
-    .fgetxattr = dht_fgetxattr,
-    .readv = dht_readv,
-    .flush = dht_flush,
-    .fsync = dht_fsync,
-    .inodelk = dht_inodelk,
-    .finodelk = dht_finodelk,
-    .lk = dht_lk,
-
-    /* Inode write operations */
-    .fremovexattr = dht_fremovexattr,
-    .removexattr = dht_removexattr,
-    .setxattr = dht_setxattr,
-    .fsetxattr = dht_fsetxattr,
-    .truncate = dht_truncate,
-    .ftruncate = dht_ftruncate,
-    .writev = dht_writev,
-    .xattrop = dht_xattrop,
-    .fxattrop = dht_fxattrop,
-    .setattr = dht_setattr,
-    .fsetattr = dht_fsetattr,
-    .fallocate = dht_fallocate,
-    .discard = dht_discard,
-    .zerofill = dht_zerofill,
-};
-
-struct xlator_cbks cbks = {.release = dht_release, .forget = dht_forget};
diff --git a/xlators/cluster/dht/src/tier.h b/xlators/cluster/dht/src/tier.h
deleted file mode 100644
index f0ffdfcd769..00000000000
--- a/xlators/cluster/dht/src/tier.h
+++ /dev/null
@@ -1,110 +0,0 @@
-/*
-  Copyright (c) 2015 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#ifndef _TIER_H_
-#define _TIER_H_
-
-/******************************************************************************/
-/* This is from dht-rebalancer.c as we don't have dht-rebalancer.h */
-#include "dht-common.h"
-#include "xlator.h"
-#include <signal.h>
-#include <fnmatch.h>
-#include <signal.h>
-
-/*
- * Size of timer wheel. We would not promote or demote less
- * frequently than this number.
- */
-#define TIMER_SECS 3600
-
-#include "gfdb_data_store.h"
-#include <ctype.h>
-#include <sys/stat.h>
-
-#define PROMOTION_QFILE "promotequeryfile"
-#define DEMOTION_QFILE "demotequeryfile"
-
-#define TIER_HASHED_SUBVOL conf->subvolumes[0]
-#define TIER_UNHASHED_SUBVOL conf->subvolumes[1]
-
-#define GET_QFILE_PATH(is_promotion)                                           \
-    (is_promotion) ? promotion_qfile : demotion_qfile
-
-typedef struct tier_qfile_array {
-    int *fd_array;
-    ssize_t array_size;
-    ssize_t next_index;
-    /* Indicate the number of exhuasted FDs*/
-    ssize_t exhausted_count;
-} tier_qfile_array_t;
-
-typedef struct _query_cbk_args {
-    xlator_t *this;
-    gf_defrag_info_t *defrag;
-    /* This is write */
-    int query_fd;
-    int is_promotion;
-    int is_compaction;
-    /* This is for read */
-    tier_qfile_array_t *qfile_array;
-} query_cbk_args_t;
-
-int
-gf_run_tier(xlator_t *this, gf_defrag_info_t *defrag);
-
-typedef struct gfdb_brick_info {
-    gfdb_time_t *time_stamp;
-    gf_boolean_t _gfdb_promote;
-    query_cbk_args_t *_query_cbk_args;
-} gfdb_brick_info_t;
-
-typedef struct brick_list {
-    xlator_t *xlator;
-    char *brick_db_path;
-    char brick_name[NAME_MAX];
-    char qfile_path[PATH_MAX];
-    struct list_head list;
-} tier_brick_list_t;
-
-typedef struct _dm_thread_args {
-    xlator_t *this;
-    gf_defrag_info_t *defrag;
-    struct list_head *brick_list;
-    int freq_time;
-    int return_value;
-    int is_promotion;
-    int is_compaction;
-    gf_boolean_t is_hot_tier;
-} migration_args_t;
-
-typedef enum tier_watermark_op_ {
-    TIER_WM_NONE = 0,
-    TIER_WM_LOW,
-    TIER_WM_HI,
-    TIER_WM_MID
-} tier_watermark_op_t;
-
-#define DEFAULT_PROMOTE_FREQ_SEC 120
-#define DEFAULT_DEMOTE_FREQ_SEC 120
-#define DEFAULT_HOT_COMPACT_FREQ_SEC 604800
-#define DEFAULT_COLD_COMPACT_FREQ_SEC 604800
-#define DEFAULT_DEMOTE_DEGRADED 1
-#define DEFAULT_WRITE_FREQ_SEC 0
-#define DEFAULT_READ_FREQ_SEC 0
-#define DEFAULT_WM_LOW 75
-#define DEFAULT_WM_HI 90
-#define DEFAULT_TIER_MODE TIER_MODE_TEST
-#define DEFAULT_COMP_MODE _gf_true
-#define DEFAULT_TIER_MAX_MIGRATE_MB 1000
-#define DEFAULT_TIER_MAX_MIGRATE_FILES 5000
-#define DEFAULT_TIER_QUERY_LIMIT 100
-
-#endif
diff --git a/xlators/cluster/dht/src/tier.sym b/xlators/cluster/dht/src/tier.sym
deleted file mode 100644
index 60205d145b6..00000000000
--- a/xlators/cluster/dht/src/tier.sym
+++ /dev/null
@@ -1,9 +0,0 @@
-fops
-cbks
-class_methods
-dht_methods
-tier_methods
-options
-mem_acct_init
-reconfigure
-dumpops
diff --git a/xlators/cluster/dht/src/unittest/dht_layout_mock.c b/xlators/cluster/dht/src/unittest/dht_layout_mock.c
index 49bf18b9fe6..771452963d1 100644
--- a/xlators/cluster/dht/src/unittest/dht_layout_mock.c
+++ b/xlators/cluster/dht/src/unittest/dht_layout_mock.c
@@ -7,10 +7,10 @@
   later), or the GNU General Public License, version 2 (GPLv2), in all
   cases as published by the Free Software Foundation.
 */
-#include "glusterfs.h"
-#include "xlator.h"
+#include <glusterfs/glusterfs.h>
+#include <glusterfs/xlator.h>
 #include "dht-common.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 
 int
 dht_hash_compute(xlator_t *this, int type, const char *name, uint32_t *hash_p)
diff --git a/xlators/cluster/dht/src/unittest/dht_layout_unittest.c b/xlators/cluster/dht/src/unittest/dht_layout_unittest.c
index 72890070835..c94a1d0a2e1 100644
--- a/xlators/cluster/dht/src/unittest/dht_layout_unittest.c
+++ b/xlators/cluster/dht/src/unittest/dht_layout_unittest.c
@@ -9,8 +9,8 @@
 */
 
 #include "dht-common.h"
-#include "logging.h"
-#include "xlator.h"
+#include <glusterfs/logging.h>
+#include <glusterfs/xlator.h>
 
 #include <inttypes.h>
 #include <stdarg.h>
diff --git a/xlators/cluster/ec/src/ec-code.c b/xlators/cluster/ec/src/ec-code.c
index b3338248569..03162ae05a9 100644
--- a/xlators/cluster/ec/src/ec-code.c
+++ b/xlators/cluster/ec/src/ec-code.c
@@ -14,7 +14,7 @@
 #include <sys/stat.h>
 #include <ctype.h>
 
-#include "syscall.h"
+#include <glusterfs/syscall.h>
 
 #include "ec-mem-types.h"
 #include "ec-code.h"
diff --git a/xlators/cluster/ec/src/ec-code.h b/xlators/cluster/ec/src/ec-code.h
index 355209c3944..75fb35d93e3 100644
--- a/xlators/cluster/ec/src/ec-code.h
+++ b/xlators/cluster/ec/src/ec-code.h
@@ -11,8 +11,8 @@
 #ifndef __EC_CODE_H__
 #define __EC_CODE_H__
 
-#include "xlator.h"
-#include "list.h"
+#include <glusterfs/xlator.h>
+#include <glusterfs/list.h>
 
 #include "ec-types.h"
 #include "ec-galois.h"
diff --git a/xlators/cluster/ec/src/ec-combine.c b/xlators/cluster/ec/src/ec-combine.c
index 454c2c72c95..703a30e2485 100644
--- a/xlators/cluster/ec/src/ec-combine.c
+++ b/xlators/cluster/ec/src/ec-combine.c
@@ -11,14 +11,14 @@
 #include <fnmatch.h>
 
 #include "libxlator.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 
 #include "ec-types.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
 #include "ec-messages.h"
-#include "quota-common-utils.h"
+#include <glusterfs/quota-common-utils.h>
 
 #define EC_QUOTA_PREFIX "trusted.glusterfs.quota."
 
@@ -179,13 +179,14 @@ ec_iatt_combine(ec_fop_data_t *fop, struct iatt *dst, struct iatt *src,
                    "links: %u-%u, uid: %u-%u, gid: %u-%u, "
                    "rdev: %" PRIu64 "-%" PRIu64 ", size: %" PRIu64 "-%" PRIu64
                    ", "
-                   "mode: %o-%o)",
+                   "mode: %o-%o), %s",
                    dst[i].ia_ino, src[i].ia_ino, dst[i].ia_nlink,
                    src[i].ia_nlink, dst[i].ia_uid, src[i].ia_uid, dst[i].ia_gid,
                    src[i].ia_gid, dst[i].ia_rdev, src[i].ia_rdev,
                    dst[i].ia_size, src[i].ia_size,
                    st_mode_from_ia(dst[i].ia_prot, dst[i].ia_type),
-                   st_mode_from_ia(src[i].ia_prot, dst[i].ia_type));
+                   st_mode_from_ia(src[i].ia_prot, dst[i].ia_type),
+                   ec_msg_str(fop));
 
             return 0;
         }
@@ -342,9 +343,8 @@ out:
 }
 
 static int32_t
-ec_dict_data_concat(const char *fmt, ec_cbk_data_t *cbk, int32_t which,
-                    char *key, char *new_key, const char *def,
-                    gf_boolean_t global, ...)
+ec_dict_data_concat(ec_cbk_data_t *cbk, int32_t which, char *key, char *new_key,
+                    const char *def, gf_boolean_t global, const char *fmt, ...)
 {
     ec_t *ec = cbk->fop->xl->private;
     data_t *data[ec->nodes];
@@ -356,7 +356,7 @@ ec_dict_data_concat(const char *fmt, ec_cbk_data_t *cbk, int32_t which,
 
     ec_dict_list(data, cbk, which, key, global);
 
-    va_start(args, global);
+    va_start(args, fmt);
     err = ec_concat_prepare(cbk->fop->xl, &pre, &sep, &post, fmt, args);
     va_end(args);
 
@@ -485,22 +485,12 @@ ec_dict_data_merge(ec_cbk_data_t *cbk, int32_t which, char *key)
 
     tmp = NULL;
 
-    len = dict_serialized_length(lockinfo);
-    if (len < 0) {
-        err = len;
-
-        goto out;
-    }
-    ptr = GF_MALLOC(len, gf_common_mt_char);
-    if (ptr == NULL) {
-        err = -ENOMEM;
-
-        goto out;
-    }
-    err = dict_serialize(lockinfo, ptr);
+    err = dict_allocate_and_serialize(lockinfo, (char **)&ptr,
+                                      (unsigned int *)&len);
     if (err != 0) {
         goto out;
     }
+
     dict = (which == EC_COMBINE_XDATA) ? cbk->xdata : cbk->dict;
     err = dict_set_dynptr(dict, key, ptr, len);
     if (err != 0) {
@@ -739,14 +729,14 @@ ec_dict_data_combine(dict_t *dict, char *key, data_t *value, void *arg)
 
     if ((strcmp(key, GF_XATTR_PATHINFO_KEY) == 0) ||
         (strcmp(key, GF_XATTR_USER_PATHINFO_KEY) == 0)) {
-        return ec_dict_data_concat("(<EC:%s> { })", data->cbk, data->which, key,
-                                   NULL, NULL, _gf_false,
+        return ec_dict_data_concat(data->cbk, data->which, key, NULL, NULL,
+                                   _gf_false, _gf_false, "(<EC:%s> { })",
                                    data->cbk->fop->xl->name);
     }
 
     if (strncmp(key, GF_XATTR_CLRLK_CMD, SLEN(GF_XATTR_CLRLK_CMD)) == 0) {
-        return ec_dict_data_concat("{\n}", data->cbk, data->which, key, NULL,
-                                   NULL, _gf_false);
+        return ec_dict_data_concat(data->cbk, data->which, key, NULL, NULL,
+                                   _gf_false, "{\n}");
     }
 
     if (strncmp(key, GF_XATTR_LOCKINFO_KEY, SLEN(GF_XATTR_LOCKINFO_KEY)) == 0) {
@@ -776,9 +766,9 @@ ec_dict_data_combine(dict_t *dict, char *key, data_t *value, void *arg)
     if (XATTR_IS_NODE_UUID(key)) {
         if (data->cbk->fop->int32) {
             /* List of node uuid is requested */
-            return ec_dict_data_concat("{ }", data->cbk, data->which, key,
+            return ec_dict_data_concat(data->cbk, data->which, key,
                                        GF_XATTR_LIST_NODE_UUIDS_KEY, UUID0_STR,
-                                       _gf_true);
+                                       _gf_true, "{ }");
         } else {
             return ec_dict_data_uuid(data->cbk, data->which, key);
         }
diff --git a/xlators/cluster/ec/src/ec-common.c b/xlators/cluster/ec/src/ec-common.c
index 737f7fda882..b955efd8c2d 100644
--- a/xlators/cluster/ec/src/ec-common.c
+++ b/xlators/cluster/ec/src/ec-common.c
@@ -8,8 +8,8 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "byte-order.h"
-#include "hashfn.h"
+#include <glusterfs/byte-order.h>
+#include <glusterfs/hashfn.h>
 
 #include "ec-mem-types.h"
 #include "ec-types.h"
@@ -44,16 +44,16 @@ ec_update_fd_status(fd_t *fd, xlator_t *xl, int idx, int32_t ret_status)
     UNLOCK(&fd->lock);
 }
 
-static int
-ec_fd_ctx_need_open(fd_t *fd, xlator_t *this, uintptr_t *need_open)
+static uintptr_t
+ec_fd_ctx_need_open(fd_t *fd, xlator_t *this, uintptr_t mask)
 {
     int i = 0;
     int count = 0;
     ec_t *ec = NULL;
     ec_fd_t *fd_ctx = NULL;
+    uintptr_t need_open = 0;
 
     ec = this->private;
-    *need_open = 0;
 
     fd_ctx = ec_fd_get(fd, this);
     if (!fd_ctx)
@@ -63,9 +63,9 @@ ec_fd_ctx_need_open(fd_t *fd, xlator_t *this, uintptr_t *need_open)
     {
         for (i = 0; i < ec->nodes; i++) {
             if ((fd_ctx->fd_status[i] == EC_FD_NOT_OPENED) &&
-                (ec->xl_up & (1 << i))) {
+                ((ec->xl_up & (1 << i)) != 0) && ((mask & (1 << i)) != 0)) {
                 fd_ctx->fd_status[i] = EC_FD_OPENING;
-                *need_open |= (1 << i);
+                need_open |= (1 << i);
                 count++;
             }
         }
@@ -76,10 +76,11 @@ ec_fd_ctx_need_open(fd_t *fd, xlator_t *this, uintptr_t *need_open)
      * then ignore fixing the fd as it has been
      * requested from heal operation.
      */
-    if (count >= ec->fragments)
-        count = 0;
+    if (count >= ec->fragments) {
+        need_open = 0;
+    }
 
-    return count;
+    return need_open;
 }
 
 static gf_boolean_t
@@ -96,11 +97,11 @@ ec_is_fd_fixable(fd_t *fd)
 }
 
 static void
-ec_fix_open(ec_fop_data_t *fop)
+ec_fix_open(ec_fop_data_t *fop, uintptr_t mask)
 {
-    int call_count = 0;
     uintptr_t need_open = 0;
     int ret = 0;
+    int32_t flags = 0;
     loc_t loc = {
         0,
     };
@@ -109,9 +110,10 @@ ec_fix_open(ec_fop_data_t *fop)
         goto out;
 
     /* Evaluate how many remote fd's to be opened */
-    call_count = ec_fd_ctx_need_open(fop->fd, fop->xl, &need_open);
-    if (!call_count)
+    need_open = ec_fd_ctx_need_open(fop->fd, fop->xl, mask);
+    if (need_open == 0) {
         goto out;
+    }
 
     loc.inode = inode_ref(fop->fd->inode);
     gf_uuid_copy(loc.gfid, fop->fd->inode->gfid);
@@ -120,12 +122,15 @@ ec_fix_open(ec_fop_data_t *fop)
         goto out;
     }
 
+    flags = fop->fd->flags & (~(O_TRUNC | O_APPEND | O_CREAT | O_EXCL));
     if (IA_IFDIR == fop->fd->inode->ia_type) {
-        ec_opendir(fop->frame, fop->xl, need_open, EC_MINIMUM_ONE, NULL, NULL,
+        ec_opendir(fop->frame, fop->xl, need_open,
+                   EC_MINIMUM_ONE | EC_FOP_NO_PROPAGATE_ERROR, NULL, NULL,
                    &fop->loc[0], fop->fd, NULL);
     } else {
-        ec_open(fop->frame, fop->xl, need_open, EC_MINIMUM_ONE, NULL, NULL,
-                &loc, fop->fd->flags, fop->fd, NULL);
+        ec_open(fop->frame, fop->xl, need_open,
+                EC_MINIMUM_ONE | EC_FOP_NO_PROPAGATE_ERROR, NULL, NULL, &loc,
+                flags, fop->fd, NULL);
     }
 
 out:
@@ -225,7 +230,7 @@ ec_child_next(ec_t *ec, ec_fop_data_t *fop, uint32_t idx)
 int32_t
 ec_heal_report(call_frame_t *frame, void *cookie, xlator_t *this,
                int32_t op_ret, int32_t op_errno, uintptr_t mask, uintptr_t good,
-               uintptr_t bad, dict_t *xdata)
+               uintptr_t bad, uint32_t pending, dict_t *xdata)
 {
     if (op_ret < 0) {
         gf_msg(this->name, GF_LOG_DEBUG, op_errno, EC_MSG_HEAL_FAIL,
@@ -311,16 +316,19 @@ ec_check_status(ec_fop_data_t *fop)
         }
     }
 
-    gf_msg(fop->xl->name, GF_LOG_WARNING, 0, EC_MSG_OP_FAIL_ON_SUBVOLS,
-           "Operation failed on %d of %d subvolumes.(up=%s, mask=%s, "
-           "remaining=%s, good=%s, bad=%s)",
-           gf_bits_count(ec->xl_up & ~(fop->remaining | fop->good)), ec->nodes,
-           ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes),
-           ec_bin(str2, sizeof(str2), fop->mask, ec->nodes),
-           ec_bin(str3, sizeof(str3), fop->remaining, ec->nodes),
-           ec_bin(str4, sizeof(str4), fop->good, ec->nodes),
-           ec_bin(str5, sizeof(str5), ec->xl_up & ~(fop->remaining | fop->good),
-                  ec->nodes));
+    gf_msg(
+        fop->xl->name, GF_LOG_WARNING, 0, EC_MSG_OP_FAIL_ON_SUBVOLS,
+        "Operation failed on %d of %d subvolumes.(up=%s, mask=%s, "
+        "remaining=%s, good=%s, bad=%s,"
+        "(Least significant bit represents first client/brick of subvol), %s)",
+        gf_bits_count(ec->xl_up & ~(fop->remaining | fop->good)), ec->nodes,
+        ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes),
+        ec_bin(str2, sizeof(str2), fop->mask, ec->nodes),
+        ec_bin(str3, sizeof(str3), fop->remaining, ec->nodes),
+        ec_bin(str4, sizeof(str4), fop->good, ec->nodes),
+        ec_bin(str5, sizeof(str5), ec->xl_up & ~(fop->remaining | fop->good),
+               ec->nodes),
+        ec_msg_str(fop));
     if (fop->use_fd) {
         if (fop->fd != NULL) {
             ec_fheal(NULL, fop->xl, -1, EC_MINIMUM_ONE, ec_heal_report, NULL,
@@ -494,12 +502,16 @@ ec_resume(ec_fop_data_t *fop, int32_t error)
 }
 
 void
-ec_resume_parent(ec_fop_data_t *fop, int32_t error)
+ec_resume_parent(ec_fop_data_t *fop)
 {
     ec_fop_data_t *parent;
+    int32_t error = 0;
 
     parent = fop->parent;
     if (parent != NULL) {
+        if ((fop->fop_flags & EC_FOP_NO_PROPAGATE_ERROR) == 0) {
+            error = fop->error;
+        }
         ec_trace("RESUME_PARENT", fop, "error=%u", error);
         fop->parent = NULL;
         ec_resume(parent, error);
@@ -592,6 +604,8 @@ ec_internal_op(ec_fop_data_t *fop)
         return _gf_true;
     if (fop->id == GF_FOP_FXATTROP)
         return _gf_true;
+    if (fop->id == GF_FOP_OPEN)
+        return _gf_true;
     return _gf_false;
 }
 
@@ -602,10 +616,10 @@ ec_msg_str(ec_fop_data_t *fop)
     loc_t *loc2 = NULL;
     char gfid1[64] = {0};
     char gfid2[64] = {0};
+    ec_fop_data_t *parent = fop->parent;
 
     if (fop->errstr)
         return fop->errstr;
-
     if (!fop->use_fd) {
         loc1 = &fop->loc[0];
         loc2 = &fop->loc[1];
@@ -613,24 +627,46 @@ ec_msg_str(ec_fop_data_t *fop)
         if (fop->id == GF_FOP_RENAME) {
             gf_asprintf(&fop->errstr,
                         "FOP : '%s' failed on '%s' and '%s' with gfids "
-                        "%s and %s respectively",
+                        "%s and %s respectively. Parent FOP: %s",
                         ec_fop_name(fop->id), loc1->path, loc2->path,
                         uuid_utoa_r(loc1->gfid, gfid1),
-                        uuid_utoa_r(loc2->gfid, gfid2));
+                        uuid_utoa_r(loc2->gfid, gfid2),
+                        parent ? ec_fop_name(parent->id) : "No Parent");
         } else {
-            gf_asprintf(&fop->errstr, "FOP : '%s' failed on '%s' with gfid %s",
-                        ec_fop_name(fop->id), loc1->path,
-                        uuid_utoa_r(loc1->gfid, gfid1));
+            gf_asprintf(
+                &fop->errstr,
+                "FOP : '%s' failed on '%s' with gfid %s. Parent FOP: %s",
+                ec_fop_name(fop->id), loc1->path,
+                uuid_utoa_r(loc1->gfid, gfid1),
+                parent ? ec_fop_name(parent->id) : "No Parent");
         }
     } else {
-        gf_asprintf(&fop->errstr, "FOP : '%s' failed on gfid %s",
-                    ec_fop_name(fop->id),
-                    uuid_utoa_r(fop->fd->inode->gfid, gfid1));
+        gf_asprintf(
+            &fop->errstr, "FOP : '%s' failed on gfid %s. Parent FOP: %s",
+            ec_fop_name(fop->id), uuid_utoa_r(fop->fd->inode->gfid, gfid1),
+            parent ? ec_fop_name(parent->id) : "No Parent");
     }
     return fop->errstr;
 }
 
-int32_t
+static void
+ec_log_insufficient_vol(ec_fop_data_t *fop, int32_t have, uint32_t need,
+                        int32_t loglevel)
+{
+    ec_t *ec = fop->xl->private;
+    char str1[32], str2[32], str3[32];
+
+    gf_msg(ec->xl->name, loglevel, 0, EC_MSG_CHILDS_INSUFFICIENT,
+           "Insufficient available children for this request: "
+           "Have : %d, Need : %u : Child UP : %s "
+           "Mask: %s, Healing : %s : %s ",
+           have, need, ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes),
+           ec_bin(str2, sizeof(str2), fop->mask, ec->nodes),
+           ec_bin(str3, sizeof(str3), fop->healing, ec->nodes),
+           ec_msg_str(fop));
+}
+
+static int32_t
 ec_child_select(ec_fop_data_t *fop)
 {
     ec_t *ec = fop->xl->private;
@@ -644,6 +680,9 @@ ec_child_select(ec_fop_data_t *fop)
      * unlock should go on all subvols where lock is performed*/
     if (fop->parent && !ec_internal_op(fop)) {
         fop->mask &= (fop->parent->mask & ~fop->parent->healing);
+        if (ec_is_data_fop(fop->id)) {
+            fop->healing |= fop->parent->healing;
+        }
     }
 
     if ((fop->mask & ~ec->xl_up) != 0) {
@@ -684,15 +723,18 @@ ec_child_select(ec_fop_data_t *fop)
     ec_trace("SELECT", fop, "");
 
     if ((num < fop->minimum) && (num < ec->fragments)) {
-        gf_msg(ec->xl->name, GF_LOG_ERROR, 0, EC_MSG_CHILDS_INSUFFICIENT,
-               "Insufficient available children "
-               "for this request (have %d, need "
-               "%d). %s",
-               num, fop->minimum, ec_msg_str(fop));
+        ec_log_insufficient_vol(fop, num, fop->minimum, GF_LOG_ERROR);
         return 0;
     }
 
-    ec_sleep(fop);
+    if (!fop->parent && fop->lock_count &&
+        (fop->locks[0].update[EC_DATA_TXN] ||
+         fop->locks[0].update[EC_METADATA_TXN])) {
+        if (ec->quorum_count && (num < ec->quorum_count)) {
+            ec_log_insufficient_vol(fop, num, ec->quorum_count, GF_LOG_ERROR);
+            return 0;
+        }
+    }
 
     return 1;
 }
@@ -772,6 +814,8 @@ ec_dispatch_one(ec_fop_data_t *fop)
     ec_dispatch_start(fop);
 
     if (ec_child_select(fop)) {
+        ec_sleep(fop);
+
         fop->expected = 1;
         fop->first = ec_select_first_by_read_policy(fop->xl->private, fop);
 
@@ -806,6 +850,8 @@ ec_dispatch_inc(ec_fop_data_t *fop)
     ec_dispatch_start(fop);
 
     if (ec_child_select(fop)) {
+        ec_sleep(fop);
+
         fop->expected = gf_bits_count(fop->remaining);
         fop->first = 0;
 
@@ -819,6 +865,8 @@ ec_dispatch_all(ec_fop_data_t *fop)
     ec_dispatch_start(fop);
 
     if (ec_child_select(fop)) {
+        ec_sleep(fop);
+
         fop->expected = gf_bits_count(fop->remaining);
         fop->first = 0;
 
@@ -837,6 +885,8 @@ ec_dispatch_min(ec_fop_data_t *fop)
     ec_dispatch_start(fop);
 
     if (ec_child_select(fop)) {
+        ec_sleep(fop);
+
         fop->expected = count = ec->fragments;
         fop->first = ec_select_first_by_read_policy(fop->xl->private, fop);
         idx = fop->first - 1;
@@ -851,6 +901,23 @@ ec_dispatch_min(ec_fop_data_t *fop)
     }
 }
 
+void
+ec_succeed_all(ec_fop_data_t *fop)
+{
+    ec_dispatch_start(fop);
+
+    if (ec_child_select(fop)) {
+        fop->expected = gf_bits_count(fop->remaining);
+        fop->first = 0;
+
+        /* Simulate a successful execution on all bricks */
+        ec_trace("SUCCEED", fop, "");
+
+        fop->good = fop->remaining;
+        fop->remaining = 0;
+    }
+}
+
 ec_lock_t *
 ec_lock_allocate(ec_fop_data_t *fop, loc_t *loc)
 {
@@ -1372,27 +1439,28 @@ ec_get_size_version(ec_lock_link_t *link)
         !ec_is_data_fop(fop->id))
         link->optimistic_changelog = _gf_true;
 
+    memset(&loc, 0, sizeof(loc));
+
+    LOCK(&lock->loc.inode->lock);
+
     set_dirty = ec_set_dirty_flag(link, ctx, dirty);
 
     /* If ec metadata has already been retrieved, do not try again. */
-    if (ctx->have_info && (!set_dirty)) {
+    if (ctx->have_info) {
         if (ec_is_data_fop(fop->id)) {
             fop->healing |= lock->healing;
         }
-        return;
+        if (!set_dirty)
+            goto unlock;
     }
 
     /* Determine if there's something we need to retrieve for the current
      * operation. */
     if (!set_dirty && !lock->query && (lock->loc.inode->ia_type != IA_IFREG) &&
         (lock->loc.inode->ia_type != IA_INVAL)) {
-        return;
+        goto unlock;
     }
 
-    memset(&loc, 0, sizeof(loc));
-
-    LOCK(&lock->loc.inode->lock);
-
     changed_flags = ec_set_xattrop_flags_and_params(lock, link, dirty);
     if (link->waiting_flags) {
         /* This fop needs to wait until all its flags are cleared which
@@ -1403,6 +1471,7 @@ ec_get_size_version(ec_lock_link_t *link)
         GF_ASSERT(!changed_flags);
     }
 
+unlock:
     UNLOCK(&lock->loc.inode->lock);
 
     if (!changed_flags)
@@ -1814,6 +1883,10 @@ ec_lock_acquired(ec_lock_link_t *link)
     LOCK(&lock->loc.inode->lock);
 
     lock->acquired = _gf_true;
+    if (lock->contention) {
+        lock->release = _gf_true;
+        lock->contention = _gf_false;
+    }
 
     ec_lock_update_fd(lock, fop);
     ec_lock_wake_shared(lock, &list);
@@ -1824,7 +1897,8 @@ ec_lock_acquired(ec_lock_link_t *link)
 
     if (fop->use_fd &&
         (link->update[EC_DATA_TXN] || link->update[EC_METADATA_TXN])) {
-        ec_fix_open(fop);
+        /* Try to reopen closed fd's only if lock has succeeded. */
+        ec_fix_open(fop, lock->mask);
     }
 
     ec_lock_resume_shared(&list);
@@ -1838,15 +1912,20 @@ ec_locked(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
     ec_lock_link_t *link = NULL;
     ec_lock_t *lock = NULL;
 
+    link = fop->data;
+    lock = link->lock;
     if (op_ret >= 0) {
-        link = fop->data;
-        lock = link->lock;
         lock->mask = lock->good_mask = fop->good;
         lock->healing = 0;
 
         ec_lock_acquired(link);
         ec_lock(fop->parent);
     } else {
+        LOCK(&lock->loc.inode->lock);
+        {
+            lock->contention = _gf_false;
+        }
+        UNLOCK(&lock->loc.inode->lock);
         gf_msg(this->name, GF_LOG_WARNING, op_errno, EC_MSG_PREOP_LOCK_FAILED,
                "Failed to complete preop lock");
     }
@@ -2177,7 +2256,7 @@ ec_unlocked(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
 
     if (op_ret < 0) {
         gf_msg(this->name, GF_LOG_WARNING, op_errno, EC_MSG_UNLOCK_FAILED,
-               "entry/inode unlocking failed (%s)", ec_fop_name(link->fop->id));
+               "entry/inode unlocking failed :(%s)", ec_msg_str(link->fop));
     } else {
         ec_trace("UNLOCKED", link->fop, "lock=%p", link->lock);
     }
@@ -2214,6 +2293,23 @@ ec_unlock_lock(ec_lock_link_t *link)
     }
 }
 
+void
+ec_inode_bad_inc(inode_t *inode, xlator_t *xl)
+{
+    ec_inode_t *ctx = NULL;
+
+    LOCK(&inode->lock);
+    {
+        ctx = __ec_inode_get(inode, xl);
+        if (ctx == NULL) {
+            goto unlock;
+        }
+        ctx->bad_version++;
+    }
+unlock:
+    UNLOCK(&inode->lock);
+}
+
 int32_t
 ec_update_size_version_done(call_frame_t *frame, void *cookie, xlator_t *this,
                             int32_t op_ret, int32_t op_errno, dict_t *xattr,
@@ -2229,6 +2325,12 @@ ec_update_size_version_done(call_frame_t *frame, void *cookie, xlator_t *this,
     ctx = lock->ctx;
 
     if (op_ret < 0) {
+        if (link->lock->fd == NULL) {
+            ec_inode_bad_inc(link->lock->loc.inode, this);
+        } else {
+            ec_inode_bad_inc(link->lock->fd->inode, this);
+        }
+
         gf_msg(fop->xl->name, fop_log_level(fop->id, op_errno), op_errno,
                EC_MSG_SIZE_VERS_UPDATE_FAIL,
                "Failed to update version and size. %s", ec_msg_str(fop));
@@ -2371,37 +2473,47 @@ ec_update_info(ec_lock_link_t *link)
     uint64_t dirty[2] = {0, 0};
     uint64_t size;
     ec_t *ec = NULL;
+    uintptr_t mask;
 
     lock = link->lock;
     ctx = lock->ctx;
     ec = link->fop->xl->private;
 
     /* pre_version[*] will be 0 if have_version is false */
-    version[0] = ctx->post_version[0] - ctx->pre_version[0];
-    version[1] = ctx->post_version[1] - ctx->pre_version[1];
+    version[EC_DATA_TXN] = ctx->post_version[EC_DATA_TXN] -
+                           ctx->pre_version[EC_DATA_TXN];
+    version[EC_METADATA_TXN] = ctx->post_version[EC_METADATA_TXN] -
+                               ctx->pre_version[EC_METADATA_TXN];
 
     size = ctx->post_size - ctx->pre_size;
     /* If we set the dirty flag for update fop, we have to unset it.
      * If fop has failed on some bricks, leave the dirty as marked. */
+
     if (lock->unlock_now) {
+        if (version[EC_DATA_TXN]) {
+            /*A data fop will have difference in post and pre version
+             *and for data fop we send writes on healing bricks also */
+            mask = lock->good_mask | lock->healing;
+        } else {
+            mask = lock->good_mask;
+        }
         /* Ensure that nodes are up while doing final
          * metadata update.*/
-        if (!(ec->node_mask & ~lock->good_mask) &&
-            !(ec->node_mask & ~ec->xl_up)) {
-            if (ctx->dirty[0] != 0) {
-                dirty[0] = -1;
+        if (!(ec->node_mask & ~(mask)) && !(ec->node_mask & ~ec->xl_up)) {
+            if (ctx->dirty[EC_DATA_TXN] != 0) {
+                dirty[EC_DATA_TXN] = -1;
             }
-            if (ctx->dirty[1] != 0) {
-                dirty[1] = -1;
+            if (ctx->dirty[EC_METADATA_TXN] != 0) {
+                dirty[EC_METADATA_TXN] = -1;
             }
             /*If everything is fine and we already
              *have version xattr set on entry, there
              *is no need to update version again*/
-            if (ctx->pre_version[0]) {
-                version[0] = 0;
+            if (ctx->pre_version[EC_DATA_TXN]) {
+                version[EC_DATA_TXN] = 0;
             }
-            if (ctx->pre_version[1]) {
-                version[1] = 0;
+            if (ctx->pre_version[EC_METADATA_TXN]) {
+                version[EC_METADATA_TXN] = 0;
             }
         } else {
             link->optimistic_changelog = _gf_false;
@@ -2410,8 +2522,8 @@ ec_update_info(ec_lock_link_t *link)
         memset(ctx->dirty, 0, sizeof(ctx->dirty));
     }
 
-    if ((version[0] != 0) || (version[1] != 0) || (dirty[0] != 0) ||
-        (dirty[1] != 0)) {
+    if ((version[EC_DATA_TXN] != 0) || (version[EC_METADATA_TXN] != 0) ||
+        (dirty[EC_DATA_TXN] != 0) || (dirty[EC_METADATA_TXN] != 0)) {
         ec_update_size_version(link, version, size, dirty);
         return _gf_true;
     }
@@ -2453,13 +2565,20 @@ ec_lock_release(ec_t *ec, inode_t *inode)
         goto done;
     }
     lock = ctx->inode_lock;
-    if ((lock == NULL) || !lock->acquired || lock->release) {
+    if ((lock == NULL) || lock->release) {
         goto done;
     }
 
     gf_msg_debug(ec->xl->name, 0, "Releasing inode %p due to lock contention",
                  inode);
 
+    if (!lock->acquired) {
+        /* This happens if some bricks already got the lock while inodelk is in
+         * progress.  Set release to true after lock is acquired*/
+        lock->contention = _gf_true;
+        goto done;
+    }
+
     /* The lock is not marked to be released, so the frozen list should be
      * empty. */
     GF_ASSERT(list_empty(&lock->frozen));
@@ -2911,3 +3030,13 @@ ec_manager(ec_fop_data_t *fop, int32_t error)
 
     __ec_manager(fop, error);
 }
+
+gf_boolean_t
+__ec_is_last_fop(ec_t *ec)
+{
+    if ((list_empty(&ec->pending_fops)) &&
+        (GF_ATOMIC_GET(ec->async_fop_count) == 0)) {
+        return _gf_true;
+    }
+    return _gf_false;
+}
diff --git a/xlators/cluster/ec/src/ec-common.h b/xlators/cluster/ec/src/ec-common.h
index bf41c0086f8..51493612ac6 100644
--- a/xlators/cluster/ec/src/ec-common.h
+++ b/xlators/cluster/ec/src/ec-common.h
@@ -11,8 +11,7 @@
 #ifndef __EC_COMMON_H__
 #define __EC_COMMON_H__
 
-#include "xlator.h"
-
+#include "glusterfs/compat-errno.h"  // for ENODATA on BSD
 #include "ec-data.h"
 
 typedef enum { EC_DATA_TXN, EC_METADATA_TXN } ec_txn_t;
@@ -26,6 +25,30 @@ typedef enum { EC_DATA_TXN, EC_METADATA_TXN } ec_txn_t;
 
 #define EC_FLAG_LOCK_SHARED 0x0001
 
+#define QUORUM_CBK(fn, fop, frame, cookie, this, op_ret, op_errno, params...)  \
+    do {                                                                       \
+        ec_t *__ec = fop->xl->private;                                         \
+        int32_t __op_ret = 0;                                                  \
+        int32_t __op_errno = 0;                                                \
+        int32_t __success_count = gf_bits_count(fop->good);                    \
+                                                                               \
+        __op_ret = op_ret;                                                     \
+        __op_errno = op_errno;                                                 \
+        if (!fop->parent && frame &&                                           \
+            (GF_CLIENT_PID_SELF_HEALD != frame->root->pid) &&                  \
+            __ec->quorum_count && (__success_count < __ec->quorum_count) &&    \
+            op_ret >= 0) {                                                     \
+            __op_ret = -1;                                                     \
+            __op_errno = EIO;                                                  \
+            gf_msg(__ec->xl->name, GF_LOG_ERROR, 0,                            \
+                   EC_MSG_CHILDS_INSUFFICIENT,                                 \
+                   "Insufficient available children for this request "         \
+                   "(have %d, need %d). %s",                                   \
+                   __success_count, __ec->quorum_count, ec_msg_str(fop));      \
+        }                                                                      \
+        fn(frame, cookie, this, __op_ret, __op_errno, params);                 \
+    } while (0)
+
 enum _ec_xattrop_flags {
     EC_FLAG_XATTROP,
     EC_FLAG_DATA_DIRTY,
@@ -54,9 +77,12 @@ enum _ec_xattrop_flags {
 
 #define EC_SELFHEAL_BIT 62
 
-#define EC_MINIMUM_ONE -1
-#define EC_MINIMUM_MIN -2
-#define EC_MINIMUM_ALL -3
+#define EC_MINIMUM_ONE (1 << 6)
+#define EC_MINIMUM_MIN (2 << 6)
+#define EC_MINIMUM_ALL (3 << 6)
+#define EC_FOP_NO_PROPAGATE_ERROR (1 << 8)
+#define EC_FOP_MINIMUM(_flags) ((_flags)&255)
+#define EC_FOP_FLAGS(_flags) ((_flags) & ~255)
 
 #define EC_UPDATE_DATA 1
 #define EC_UPDATE_META 2
@@ -163,11 +189,14 @@ void
 ec_dispatch_one(ec_fop_data_t *fop);
 
 void
+ec_succeed_all(ec_fop_data_t *fop);
+
+void
 ec_sleep(ec_fop_data_t *fop);
 void
 ec_resume(ec_fop_data_t *fop, int32_t error);
 void
-ec_resume_parent(ec_fop_data_t *fop, int32_t error);
+ec_resume_parent(ec_fop_data_t *fop);
 
 void
 ec_manager(ec_fop_data_t *fop, int32_t error);
@@ -190,4 +219,16 @@ ec_lock_unlocked(call_frame_t *frame, void *cookie, xlator_t *this,
 void
 ec_update_fd_status(fd_t *fd, xlator_t *xl, int child_index,
                     int32_t ret_status);
+gf_boolean_t
+ec_is_entry_healing(ec_fop_data_t *fop);
+void
+ec_set_entry_healing(ec_fop_data_t *fop);
+void
+ec_reset_entry_healing(ec_fop_data_t *fop);
+char *
+ec_msg_str(ec_fop_data_t *fop);
+gf_boolean_t
+__ec_is_last_fop(ec_t *ec);
+void
+ec_lock_update_good(ec_lock_t *lock, ec_fop_data_t *fop);
 #endif /* __EC_COMMON_H__ */
diff --git a/xlators/cluster/ec/src/ec-data.c b/xlators/cluster/ec/src/ec-data.c
index fae8843a679..06388833546 100644
--- a/xlators/cluster/ec/src/ec-data.c
+++ b/xlators/cluster/ec/src/ec-data.c
@@ -8,7 +8,6 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "ec-mem-types.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-data.h"
@@ -98,7 +97,7 @@ ec_cbk_data_destroy(ec_cbk_data_t *cbk)
 
 ec_fop_data_t *
 ec_fop_data_allocate(call_frame_t *frame, xlator_t *this, int32_t id,
-                     uint32_t flags, uintptr_t target, int32_t minimum,
+                     uint32_t flags, uintptr_t target, uint32_t fop_flags,
                      ec_wind_f wind, ec_handler_f handler, ec_cbk_t cbks,
                      void *data)
 {
@@ -151,7 +150,8 @@ ec_fop_data_allocate(call_frame_t *frame, xlator_t *this, int32_t id,
     fop->refs = 1;
 
     fop->flags = flags;
-    fop->minimum = minimum;
+    fop->minimum = EC_FOP_MINIMUM(fop_flags);
+    fop->fop_flags = EC_FOP_FLAGS(fop_flags);
     fop->mask = target;
 
     fop->wind = wind;
@@ -201,11 +201,13 @@ ec_handle_last_pending_fop_completion(ec_fop_data_t *fop, gf_boolean_t *notify)
 {
     ec_t *ec = fop->xl->private;
 
+    *notify = _gf_false;
+
     if (!list_empty(&fop->pending_list)) {
         LOCK(&ec->lock);
         {
             list_del_init(&fop->pending_list);
-            *notify = list_empty(&ec->pending_fops);
+            *notify = __ec_is_last_fop(ec);
         }
         UNLOCK(&ec->lock);
     }
@@ -271,7 +273,7 @@ ec_fop_data_release(ec_fop_data_t *fop)
         loc_wipe(&fop->loc[1]);
         GF_FREE(fop->errstr);
 
-        ec_resume_parent(fop, fop->error);
+        ec_resume_parent(fop);
 
         ec_fop_cleanup(fop);
 
diff --git a/xlators/cluster/ec/src/ec-data.h b/xlators/cluster/ec/src/ec-data.h
index 112536d554c..c8a74ffe1ed 100644
--- a/xlators/cluster/ec/src/ec-data.h
+++ b/xlators/cluster/ec/src/ec-data.h
@@ -18,7 +18,7 @@ ec_cbk_data_allocate(call_frame_t *frame, xlator_t *this, ec_fop_data_t *fop,
                      int32_t id, int32_t idx, int32_t op_ret, int32_t op_errno);
 ec_fop_data_t *
 ec_fop_data_allocate(call_frame_t *frame, xlator_t *this, int32_t id,
-                     uint32_t flags, uintptr_t target, int32_t minimum,
+                     uint32_t flags, uintptr_t target, uint32_t fop_flags,
                      ec_wind_f wind, ec_handler_f handler, ec_cbk_t cbks,
                      void *data);
 void
diff --git a/xlators/cluster/ec/src/ec-dir-read.c b/xlators/cluster/ec/src/ec-dir-read.c
index ec4cefb1e78..f71dcfac293 100644
--- a/xlators/cluster/ec/src/ec-dir-read.c
+++ b/xlators/cluster/ec/src/ec-dir-read.c
@@ -8,15 +8,11 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-
 #include "ec.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
-#include "ec-method.h"
 #include "ec-fops.h"
 
 /****************************************************************
@@ -127,13 +123,15 @@ ec_manager_opendir(ec_fop_data_t *fop, int32_t state)
 
                 return EC_STATE_REPORT;
             }
-            err = ec_loc_from_loc(fop->xl, &ctx->loc, &fop->loc[0]);
-            if (err != 0) {
-                UNLOCK(&fop->fd->lock);
+            if (!ctx->loc.inode) {
+                err = ec_loc_from_loc(fop->xl, &ctx->loc, &fop->loc[0]);
+                if (err != 0) {
+                    UNLOCK(&fop->fd->lock);
 
-                fop->error = -err;
+                    fop->error = -err;
 
-                return EC_STATE_REPORT;
+                    return EC_STATE_REPORT;
+                }
             }
 
             UNLOCK(&fop->fd->lock);
@@ -219,7 +217,7 @@ ec_manager_opendir(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_opendir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_opendir_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_opendir_cbk_t func, void *data, loc_t *loc,
            fd_t *fd, dict_t *xdata)
 {
     ec_cbk_t callback = {.opendir = func};
@@ -233,7 +231,7 @@ ec_opendir(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_OPENDIR, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_opendir,
+                               target, fop_flags, ec_wind_opendir,
                                ec_manager_opendir, callback, data);
     if (fop == NULL) {
         goto out;
@@ -388,9 +386,16 @@ ec_manager_readdir(ec_fop_data_t *fop, int32_t state)
             /* Return error if opendir has not been successfully called on
              * any subvolume. */
             ctx = ec_fd_get(fop->fd, fop->xl);
-            if ((ctx == NULL) || (ctx->open == 0)) {
-                fop->error = EINVAL;
+            if (ctx == NULL) {
+                fop->error = ENOMEM;
+            } else if (ctx->open == 0) {
+                fop->error = EBADFD;
+            }
 
+            if (fop->error) {
+                gf_msg(fop->xl->name, GF_LOG_ERROR, fop->error,
+                       EC_MSG_INVALID_REQUEST, "EC is not winding readdir: %s",
+                       ec_msg_str(fop));
                 return EC_STATE_REPORT;
             }
 
@@ -515,7 +520,7 @@ ec_manager_readdir(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_readdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_readdir_cbk_t func, void *data, fd_t *fd,
+           uint32_t fop_flags, fop_readdir_cbk_t func, void *data, fd_t *fd,
            size_t size, off_t offset, dict_t *xdata)
 {
     ec_cbk_t callback = {.readdir = func};
@@ -529,7 +534,7 @@ ec_readdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_READDIR, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_readdir,
+                               target, fop_flags, ec_wind_readdir,
                                ec_manager_readdir, callback, data);
     if (fop == NULL) {
         goto out;
@@ -585,7 +590,7 @@ ec_wind_readdirp(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_readdirp(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_readdirp_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_readdirp_cbk_t func, void *data, fd_t *fd,
             size_t size, off_t offset, dict_t *xdata)
 {
     ec_cbk_t callback = {.readdirp = func};
@@ -599,7 +604,7 @@ ec_readdirp(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(
-        frame, this, GF_FOP_READDIRP, EC_FLAG_LOCK_SHARED, target, minimum,
+        frame, this, GF_FOP_READDIRP, EC_FLAG_LOCK_SHARED, target, fop_flags,
         ec_wind_readdirp, ec_manager_readdir, callback, data);
     if (fop == NULL) {
         goto out;
diff --git a/xlators/cluster/ec/src/ec-dir-write.c b/xlators/cluster/ec/src/ec-dir-write.c
index f5c38e80dd7..53d27d895c3 100644
--- a/xlators/cluster/ec/src/ec-dir-write.c
+++ b/xlators/cluster/ec/src/ec-dir-write.c
@@ -8,9 +8,6 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-
 #include "ec.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
@@ -218,10 +215,10 @@ ec_manager_create(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.create != NULL) {
-                fop->cbks.create(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                 cbk->op_errno, fop->fd, fop->loc[0].inode,
-                                 &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
-                                 cbk->xdata);
+                QUORUM_CBK(fop->cbks.create, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, fop->fd,
+                           fop->loc[0].inode, &cbk->iatt[0], &cbk->iatt[1],
+                           &cbk->iatt[2], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -262,7 +259,7 @@ ec_manager_create(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_create(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_create_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_create_cbk_t func, void *data, loc_t *loc,
           int32_t flags, mode_t mode, mode_t umask, fd_t *fd, dict_t *xdata)
 {
     ec_cbk_t callback = {.create = func};
@@ -275,7 +272,7 @@ ec_create(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_CREATE, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_CREATE, 0, target, fop_flags,
                                ec_wind_create, ec_manager_create, callback,
                                data);
     if (fop == NULL) {
@@ -390,9 +387,10 @@ ec_manager_link(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.link != NULL) {
-                fop->cbks.link(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                               cbk->op_errno, fop->loc[0].inode, &cbk->iatt[0],
-                               &cbk->iatt[1], &cbk->iatt[2], cbk->xdata);
+                QUORUM_CBK(fop->cbks.link, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, fop->loc[0].inode,
+                           &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
+                           cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -432,9 +430,9 @@ ec_manager_link(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_link(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_link_cbk_t func, void *data, loc_t *oldloc, loc_t *newloc,
-        dict_t *xdata)
+ec_link(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_link_cbk_t func, void *data, loc_t *oldloc,
+        loc_t *newloc, dict_t *xdata)
 {
     ec_cbk_t callback = {.link = func};
     ec_fop_data_t *fop = NULL;
@@ -446,7 +444,7 @@ ec_link(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_LINK, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_LINK, 0, target, fop_flags,
                                ec_wind_link, ec_manager_link, callback, data);
     if (fop == NULL) {
         goto out;
@@ -569,9 +567,10 @@ ec_manager_mkdir(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.mkdir != NULL) {
-                fop->cbks.mkdir(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                cbk->op_errno, fop->loc[0].inode, &cbk->iatt[0],
-                                &cbk->iatt[1], &cbk->iatt[2], cbk->xdata);
+                QUORUM_CBK(fop->cbks.mkdir, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, fop->loc[0].inode,
+                           &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
+                           cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -613,9 +612,9 @@ ec_manager_mkdir(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_mkdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_mkdir_cbk_t func, void *data, loc_t *loc, mode_t mode,
-         mode_t umask, dict_t *xdata)
+ec_mkdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_mkdir_cbk_t func, void *data, loc_t *loc,
+         mode_t mode, mode_t umask, dict_t *xdata)
 {
     ec_cbk_t callback = {.mkdir = func};
     ec_fop_data_t *fop = NULL;
@@ -627,7 +626,7 @@ ec_mkdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_MKDIR, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_MKDIR, 0, target, fop_flags,
                                ec_wind_mkdir, ec_manager_mkdir, callback, data);
     if (fop == NULL) {
         goto out;
@@ -773,9 +772,10 @@ ec_manager_mknod(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.mknod != NULL) {
-                fop->cbks.mknod(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                cbk->op_errno, fop->loc[0].inode, &cbk->iatt[0],
-                                &cbk->iatt[1], &cbk->iatt[2], cbk->xdata);
+                QUORUM_CBK(fop->cbks.mknod, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, fop->loc[0].inode,
+                           &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
+                           cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -815,9 +815,9 @@ ec_manager_mknod(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_mknod(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_mknod_cbk_t func, void *data, loc_t *loc, mode_t mode, dev_t rdev,
-         mode_t umask, dict_t *xdata)
+ec_mknod(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_mknod_cbk_t func, void *data, loc_t *loc,
+         mode_t mode, dev_t rdev, mode_t umask, dict_t *xdata)
 {
     ec_cbk_t callback = {.mknod = func};
     ec_fop_data_t *fop = NULL;
@@ -829,7 +829,7 @@ ec_mknod(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_MKNOD, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_MKNOD, 0, target, fop_flags,
                                ec_wind_mknod, ec_manager_mknod, callback, data);
     if (fop == NULL) {
         goto out;
@@ -931,10 +931,10 @@ ec_manager_rename(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.rename != NULL) {
-                fop->cbks.rename(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                 cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                 &cbk->iatt[2], &cbk->iatt[3], &cbk->iatt[4],
-                                 cbk->xdata);
+                QUORUM_CBK(fop->cbks.rename, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], &cbk->iatt[2], &cbk->iatt[3],
+                           &cbk->iatt[4], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -975,7 +975,7 @@ ec_manager_rename(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_rename(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_rename_cbk_t func, void *data, loc_t *oldloc,
+          uint32_t fop_flags, fop_rename_cbk_t func, void *data, loc_t *oldloc,
           loc_t *newloc, dict_t *xdata)
 {
     ec_cbk_t callback = {.rename = func};
@@ -988,7 +988,7 @@ ec_rename(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_RENAME, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_RENAME, 0, target, fop_flags,
                                ec_wind_rename, ec_manager_rename, callback,
                                data);
     if (fop == NULL) {
@@ -1083,9 +1083,9 @@ ec_manager_rmdir(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.rmdir != NULL) {
-                fop->cbks.rmdir(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                cbk->xdata);
+                QUORUM_CBK(fop->cbks.rmdir, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -1125,9 +1125,9 @@ ec_manager_rmdir(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_rmdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_rmdir_cbk_t func, void *data, loc_t *loc, int xflags,
-         dict_t *xdata)
+ec_rmdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_rmdir_cbk_t func, void *data, loc_t *loc,
+         int xflags, dict_t *xdata)
 {
     ec_cbk_t callback = {.rmdir = func};
     ec_fop_data_t *fop = NULL;
@@ -1139,7 +1139,7 @@ ec_rmdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_RMDIR, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_RMDIR, 0, target, fop_flags,
                                ec_wind_rmdir, ec_manager_rmdir, callback, data);
     if (fop == NULL) {
         goto out;
@@ -1237,10 +1237,10 @@ ec_manager_symlink(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.symlink != NULL) {
-                fop->cbks.symlink(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                  cbk->op_errno, fop->loc[0].inode,
-                                  &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
-                                  cbk->xdata);
+                QUORUM_CBK(fop->cbks.symlink, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, fop->loc[0].inode,
+                           &cbk->iatt[0], &cbk->iatt[1], &cbk->iatt[2],
+                           cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -1281,7 +1281,7 @@ ec_manager_symlink(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_symlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_symlink_cbk_t func, void *data,
+           uint32_t fop_flags, fop_symlink_cbk_t func, void *data,
            const char *linkname, loc_t *loc, mode_t umask, dict_t *xdata)
 {
     ec_cbk_t callback = {.symlink = func};
@@ -1294,9 +1294,9 @@ ec_symlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_SYMLINK, 0, target, minimum,
-                               ec_wind_symlink, ec_manager_symlink, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_SYMLINK, 0, target,
+                               fop_flags, ec_wind_symlink, ec_manager_symlink,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1392,9 +1392,9 @@ ec_manager_unlink(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.unlink != NULL) {
-                fop->cbks.unlink(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                 cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                 cbk->xdata);
+                QUORUM_CBK(fop->cbks.unlink, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -1435,7 +1435,7 @@ ec_manager_unlink(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_unlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_unlink_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_unlink_cbk_t func, void *data, loc_t *loc,
           int xflags, dict_t *xdata)
 {
     ec_cbk_t callback = {.unlink = func};
@@ -1448,7 +1448,7 @@ ec_unlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_UNLINK, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_UNLINK, 0, target, fop_flags,
                                ec_wind_unlink, ec_manager_unlink, callback,
                                data);
     if (fop == NULL) {
diff --git a/xlators/cluster/ec/src/ec-fops.h b/xlators/cluster/ec/src/ec-fops.h
index 2858d829c73..07edf8a7fec 100644
--- a/xlators/cluster/ec/src/ec-fops.h
+++ b/xlators/cluster/ec/src/ec-fops.h
@@ -11,240 +11,244 @@
 #ifndef __EC_FOPS_H__
 #define __EC_FOPS_H__
 
-#include "xlator.h"
+#include <glusterfs/xlator.h>
 
 #include "ec-types.h"
 #include "ec-common.h"
 
 void
 ec_access(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_access_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_access_cbk_t func, void *data, loc_t *loc,
           int32_t mask, dict_t *xdata);
 
 void
 ec_create(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_create_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_create_cbk_t func, void *data, loc_t *loc,
           int32_t flags, mode_t mode, mode_t umask, fd_t *fd, dict_t *xdata);
 
 void
 ec_entrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_entrylk_cbk_t func, void *data,
+           uint32_t fop_flags, fop_entrylk_cbk_t func, void *data,
            const char *volume, loc_t *loc, const char *basename,
            entrylk_cmd cmd, entrylk_type type, dict_t *xdata);
 
 void
 ec_fentrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fentrylk_cbk_t func, void *data,
+            uint32_t fop_flags, fop_fentrylk_cbk_t func, void *data,
             const char *volume, fd_t *fd, const char *basename, entrylk_cmd cmd,
             entrylk_type type, dict_t *xdata);
 
 void
-ec_flush(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_flush_cbk_t func, void *data, fd_t *fd, dict_t *xdata);
+ec_flush(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_flush_cbk_t func, void *data, fd_t *fd,
+         dict_t *xdata);
 
 void
-ec_fsync(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fsync_cbk_t func, void *data, fd_t *fd, int32_t datasync,
-         dict_t *xdata);
+ec_fsync(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fsync_cbk_t func, void *data, fd_t *fd,
+         int32_t datasync, dict_t *xdata);
 
 void
 ec_fsyncdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fsyncdir_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fsyncdir_cbk_t func, void *data, fd_t *fd,
             int32_t datasync, dict_t *xdata);
 
 void
 ec_getxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_getxattr_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_getxattr_cbk_t func, void *data, loc_t *loc,
             const char *name, dict_t *xdata);
 
 void
 ec_fgetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fgetxattr_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fgetxattr_cbk_t func, void *data, fd_t *fd,
              const char *name, dict_t *xdata);
 
 void
-ec_heal(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_heal_cbk_t func, void *data, loc_t *loc, int32_t partial,
-        dict_t *xdata);
+ec_heal(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_heal_cbk_t func, void *data, loc_t *loc,
+        int32_t partial, dict_t *xdata);
 
 void
-ec_fheal(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fheal_cbk_t func, void *data, fd_t *fd, int32_t partial,
-         dict_t *xdata);
+ec_fheal(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fheal_cbk_t func, void *data, fd_t *fd,
+         int32_t partial, dict_t *xdata);
 
 void
 ec_inodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
-           uintptr_t target, int32_t minimum, fop_inodelk_cbk_t func,
+           uintptr_t target, uint32_t fop_flags, fop_inodelk_cbk_t func,
            void *data, const char *volume, loc_t *loc, int32_t cmd,
            struct gf_flock *flock, dict_t *xdata);
 
 void
 ec_finodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
-            uintptr_t target, int32_t minimum, fop_finodelk_cbk_t func,
+            uintptr_t target, uint32_t fop_flags, fop_finodelk_cbk_t func,
             void *data, const char *volume, fd_t *fd, int32_t cmd,
             struct gf_flock *flock, dict_t *xdata);
 
 void
-ec_link(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_link_cbk_t func, void *data, loc_t *oldloc, loc_t *newloc,
-        dict_t *xdata);
+ec_link(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_link_cbk_t func, void *data, loc_t *oldloc,
+        loc_t *newloc, dict_t *xdata);
 
 void
-ec_lk(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
+ec_lk(call_frame_t *frame, xlator_t *this, uintptr_t target, uint32_t fop_flags,
       fop_lk_cbk_t func, void *data, fd_t *fd, int32_t cmd,
       struct gf_flock *flock, dict_t *xdata);
 
 void
 ec_lookup(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_lookup_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_lookup_cbk_t func, void *data, loc_t *loc,
           dict_t *xdata);
 
 void
-ec_mkdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_mkdir_cbk_t func, void *data, loc_t *loc, mode_t mode,
-         mode_t umask, dict_t *xdata);
+ec_mkdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_mkdir_cbk_t func, void *data, loc_t *loc,
+         mode_t mode, mode_t umask, dict_t *xdata);
 
 void
-ec_mknod(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_mknod_cbk_t func, void *data, loc_t *loc, mode_t mode, dev_t rdev,
-         mode_t umask, dict_t *xdata);
+ec_mknod(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_mknod_cbk_t func, void *data, loc_t *loc,
+         mode_t mode, dev_t rdev, mode_t umask, dict_t *xdata);
 
 void
-ec_open(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_open_cbk_t func, void *data, loc_t *loc, int32_t flags, fd_t *fd,
-        dict_t *xdata);
+ec_open(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_open_cbk_t func, void *data, loc_t *loc,
+        int32_t flags, fd_t *fd, dict_t *xdata);
 
 void
 ec_opendir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_opendir_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_opendir_cbk_t func, void *data, loc_t *loc,
            fd_t *fd, dict_t *xdata);
 
 void
 ec_readdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_readdir_cbk_t func, void *data, fd_t *fd,
+           uint32_t fop_flags, fop_readdir_cbk_t func, void *data, fd_t *fd,
            size_t size, off_t offset, dict_t *xdata);
 
 void
 ec_readdirp(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_readdirp_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_readdirp_cbk_t func, void *data, fd_t *fd,
             size_t size, off_t offset, dict_t *xdata);
 
 void
 ec_readlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_readlink_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_readlink_cbk_t func, void *data, loc_t *loc,
             size_t size, dict_t *xdata);
 
 void
-ec_readv(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_readv_cbk_t func, void *data, fd_t *fd, size_t size, off_t offset,
-         uint32_t flags, dict_t *xdata);
+ec_readv(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_readv_cbk_t func, void *data, fd_t *fd,
+         size_t size, off_t offset, uint32_t flags, dict_t *xdata);
 
 void
 ec_removexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-               int32_t minimum, fop_removexattr_cbk_t func, void *data,
+               uint32_t fop_flags, fop_removexattr_cbk_t func, void *data,
                loc_t *loc, const char *name, dict_t *xdata);
 
 void
 ec_fremovexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-                int32_t minimum, fop_fremovexattr_cbk_t func, void *data,
+                uint32_t fop_flags, fop_fremovexattr_cbk_t func, void *data,
                 fd_t *fd, const char *name, dict_t *xdata);
 
 void
 ec_rename(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_rename_cbk_t func, void *data, loc_t *oldloc,
+          uint32_t fop_flags, fop_rename_cbk_t func, void *data, loc_t *oldloc,
           loc_t *newloc, dict_t *xdata);
 
 void
-ec_rmdir(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_rmdir_cbk_t func, void *data, loc_t *loc, int xflags,
-         dict_t *xdata);
+ec_rmdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_rmdir_cbk_t func, void *data, loc_t *loc,
+         int xflags, dict_t *xdata);
 
 void
 ec_setattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_setattr_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_setattr_cbk_t func, void *data, loc_t *loc,
            struct iatt *stbuf, int32_t valid, dict_t *xdata);
 
 void
 ec_fsetattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fsetattr_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fsetattr_cbk_t func, void *data, fd_t *fd,
             struct iatt *stbuf, int32_t valid, dict_t *xdata);
 
 void
 ec_setxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_setxattr_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_setxattr_cbk_t func, void *data, loc_t *loc,
             dict_t *dict, int32_t flags, dict_t *xdata);
 
 void
 ec_fsetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fsetxattr_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fsetxattr_cbk_t func, void *data, fd_t *fd,
              dict_t *dict, int32_t flags, dict_t *xdata);
 
 void
-ec_stat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_stat_cbk_t func, void *data, loc_t *loc, dict_t *xdata);
+ec_stat(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_stat_cbk_t func, void *data, loc_t *loc,
+        dict_t *xdata);
 
 void
-ec_fstat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fstat_cbk_t func, void *data, fd_t *fd, dict_t *xdata);
+ec_fstat(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fstat_cbk_t func, void *data, fd_t *fd,
+         dict_t *xdata);
 
 void
 ec_statfs(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_statfs_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_statfs_cbk_t func, void *data, loc_t *loc,
           dict_t *xdata);
 
 void
 ec_symlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_symlink_cbk_t func, void *data,
+           uint32_t fop_flags, fop_symlink_cbk_t func, void *data,
            const char *linkname, loc_t *loc, mode_t umask, dict_t *xdata);
 
 void
 ec_fallocate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fallocate_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fallocate_cbk_t func, void *data, fd_t *fd,
              int32_t mode, off_t offset, size_t len, dict_t *xdata);
 
 void
 ec_discard(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_discard_cbk_t func, void *data, fd_t *fd,
+           uint32_t fop_flags, fop_discard_cbk_t func, void *data, fd_t *fd,
            off_t offset, size_t len, dict_t *xdata);
 
 void
 ec_truncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_truncate_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_truncate_cbk_t func, void *data, loc_t *loc,
             off_t offset, dict_t *xdata);
 
 void
 ec_ftruncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_ftruncate_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_ftruncate_cbk_t func, void *data, fd_t *fd,
              off_t offset, dict_t *xdata);
 
 void
 ec_unlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_unlink_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_unlink_cbk_t func, void *data, loc_t *loc,
           int xflags, dict_t *xdata);
 
 void
 ec_writev(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_writev_cbk_t func, void *data, fd_t *fd,
+          uint32_t fop_flags, fop_writev_cbk_t func, void *data, fd_t *fd,
           struct iovec *vector, int32_t count, off_t offset, uint32_t flags,
           struct iobref *iobref, dict_t *xdata);
 
 void
 ec_xattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_xattrop_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_xattrop_cbk_t func, void *data, loc_t *loc,
            gf_xattrop_flags_t optype, dict_t *xattr, dict_t *xdata);
 
 void
 ec_fxattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fxattrop_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fxattrop_cbk_t func, void *data, fd_t *fd,
             gf_xattrop_flags_t optype, dict_t *xattr, dict_t *xdata);
 
 void
-ec_seek(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_seek_cbk_t func, void *data, fd_t *fd, off_t offset,
-        gf_seek_what_t what, dict_t *xdata);
+ec_seek(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_seek_cbk_t func, void *data, fd_t *fd,
+        off_t offset, gf_seek_what_t what, dict_t *xdata);
 
 void
-ec_ipc(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-       fop_ipc_cbk_t func, void *data, int32_t op, dict_t *xdata);
+ec_ipc(call_frame_t *frame, xlator_t *this, uintptr_t target,
+       uint32_t fop_flags, fop_ipc_cbk_t func, void *data, int32_t op,
+       dict_t *xdata);
 
 #endif /* __EC_FOPS_H__ */
diff --git a/xlators/cluster/ec/src/ec-galois.c b/xlators/cluster/ec/src/ec-galois.c
index 8cb4dc2e4e3..6e4990c71f5 100644
--- a/xlators/cluster/ec/src/ec-galois.c
+++ b/xlators/cluster/ec/src/ec-galois.c
@@ -10,9 +10,6 @@
 
 #include <string.h>
 
-#include "mem-pool.h"
-#include "list.h"
-
 #include "ec-mem-types.h"
 #include "ec-gf8.h"
 #include "ec-helpers.h"
diff --git a/xlators/cluster/ec/src/ec-generic.c b/xlators/cluster/ec/src/ec-generic.c
index 2fd10526494..884deb93669 100644
--- a/xlators/cluster/ec/src/ec-generic.c
+++ b/xlators/cluster/ec/src/ec-generic.c
@@ -8,16 +8,13 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 
 #include "ec.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
-#include "ec-method.h"
 #include "ec-fops.h"
 
 /* FOP: flush */
@@ -150,9 +147,41 @@ ec_manager_flush(ec_fop_data_t *fop, int32_t state)
     }
 }
 
+static int32_t
+ec_validate_fd(fd_t *fd, xlator_t *xl)
+{
+    uint64_t iversion = 0;
+    uint64_t fversion = 0;
+    ec_inode_t *inode_ctx = NULL;
+    ec_fd_t *fd_ctx = NULL;
+
+    LOCK(&fd->lock);
+    {
+        fd_ctx = __ec_fd_get(fd, xl);
+        if (fd_ctx) {
+            fversion = fd_ctx->bad_version;
+        }
+    }
+    UNLOCK(&fd->lock);
+
+    LOCK(&fd->inode->lock);
+    {
+        inode_ctx = __ec_inode_get(fd->inode, xl);
+        if (inode_ctx) {
+            iversion = inode_ctx->bad_version;
+        }
+    }
+    UNLOCK(&fd->inode->lock);
+    if (fversion < iversion) {
+        return EBADF;
+    }
+    return 0;
+}
+
 void
-ec_flush(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_flush_cbk_t func, void *data, fd_t *fd, dict_t *xdata)
+ec_flush(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_flush_cbk_t func, void *data, fd_t *fd,
+         dict_t *xdata)
 {
     ec_cbk_t callback = {.flush = func};
     ec_fop_data_t *fop = NULL;
@@ -164,7 +193,17 @@ ec_flush(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FLUSH, 0, target, minimum,
+    if (fd) {
+        error = ec_validate_fd(fd, this);
+        if (error) {
+            gf_msg(this->name, GF_LOG_ERROR, EBADF, EC_MSG_FD_BAD,
+                   "Failing %s on %s", gf_fop_list[GF_FOP_FLUSH],
+                   fd->inode ? uuid_utoa(fd->inode->gfid) : "");
+            goto out;
+        }
+    }
+
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FLUSH, 0, target, fop_flags,
                                ec_wind_flush, ec_manager_flush, callback, data);
     if (fop == NULL) {
         goto out;
@@ -366,9 +405,9 @@ ec_manager_fsync(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_fsync(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fsync_cbk_t func, void *data, fd_t *fd, int32_t datasync,
-         dict_t *xdata)
+ec_fsync(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fsync_cbk_t func, void *data, fd_t *fd,
+         int32_t datasync, dict_t *xdata)
 {
     ec_cbk_t callback = {.fsync = func};
     ec_fop_data_t *fop = NULL;
@@ -380,7 +419,17 @@ ec_fsync(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSYNC, 0, target, minimum,
+    if (fd) {
+        error = ec_validate_fd(fd, this);
+        if (error) {
+            gf_msg(this->name, GF_LOG_ERROR, EBADF, EC_MSG_FD_BAD,
+                   "Failing %s on %s", gf_fop_list[GF_FOP_FSYNC],
+                   fd->inode ? uuid_utoa(fd->inode->gfid) : "");
+            goto out;
+        }
+    }
+
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSYNC, 0, target, fop_flags,
                                ec_wind_fsync, ec_manager_fsync, callback, data);
     if (fop == NULL) {
         goto out;
@@ -553,7 +602,7 @@ ec_manager_fsyncdir(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_fsyncdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fsyncdir_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fsyncdir_cbk_t func, void *data, fd_t *fd,
             int32_t datasync, dict_t *xdata)
 {
     ec_cbk_t callback = {.fsyncdir = func};
@@ -566,9 +615,9 @@ ec_fsyncdir(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSYNCDIR, 0, target, minimum,
-                               ec_wind_fsyncdir, ec_manager_fsyncdir, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSYNCDIR, 0, target,
+                               fop_flags, ec_wind_fsyncdir, ec_manager_fsyncdir,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -848,7 +897,7 @@ ec_manager_lookup(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_lookup(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_lookup_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_lookup_cbk_t func, void *data, loc_t *loc,
           dict_t *xdata)
 {
     ec_cbk_t callback = {.lookup = func};
@@ -862,7 +911,7 @@ ec_lookup(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_LOOKUP, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_lookup,
+                               target, fop_flags, ec_wind_lookup,
                                ec_manager_lookup, callback, data);
     if (fop == NULL) {
         goto out;
@@ -1033,7 +1082,7 @@ ec_manager_statfs(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_statfs(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_statfs_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_statfs_cbk_t func, void *data, loc_t *loc,
           dict_t *xdata)
 {
     ec_cbk_t callback = {.statfs = func};
@@ -1047,7 +1096,7 @@ ec_statfs(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_STATFS, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_statfs,
+                               target, fop_flags, ec_wind_statfs,
                                ec_manager_statfs, callback, data);
     if (fop == NULL) {
         goto out;
@@ -1270,7 +1319,7 @@ ec_manager_xattrop(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_xattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_xattrop_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_xattrop_cbk_t func, void *data, loc_t *loc,
            gf_xattrop_flags_t optype, dict_t *xattr, dict_t *xdata)
 {
     ec_cbk_t callback = {.xattrop = func};
@@ -1283,9 +1332,9 @@ ec_xattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_XATTROP, 0, target, minimum,
-                               ec_wind_xattrop, ec_manager_xattrop, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_XATTROP, 0, target,
+                               fop_flags, ec_wind_xattrop, ec_manager_xattrop,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1343,7 +1392,7 @@ ec_wind_fxattrop(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fxattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fxattrop_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fxattrop_cbk_t func, void *data, fd_t *fd,
             gf_xattrop_flags_t optype, dict_t *xattr, dict_t *xdata)
 {
     ec_cbk_t callback = {.fxattrop = func};
@@ -1356,9 +1405,9 @@ ec_fxattrop(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FXATTROP, 0, target, minimum,
-                               ec_wind_fxattrop, ec_manager_xattrop, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FXATTROP, 0, target,
+                               fop_flags, ec_wind_fxattrop, ec_manager_xattrop,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1507,8 +1556,9 @@ ec_manager_ipc(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_ipc(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-       fop_ipc_cbk_t func, void *data, int32_t op, dict_t *xdata)
+ec_ipc(call_frame_t *frame, xlator_t *this, uintptr_t target,
+       uint32_t fop_flags, fop_ipc_cbk_t func, void *data, int32_t op,
+       dict_t *xdata)
 {
     ec_cbk_t callback = {.ipc = func};
     ec_fop_data_t *fop = NULL;
@@ -1520,7 +1570,7 @@ ec_ipc(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_IPC, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_IPC, 0, target, fop_flags,
                                ec_wind_ipc, ec_manager_ipc, callback, data);
     if (fop == NULL) {
         goto out;
diff --git a/xlators/cluster/ec/src/ec-heal.c b/xlators/cluster/ec/src/ec-heal.c
index 9eb5b856932..7d991f04aac 100644
--- a/xlators/cluster/ec/src/ec-heal.c
+++ b/xlators/cluster/ec/src/ec-heal.c
@@ -8,16 +8,14 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-#include "compat-errno.h"
-#include "byte-order.h"
-#include "syncop.h"
-#include "syncop-utils.h"
-#include "cluster-syncop.h"
+#include <glusterfs/defaults.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/byte-order.h>
+#include <glusterfs/syncop.h>
+#include <glusterfs/syncop-utils.h>
+#include <glusterfs/cluster-syncop.h>
 
 #include "ec.h"
-#include "ec-mem-types.h"
 #include "ec-types.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
@@ -72,6 +70,7 @@ struct ec_name_data {
     char *name;
     inode_t *parent;
     default_args_cbk_t *replies;
+    uint32_t heal_pending;
 };
 
 static char *ec_ignore_xattrs[] = {GF_SELINUX_XATTR_KEY, QUOTA_SIZE_KEY, NULL};
@@ -103,6 +102,48 @@ ec_sh_key_match(dict_t *dict, char *key, data_t *val, void *mdata)
 }
 /* FOP: heal */
 
+void
+ec_set_entry_healing(ec_fop_data_t *fop)
+{
+    ec_inode_t *ctx = NULL;
+    loc_t *loc = NULL;
+
+    if (!fop)
+        return;
+
+    loc = &fop->loc[0];
+    LOCK(&loc->inode->lock);
+    {
+        ctx = __ec_inode_get(loc->inode, fop->xl);
+        if (ctx) {
+            ctx->heal_count += 1;
+        }
+    }
+    UNLOCK(&loc->inode->lock);
+}
+
+void
+ec_reset_entry_healing(ec_fop_data_t *fop)
+{
+    ec_inode_t *ctx = NULL;
+    loc_t *loc = NULL;
+    int32_t heal_count = 0;
+    if (!fop)
+        return;
+
+    loc = &fop->loc[0];
+    LOCK(&loc->inode->lock);
+    {
+        ctx = __ec_inode_get(loc->inode, fop->xl);
+        if (ctx) {
+            ctx->heal_count += -1;
+            heal_count = ctx->heal_count;
+        }
+    }
+    UNLOCK(&loc->inode->lock);
+    GF_ASSERT(heal_count >= 0);
+}
+
 uintptr_t
 ec_heal_check(ec_fop_data_t *fop, uintptr_t *pgood)
 {
@@ -325,16 +366,16 @@ ec_heal_data_block(ec_heal_t *heal)
 /* FOP: fheal */
 
 void
-ec_fheal(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fheal_cbk_t func, void *data, fd_t *fd, int32_t partial,
-         dict_t *xdata)
+ec_fheal(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fheal_cbk_t func, void *data, fd_t *fd,
+         int32_t partial, dict_t *xdata)
 {
     ec_fd_t *ctx = ec_fd_get(fd, this);
 
     if (ctx != NULL) {
         gf_msg_trace("ec", 0, "FHEAL ctx: flags=%X, open=%" PRIXPTR, ctx->flags,
                      ctx->open);
-        ec_heal(frame, this, target, minimum, func, data, &ctx->loc, partial,
+        ec_heal(frame, this, target, fop_flags, func, data, &ctx->loc, partial,
                 xdata);
     }
 }
@@ -954,6 +995,7 @@ ec_set_new_entry_dirty(ec_t *ec, loc_t *loc, struct iatt *ia,
         ret = -ENOTCONN;
         goto out;
     }
+
 out:
     if (xattr)
         dict_unref(xattr);
@@ -977,6 +1019,7 @@ ec_delete_stale_name(dict_t *gfid_db, char *key, data_t *d, void *data)
     int estale_count = 0;
     int i = 0;
     call_frame_t *frame = name_data->frame;
+    uuid_t gfid;
 
     ec = name_data->frame->this->private;
     EC_REPLIES_ALLOC(replies, ec->nodes);
@@ -985,12 +1028,16 @@ ec_delete_stale_name(dict_t *gfid_db, char *key, data_t *d, void *data)
         goto out;
     }
 
+    loc.parent = inode_ref(name_data->parent);
     loc.inode = inode_new(name_data->parent->table);
     if (!loc.inode) {
         ret = -ENOMEM;
         goto out;
     }
-    gf_uuid_parse(key, loc.gfid);
+
+    gf_uuid_parse(key, gfid);
+    gf_uuid_copy(loc.pargfid, name_data->parent->gfid);
+    loc.name = name_data->name;
     output = alloca0(ec->nodes);
     ret = cluster_lookup(ec->xl_list, name_data->participants, ec->nodes,
                          replies, output, name_data->frame, ec->xl, &loc, NULL);
@@ -1003,6 +1050,11 @@ ec_delete_stale_name(dict_t *gfid_db, char *key, data_t *d, void *data)
                 estale_count++;
             else
                 name_data->participants[i] = 0;
+        } else if (gf_uuid_compare(gfid, replies[i].stat.ia_gfid)) {
+            estale_count++;
+            gf_msg_debug(ec->xl->name, 0, "%s/%s: different gfid as %s",
+                         uuid_utoa(name_data->parent->gfid), name_data->name,
+                         key);
         }
     }
 
@@ -1122,6 +1174,7 @@ ec_create_name(call_frame_t *frame, ec_t *ec, inode_t *parent, char *name,
     dict_t *xdata = NULL;
     char *linkname = NULL;
     ec_config_t config;
+
     /* There should be just one gfid key */
     EC_REPLIES_ALLOC(replies, ec->nodes);
     if (gfid_db->count != 1) {
@@ -1366,6 +1419,11 @@ __ec_heal_name(call_frame_t *frame, ec_t *ec, inode_t *parent, char *name,
 
     ret = ec_create_name(frame, ec, parent, name, replies, gfid_db, enoent,
                          participants);
+    if (ret >= 0) {
+        /* If ec_create_name() succeeded we return 1 to indicate that a new
+         * file has been created and it will need to be healed. */
+        ret = 1;
+    }
 out:
     cluster_replies_wipe(replies, ec->nodes);
     loc_wipe(&loc);
@@ -1443,18 +1501,22 @@ ec_name_heal_handler(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
     ret = ec_heal_name(name_data->frame, ec, parent->inode, entry->d_name,
                        name_on);
 
-    if (ret < 0)
+    if (ret < 0) {
         memset(name_on, 0, ec->nodes);
+    } else {
+        name_data->heal_pending += ret;
+    }
 
     for (i = 0; i < ec->nodes; i++)
         if (name_data->participants[i] && !name_on[i])
             name_data->failed_on[i] = 1;
+
     return 0;
 }
 
 int
 ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
-              unsigned char *participants)
+              unsigned char *participants, uint32_t *pending)
 {
     int i = 0;
     int j = 0;
@@ -1467,7 +1529,7 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
     name_data.frame = frame;
     name_data.participants = participants;
     name_data.failed_on = alloca0(ec->nodes);
-    ;
+    name_data.heal_pending = 0;
 
     for (i = 0; i < ec->nodes; i++) {
         if (!participants[i])
@@ -1486,6 +1548,8 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
             break;
         }
     }
+    *pending += name_data.heal_pending;
+
     loc_wipe(&loc);
     return ret;
 }
@@ -1493,7 +1557,7 @@ ec_heal_names(call_frame_t *frame, ec_t *ec, inode_t *inode,
 int
 __ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
                 unsigned char *heal_on, unsigned char *sources,
-                unsigned char *healed_sinks)
+                unsigned char *healed_sinks, uint32_t *pending)
 {
     unsigned char *locked_on = NULL;
     unsigned char *output = NULL;
@@ -1538,7 +1602,7 @@ unlock:
         if (sources[i] || healed_sinks[i])
             participants[i] = 1;
     }
-    ret = ec_heal_names(frame, ec, inode, participants);
+    ret = ec_heal_names(frame, ec, inode, participants, pending);
 
     if (EC_COUNT(participants, ec->nodes) <= ec->fragments)
         goto out;
@@ -1559,7 +1623,8 @@ out:
 
 int
 ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
-              unsigned char *sources, unsigned char *healed_sinks)
+              unsigned char *sources, unsigned char *healed_sinks,
+              uint32_t *pending)
 {
     unsigned char *locked_on = NULL;
     unsigned char *up_subvols = NULL;
@@ -1590,7 +1655,7 @@ ec_heal_entry(call_frame_t *frame, ec_t *ec, inode_t *inode,
             goto unlock;
         }
         ret = __ec_heal_entry(frame, ec, inode, locked_on, sources,
-                              healed_sinks);
+                              healed_sinks, pending);
     }
 unlock:
     cluster_uninodelk(ec->xl_list, locked_on, ec->nodes, replies, output, frame,
@@ -1909,16 +1974,16 @@ ec_manager_heal_block(ec_fop_data_t *fop, int32_t state)
 
         case EC_STATE_REPORT:
             if (fop->cbks.heal) {
-                fop->cbks.heal(fop->req_frame, fop, fop->xl, 0, 0,
+                fop->cbks.heal(fop->req_frame, fop->data, fop->xl, 0, 0,
                                (heal->good | heal->bad), heal->good, heal->bad,
-                               NULL);
+                               0, NULL);
             }
 
             return EC_STATE_END;
         case -EC_STATE_REPORT:
             if (fop->cbks.heal) {
-                fop->cbks.heal(fop->req_frame, fop, fop->xl, -1, fop->error, 0,
-                               0, 0, NULL);
+                fop->cbks.heal(fop->req_frame, fop->data, fop->xl, -1,
+                               fop->error, 0, 0, 0, 0, NULL);
             }
 
             return EC_STATE_END;
@@ -1933,7 +1998,7 @@ ec_manager_heal_block(ec_fop_data_t *fop, int32_t state)
 /*Takes lock */
 void
 ec_heal_block(call_frame_t *frame, xlator_t *this, uintptr_t target,
-              int32_t minimum, fop_heal_cbk_t func, ec_heal_t *heal)
+              uint32_t fop_flags, fop_heal_cbk_t func, ec_heal_t *heal)
 {
     ec_cbk_t callback = {.heal = func};
     ec_fop_data_t *fop = NULL;
@@ -1944,7 +2009,7 @@ ec_heal_block(call_frame_t *frame, xlator_t *this, uintptr_t target,
     VALIDATE_OR_GOTO(this, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, EC_FOP_HEAL, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, EC_FOP_HEAL, 0, target, fop_flags,
                                NULL, ec_manager_heal_block, callback, heal);
     if (fop == NULL)
         goto out;
@@ -1955,19 +2020,21 @@ out:
     if (fop != NULL) {
         ec_manager(fop, error);
     } else {
-        func(frame, NULL, this, -1, error, 0, 0, 0, NULL);
+        func(frame, heal, this, -1, error, 0, 0, 0, 0, NULL);
     }
 }
 
 int32_t
 ec_heal_block_done(call_frame_t *frame, void *cookie, xlator_t *this,
                    int32_t op_ret, int32_t op_errno, uintptr_t mask,
-                   uintptr_t good, uintptr_t bad, dict_t *xdata)
+                   uintptr_t good, uintptr_t bad, uint32_t pending,
+                   dict_t *xdata)
 {
-    ec_fop_data_t *fop = cookie;
-    ec_heal_t *heal = fop->data;
+    ec_heal_t *heal = cookie;
 
-    fop->heal = NULL;
+    if (heal->fop) {
+        heal->fop->heal = NULL;
+    }
     heal->fop = NULL;
     heal->error = op_ret < 0 ? op_errno : 0;
     syncbarrier_wake(heal->data);
@@ -2259,9 +2326,10 @@ ec_restore_time_and_adjust_versions(call_frame_t *frame, ec_t *ec, fd_t *fd,
 
         loc.inode = inode_ref(fd->inode);
         gf_uuid_copy(loc.gfid, fd->inode->gfid);
-        ret = cluster_setattr(ec->xl_list, healed_sinks, ec->nodes, replies,
-                              output, frame, ec->xl, &loc, &source_buf,
-                              GF_SET_ATTR_ATIME | GF_SET_ATTR_MTIME, NULL);
+        ret = cluster_setattr(
+            ec->xl_list, healed_sinks, ec->nodes, replies, output, frame,
+            ec->xl, &loc, &source_buf,
+            GF_SET_ATTR_ATIME | GF_SET_ATTR_MTIME | GF_SET_ATTR_CTIME, NULL);
         EC_INTERSECT(healed_sinks, healed_sinks, output, ec->nodes);
         if (EC_COUNT(healed_sinks, ec->nodes) == 0) {
             ret = -ENOTCONN;
@@ -2429,6 +2497,58 @@ out:
     return ret;
 }
 
+int
+ec_heal_purge_stale_index(call_frame_t *frame, ec_t *ec, inode_t *inode)
+{
+    int i = 0;
+    int ret = 0;
+    dict_t **xattr = NULL;
+    loc_t loc = {0};
+    uint64_t dirty_xattr[EC_VERSION_SIZE] = {0};
+    unsigned char *on = NULL;
+    default_args_cbk_t *replies = NULL;
+    dict_t *dict = NULL;
+
+    /* Allocate the required memory */
+    loc.inode = inode_ref(inode);
+    gf_uuid_copy(loc.gfid, inode->gfid);
+    on = alloca0(ec->nodes);
+    EC_REPLIES_ALLOC(replies, ec->nodes);
+    xattr = GF_CALLOC(ec->nodes, sizeof(*xattr), gf_common_mt_pointer);
+    if (!xattr) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    dict = dict_new();
+    if (!dict) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    for (i = 0; i < ec->nodes; i++) {
+        xattr[i] = dict;
+        on[i] = 1;
+    }
+    ret = dict_set_static_bin(dict, EC_XATTR_DIRTY, dirty_xattr,
+                              (sizeof(*dirty_xattr) * EC_VERSION_SIZE));
+    if (ret < 0) {
+        ret = -ENOMEM;
+        goto out;
+    }
+    PARALLEL_FOP_ONLIST(ec->xl_list, on, ec->nodes, replies, frame,
+                        ec_wind_xattrop_parallel, &loc, GF_XATTROP_ADD_ARRAY64,
+                        xattr, NULL);
+out:
+    if (dict) {
+        dict_unref(dict);
+    }
+    if (xattr) {
+        GF_FREE(xattr);
+    }
+    cluster_replies_wipe(replies, ec->nodes);
+    loc_wipe(&loc);
+    return ret;
+}
+
 void
 ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
 {
@@ -2446,6 +2566,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
     intptr_t mbad = 0;
     intptr_t good = 0;
     intptr_t bad = 0;
+    uint32_t pending = 0;
     ec_fop_data_t *fop = data;
     gf_boolean_t blocking = _gf_false;
     ec_heal_need_t need_heal = EC_HEAL_NONEED;
@@ -2481,7 +2602,7 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
     if (loc->name && strlen(loc->name)) {
         ret = ec_heal_name(frame, ec, loc->parent, (char *)loc->name,
                            participants);
-        if (ret == 0) {
+        if (ret >= 0) {
             gf_msg_debug(this->name, 0,
                          "%s: name heal "
                          "successful on %" PRIXPTR,
@@ -2499,32 +2620,34 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
 
     /* Mount triggers heal only when it detects that it must need heal, shd
      * triggers heals periodically which need not be thorough*/
-    ec_heal_inspect(frame, ec, loc->inode, up_subvols, _gf_false,
-                    !ec->shd.iamshd, &need_heal);
-
-    if (need_heal == EC_HEAL_NONEED) {
-        gf_msg(ec->xl->name, GF_LOG_DEBUG, 0, EC_MSG_HEAL_FAIL,
-               "Heal is not required for : %s ", uuid_utoa(loc->gfid));
-        goto out;
+    if (ec->shd.iamshd && (ret <= 0)) {
+        ec_heal_inspect(frame, ec, loc->inode, up_subvols, _gf_false, _gf_false,
+                        &need_heal);
+
+        if (need_heal == EC_HEAL_PURGE_INDEX) {
+            gf_msg(ec->xl->name, GF_LOG_INFO, 0, EC_MSG_HEAL_FAIL,
+                   "Index entry needs to be purged for: %s ",
+                   uuid_utoa(loc->gfid));
+            /* We need to send zero-xattrop so that stale index entry could be
+             * removed. We need not take lock on this entry to do so as
+             * xattrop on a brick is atomic. */
+            ec_heal_purge_stale_index(frame, ec, loc->inode);
+            goto out;
+        } else if (need_heal == EC_HEAL_NONEED) {
+            gf_msg(ec->xl->name, GF_LOG_DEBUG, 0, EC_MSG_HEAL_FAIL,
+                   "Heal is not required for : %s ", uuid_utoa(loc->gfid));
+            goto out;
+        }
     }
 
-    msources = alloca0(ec->nodes);
-    mhealed_sinks = alloca0(ec->nodes);
-    ret = ec_heal_metadata(frame, ec, loc->inode, msources, mhealed_sinks);
-    if (ret == 0) {
-        mgood = ec_char_array_to_mask(msources, ec->nodes);
-        mbad = ec_char_array_to_mask(mhealed_sinks, ec->nodes);
-    } else {
-        op_ret = -1;
-        op_errno = -ret;
-    }
     sources = alloca0(ec->nodes);
     healed_sinks = alloca0(ec->nodes);
     if (IA_ISREG(loc->inode->ia_type)) {
         ret = ec_heal_data(frame, ec, blocking, loc->inode, sources,
                            healed_sinks);
     } else if (IA_ISDIR(loc->inode->ia_type) && !partial) {
-        ret = ec_heal_entry(frame, ec, loc->inode, sources, healed_sinks);
+        ret = ec_heal_entry(frame, ec, loc->inode, sources, healed_sinks,
+                            &pending);
     } else {
         ret = 0;
         memcpy(sources, participants, ec->nodes);
@@ -2538,15 +2661,27 @@ ec_heal_do(xlator_t *this, void *data, loc_t *loc, int32_t partial)
         op_ret = -1;
         op_errno = -ret;
     }
+    msources = alloca0(ec->nodes);
+    mhealed_sinks = alloca0(ec->nodes);
+    ret = ec_heal_metadata(frame, ec, loc->inode, msources, mhealed_sinks);
+    if (ret == 0) {
+        mgood = ec_char_array_to_mask(msources, ec->nodes);
+        mbad = ec_char_array_to_mask(mhealed_sinks, ec->nodes);
+    } else {
+        op_ret = -1;
+        op_errno = -ret;
+    }
 
 out:
+    ec_reset_entry_healing(fop);
     if (fop->cbks.heal) {
-        fop->cbks.heal(fop->req_frame, fop, fop->xl, op_ret, op_errno,
+        fop->cbks.heal(fop->req_frame, fop->data, fop->xl, op_ret, op_errno,
                        ec_char_array_to_mask(participants, ec->nodes),
-                       mgood & good, mbad & bad, NULL);
+                       mgood & good, mbad & bad, pending, NULL);
     }
     if (frame)
         STACK_DESTROY(frame->root);
+
     return;
 }
 
@@ -2593,8 +2728,8 @@ void
 ec_heal_fail(ec_t *ec, ec_fop_data_t *fop)
 {
     if (fop->cbks.heal) {
-        fop->cbks.heal(fop->req_frame, NULL, ec->xl, -1, fop->error, 0, 0, 0,
-                       NULL);
+        fop->cbks.heal(fop->req_frame, fop->data, ec->xl, -1, fop->error, 0, 0,
+                       0, 0, NULL);
     }
     ec_fop_data_release(fop);
 }
@@ -2603,13 +2738,31 @@ void
 ec_launch_heal(ec_t *ec, ec_fop_data_t *fop)
 {
     int ret = 0;
+    call_frame_t *frame = NULL;
+
+    frame = create_frame(ec->xl, ec->xl->ctx->pool);
+    if (!frame) {
+        ret = -1;
+        goto out;
+    }
+
+    ec_owner_set(frame, frame->root);
+    /*Do heal as root*/
+    frame->root->uid = 0;
+    frame->root->gid = 0;
+    /*Mark the fops as internal*/
+    frame->root->pid = GF_CLIENT_PID_SELF_HEALD;
 
     ret = synctask_new(ec->xl->ctx->env, ec_synctask_heal_wrap, ec_heal_done,
-                       NULL, fop);
+                       frame, fop);
+out:
     if (ret < 0) {
         ec_fop_set_error(fop, ENOMEM);
         ec_heal_fail(ec, fop);
     }
+
+    if (frame)
+        STACK_DESTROY(frame->root);
 }
 
 void
@@ -2650,11 +2803,33 @@ ec_handle_healers_done(ec_fop_data_t *fop)
         ec_launch_heal(ec, heal_fop);
 }
 
+gf_boolean_t
+ec_is_entry_healing(ec_fop_data_t *fop)
+{
+    ec_inode_t *ctx = NULL;
+    int32_t heal_count = 0;
+    loc_t *loc = NULL;
+
+    loc = &fop->loc[0];
+
+    LOCK(&loc->inode->lock);
+    {
+        ctx = __ec_inode_get(loc->inode, fop->xl);
+        if (ctx) {
+            heal_count = ctx->heal_count;
+        }
+    }
+    UNLOCK(&loc->inode->lock);
+    GF_ASSERT(heal_count >= 0);
+    return heal_count;
+}
+
 void
 ec_heal_throttle(xlator_t *this, ec_fop_data_t *fop)
 {
     gf_boolean_t can_heal = _gf_true;
     ec_t *ec = this->private;
+    ec_fop_data_t *fop_rel = NULL;
 
     if (fop->req_frame == NULL) {
         LOCK(&ec->lock);
@@ -2662,8 +2837,13 @@ ec_heal_throttle(xlator_t *this, ec_fop_data_t *fop)
             if ((ec->background_heals > 0) &&
                 (ec->heal_wait_qlen + ec->background_heals) >
                     (ec->heal_waiters + ec->healers)) {
-                list_add_tail(&fop->healer, &ec->heal_waiting);
-                ec->heal_waiters++;
+                if (!ec_is_entry_healing(fop)) {
+                    list_add_tail(&fop->healer, &ec->heal_waiting);
+                    ec->heal_waiters++;
+                    ec_set_entry_healing(fop);
+                } else {
+                    fop_rel = fop;
+                }
                 fop = __ec_dequeue_heals(ec);
             } else {
                 can_heal = _gf_false;
@@ -2673,8 +2853,12 @@ ec_heal_throttle(xlator_t *this, ec_fop_data_t *fop)
     }
 
     if (can_heal) {
-        if (fop)
+        if (fop) {
+            if (fop->req_frame != NULL) {
+                ec_set_entry_healing(fop);
+            }
             ec_launch_heal(ec, fop);
+        }
     } else {
         gf_msg_debug(this->name, 0,
                      "Max number of heals are "
@@ -2682,12 +2866,15 @@ ec_heal_throttle(xlator_t *this, ec_fop_data_t *fop)
         ec_fop_set_error(fop, EBUSY);
         ec_heal_fail(ec, fop);
     }
+    if (fop_rel) {
+        ec_heal_done(0, NULL, fop_rel);
+    }
 }
 
 void
-ec_heal(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_heal_cbk_t func, void *data, loc_t *loc, int32_t partial,
-        dict_t *xdata)
+ec_heal(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_heal_cbk_t func, void *data, loc_t *loc,
+        int32_t partial, dict_t *xdata)
 {
     ec_cbk_t callback = {.heal = func};
     ec_fop_data_t *fop = NULL;
@@ -2703,7 +2890,7 @@ ec_heal(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
 
     if (frame && frame->local)
         goto fail;
-    fop = ec_fop_data_allocate(frame, this, EC_FOP_HEAL, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, EC_FOP_HEAL, 0, target, fop_flags,
                                NULL, NULL, callback, data);
 
     err = ENOMEM;
@@ -2729,15 +2916,27 @@ fail:
     if (fop)
         ec_fop_data_release(fop);
     if (func)
-        func(frame, NULL, this, -1, err, 0, 0, 0, NULL);
+        func(frame, data, this, -1, err, 0, 0, 0, 0, NULL);
 }
 
 int
 ec_replace_heal_done(int ret, call_frame_t *heal, void *opaque)
 {
     ec_t *ec = opaque;
+    gf_boolean_t last_fop = _gf_false;
 
+    if (GF_ATOMIC_DEC(ec->async_fop_count) == 0) {
+        LOCK(&ec->lock);
+        {
+            last_fop = __ec_is_last_fop(ec);
+        }
+        UNLOCK(&ec->lock);
+    }
     gf_msg_debug(ec->xl->name, 0, "getxattr on bricks is done ret %d", ret);
+
+    if (last_fop)
+        ec_pending_fops_completed(ec);
+
     return 0;
 }
 
@@ -2777,6 +2976,10 @@ ec_replace_brick_heal_wrap(void *opaque)
         itable = ec->xl->itable;
     else
         goto out;
+
+    if (xlator_is_cleanup_starting(ec->xl))
+        goto out;
+
     ret = ec_replace_heal(ec, itable->root);
 out:
     return ret;
@@ -2787,14 +2990,15 @@ ec_launch_replace_heal(ec_t *ec)
 {
     int ret = -1;
 
-    if (!ec)
-        return ret;
     ret = synctask_new(ec->xl->ctx->env, ec_replace_brick_heal_wrap,
                        ec_replace_heal_done, NULL, ec);
+
     if (ret < 0) {
         gf_msg_debug(ec->xl->name, 0, "Heal failed for replace brick ret = %d",
                      ret);
+        ec_replace_heal_done(-1, NULL, ec);
     }
+
     return ret;
 }
 
@@ -2826,7 +3030,7 @@ out:
 static int32_t
 _need_heal_calculate(ec_t *ec, uint64_t *dirty, unsigned char *sources,
                      gf_boolean_t self_locked, int32_t lock_count,
-                     ec_heal_need_t *need_heal)
+                     ec_heal_need_t *need_heal, uint64_t *versions)
 {
     int i = 0;
     int source_count = 0;
@@ -2836,11 +3040,18 @@ _need_heal_calculate(ec_t *ec, uint64_t *dirty, unsigned char *sources,
         *need_heal = EC_HEAL_NONEED;
         if (self_locked || lock_count == 0) {
             for (i = 0; i < ec->nodes; i++) {
-                if (dirty[i]) {
+                if (dirty[i] || (versions[i] != versions[0])) {
                     *need_heal = EC_HEAL_MUST;
                     goto out;
                 }
             }
+            /* If lock count is 0, all dirty flags are 0 and all the
+             * versions are macthing then why are we here. It looks
+             * like something went wrong while removing the index entries
+             * after completing a successful heal or fop. In this case
+             * we need to remove this index entry to avoid triggering heal
+             * in a loop and causing lookups again and again*/
+            *need_heal = EC_HEAL_PURGE_INDEX;
         } else {
             for (i = 0; i < ec->nodes; i++) {
                 /* Since each lock can only increment the dirty
@@ -2852,6 +3063,9 @@ _need_heal_calculate(ec_t *ec, uint64_t *dirty, unsigned char *sources,
                     *need_heal = EC_HEAL_MUST;
                     goto out;
                 }
+                if (dirty[i] != dirty[0] || (versions[i] != versions[0])) {
+                    *need_heal = EC_HEAL_MAYBE;
+                }
             }
         }
     } else {
@@ -2872,7 +3086,6 @@ ec_need_metadata_heal(ec_t *ec, inode_t *inode, default_args_cbk_t *replies,
     unsigned char *healed_sinks = NULL;
     uint64_t *meta_versions = NULL;
     int ret = 0;
-    int i = 0;
 
     sources = alloca0(ec->nodes);
     healed_sinks = alloca0(ec->nodes);
@@ -2885,15 +3098,7 @@ ec_need_metadata_heal(ec_t *ec, inode_t *inode, default_args_cbk_t *replies,
     }
 
     ret = _need_heal_calculate(ec, dirty, sources, self_locked, lock_count,
-                               need_heal);
-    if (ret == ec->nodes && *need_heal == EC_HEAL_NONEED) {
-        for (i = 1; i < ec->nodes; i++) {
-            if (meta_versions[i] != meta_versions[0]) {
-                *need_heal = EC_HEAL_MUST;
-                goto out;
-            }
-        }
-    }
+                               need_heal, meta_versions);
 out:
     return ret;
 }
@@ -2929,7 +3134,7 @@ ec_need_data_heal(ec_t *ec, inode_t *inode, default_args_cbk_t *replies,
     }
 
     ret = _need_heal_calculate(ec, dirty, sources, self_locked, lock_count,
-                               need_heal);
+                               need_heal, data_versions);
 out:
     return ret;
 }
@@ -2957,7 +3162,7 @@ ec_need_entry_heal(ec_t *ec, inode_t *inode, default_args_cbk_t *replies,
     }
 
     ret = _need_heal_calculate(ec, dirty, sources, self_locked, lock_count,
-                               need_heal);
+                               need_heal, data_versions);
 out:
     return ret;
 }
@@ -3055,10 +3260,6 @@ ec_heal_inspect(call_frame_t *frame, ec_t *ec, inode_t *inode,
 need_heal:
     ret = ec_need_heal(ec, inode, replies, lock_count, self_locked, thorough,
                        need_heal);
-
-    if (!self_locked && *need_heal == EC_HEAL_MUST) {
-        *need_heal = EC_HEAL_MAYBE;
-    }
 out:
     cluster_replies_wipe(replies, ec->nodes);
     loc_wipe(&loc);
@@ -3144,7 +3345,7 @@ ec_get_heal_info(xlator_t *this, loc_t *entry_loc, dict_t **dict_rsp)
 
     ret = ec_heal_inspect(frame, ec, loc.inode, up_subvols, _gf_false,
                           _gf_false, &need_heal);
-    if (ret == ec->nodes && need_heal == EC_HEAL_NONEED) {
+    if (ret == ec->nodes && need_heal != EC_HEAL_MAYBE) {
         goto set_heal;
     }
     need_heal = EC_HEAL_NONEED;
diff --git a/xlators/cluster/ec/src/ec-heald.c b/xlators/cluster/ec/src/ec-heald.c
index 130790c66ac..5c1586bc9c5 100644
--- a/xlators/cluster/ec/src/ec-heald.c
+++ b/xlators/cluster/ec/src/ec-heald.c
@@ -8,15 +8,14 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-#include "compat-errno.h"
+#include <glusterfs/defaults.h>
+#include <glusterfs/compat-errno.h>
 #include "ec.h"
 #include "ec-messages.h"
 #include "ec-heald.h"
 #include "ec-mem-types.h"
-#include "syncop.h"
-#include "syncop-utils.h"
+#include <glusterfs/syncop.h>
+#include <glusterfs/syncop-utils.h>
 #include "protocol-common.h"
 
 #define NTH_INDEX_HEALER(this, n)                                              \
@@ -63,7 +62,7 @@ __ec_shd_healer_wait(struct subvol_healer *healer)
     ec = healer->this->private;
 
 disabled_loop:
-    wait_till.tv_sec = time(NULL) + 60;
+    wait_till.tv_sec = gf_time() + ec->shd.timeout;
 
     while (!healer->rerun) {
         ret = pthread_cond_timedwait(&healer->cond, &healer->mutex, &wait_till);
@@ -71,6 +70,11 @@ disabled_loop:
             break;
     }
 
+    if (ec->shutdown) {
+        healer->running = _gf_false;
+        return -1;
+    }
+
     ret = healer->rerun;
     healer->rerun = 0;
 
@@ -152,19 +156,78 @@ ec_shd_index_purge(xlator_t *subvol, inode_t *inode, char *name)
     return ret;
 }
 
+static gf_boolean_t
+ec_is_heal_completed(char *status)
+{
+    char *bad_pos = NULL;
+    char *zero_pos = NULL;
+
+    if (!status) {
+        return _gf_false;
+    }
+
+    /*Logic:
+     * Status will be of the form Good: <binary>, Bad: <binary>
+     * If heal completes, if we do strchr for '0' it should be present after
+     * 'Bad:' i.e. strRchr for ':'
+     * */
+
+    zero_pos = strchr(status, '0');
+    bad_pos = strrchr(status, ':');
+    if (!zero_pos || !bad_pos) {
+        /*malformed status*/
+        return _gf_false;
+    }
+
+    if (zero_pos > bad_pos) {
+        return _gf_true;
+    }
+
+    return _gf_false;
+}
+
 int
 ec_shd_selfheal(struct subvol_healer *healer, int child, loc_t *loc,
                 gf_boolean_t full)
 {
+    dict_t *xdata = NULL;
+    dict_t *dict = NULL;
+    uint32_t count;
     int32_t ret;
+    char *heal_status = NULL;
+    ec_t *ec = healer->this->private;
+
+    GF_ATOMIC_INC(ec->stats.shd.attempted);
+    ret = syncop_getxattr(healer->this, loc, &dict, EC_XATTR_HEAL, NULL,
+                          &xdata);
+    if (ret == 0) {
+        if (dict && (dict_get_str(dict, EC_XATTR_HEAL, &heal_status) == 0)) {
+            if (ec_is_heal_completed(heal_status)) {
+                GF_ATOMIC_INC(ec->stats.shd.completed);
+            }
+        }
+    }
 
-    ret = syncop_getxattr(healer->this, loc, NULL, EC_XATTR_HEAL, NULL, NULL);
-    if (!full && (ret >= 0) && (loc->inode->ia_type == IA_IFDIR)) {
+    if (!full && (loc->inode->ia_type == IA_IFDIR)) {
         /* If we have just healed a directory, it's possible that
-         * other index entries have appeared to be healed. We put a
-         * mark so that we can check it later and restart a scan
-         * without delay. */
-        healer->rerun = _gf_true;
+         * other index entries have appeared to be healed. */
+        if ((xdata != NULL) &&
+            (dict_get_uint32(xdata, EC_XATTR_HEAL_NEW, &count) == 0) &&
+            (count > 0)) {
+            /* Force a rerun of the index healer. */
+            gf_msg_debug(healer->this->name, 0, "%d more entries to heal",
+                         count);
+
+            healer->rerun = _gf_true;
+        }
+    }
+
+    if (xdata != NULL) {
+        dict_unref(xdata);
+    }
+
+    if (dict) {
+        dict_unref(dict);
     }
 
     return ret;
@@ -241,9 +304,11 @@ ec_shd_index_sweep(struct subvol_healer *healer)
         goto out;
     }
 
+    _mask_cancellation();
     ret = syncop_mt_dir_scan(NULL, subvol, &loc, GF_CLIENT_PID_SELF_HEALD,
                              healer, ec_shd_index_heal, xdata,
                              ec->shd.max_threads, ec->shd.wait_qlength);
+    _unmask_cancellation();
 out:
     if (xdata)
         dict_unref(xdata);
@@ -263,6 +328,11 @@ ec_shd_full_heal(xlator_t *subvol, gf_dirent_t *entry, loc_t *parent,
     int ret = 0;
 
     ec = this->private;
+
+    if (this->cleanup_starting) {
+        return -ENOTCONN;
+    }
+
     if (ec->xl_up_count <= ec->fragments) {
         return -ENOTCONN;
     }
@@ -305,11 +375,15 @@ ec_shd_full_sweep(struct subvol_healer *healer, inode_t *inode)
 {
     ec_t *ec = NULL;
     loc_t loc = {0};
+    int ret = -1;
 
     ec = healer->this->private;
     loc.inode = inode;
-    return syncop_ftw(ec->xl_list[healer->subvol], &loc,
-                      GF_CLIENT_PID_SELF_HEALD, healer, ec_shd_full_heal);
+    _mask_cancellation();
+    ret = syncop_ftw(ec->xl_list[healer->subvol], &loc,
+                     GF_CLIENT_PID_SELF_HEALD, healer, ec_shd_full_heal);
+    _unmask_cancellation();
+    return ret;
 }
 
 void *
@@ -317,13 +391,16 @@ ec_shd_index_healer(void *data)
 {
     struct subvol_healer *healer = NULL;
     xlator_t *this = NULL;
+    int run = 0;
 
     healer = data;
     THIS = this = healer->this;
     ec_t *ec = this->private;
 
     for (;;) {
-        ec_shd_healer_wait(healer);
+        run = ec_shd_healer_wait(healer);
+        if (run == -1)
+            break;
 
         if (ec->xl_up_count > ec->fragments) {
             gf_msg_debug(this->name, 0, "starting index sweep on subvol %s",
@@ -352,16 +429,12 @@ ec_shd_full_healer(void *data)
 
     rootloc.inode = this->itable->root;
     for (;;) {
-        pthread_mutex_lock(&healer->mutex);
-        {
-            run = __ec_shd_healer_wait(healer);
-            if (!run)
-                healer->running = _gf_false;
-        }
-        pthread_mutex_unlock(&healer->mutex);
-
-        if (!run)
+        run = ec_shd_healer_wait(healer);
+        if (run < 0) {
             break;
+        } else if (run == 0) {
+            continue;
+        }
 
         if (ec->xl_up_count > ec->fragments) {
             gf_msg(this->name, GF_LOG_INFO, 0, EC_MSG_FULL_SWEEP_START,
@@ -429,6 +502,9 @@ unlock:
 int
 ec_shd_full_healer_spawn(xlator_t *this, int subvol)
 {
+    if (xlator_is_cleanup_starting(this))
+        return -1;
+
     return ec_shd_healer_spawn(this, NTH_FULL_HEALER(this, subvol),
                                ec_shd_full_healer);
 }
@@ -436,6 +512,9 @@ ec_shd_full_healer_spawn(xlator_t *this, int subvol)
 int
 ec_shd_index_healer_spawn(xlator_t *this, int subvol)
 {
+    if (xlator_is_cleanup_starting(this))
+        return -1;
+
     return ec_shd_healer_spawn(this, NTH_INDEX_HEALER(this, subvol),
                                ec_shd_index_healer);
 }
@@ -562,3 +641,41 @@ out:
     dict_del(output, this->name);
     return ret;
 }
+
+void
+ec_destroy_healer_object(xlator_t *this, struct subvol_healer *healer)
+{
+    if (!healer)
+        return;
+
+    pthread_cond_destroy(&healer->cond);
+    pthread_mutex_destroy(&healer->mutex);
+}
+
+void
+ec_selfheal_daemon_fini(xlator_t *this)
+{
+    struct subvol_healer *healer = NULL;
+    ec_self_heald_t *shd = NULL;
+    ec_t *priv = NULL;
+    int i = 0;
+
+    priv = this->private;
+    if (!priv)
+        return;
+
+    shd = &priv->shd;
+    if (!shd->iamshd)
+        return;
+
+    for (i = 0; i < priv->nodes; i++) {
+        healer = &shd->index_healers[i];
+        ec_destroy_healer_object(this, healer);
+
+        healer = &shd->full_healers[i];
+        ec_destroy_healer_object(this, healer);
+    }
+
+    GF_FREE(shd->index_healers);
+    GF_FREE(shd->full_healers);
+}
diff --git a/xlators/cluster/ec/src/ec-heald.h b/xlators/cluster/ec/src/ec-heald.h
index 4d141d767e5..6c7da4edc10 100644
--- a/xlators/cluster/ec/src/ec-heald.h
+++ b/xlators/cluster/ec/src/ec-heald.h
@@ -11,9 +11,9 @@
 #ifndef __EC_HEALD_H__
 #define __EC_HEALD_H__
 
-#include "xlator.h"
-
-#include "ec-types.h"
+#include "ec-types.h"           // for ec_t
+#include "glusterfs/dict.h"     // for dict_t
+#include "glusterfs/globals.h"  // for xlator_t
 
 int
 ec_xl_op(xlator_t *this, dict_t *input, dict_t *output);
@@ -24,4 +24,7 @@ ec_selfheal_daemon_init(xlator_t *this);
 void
 ec_shd_index_healer_wake(ec_t *ec);
 
+void
+ec_selfheal_daemon_fini(xlator_t *this);
+
 #endif /* __EC_HEALD_H__ */
diff --git a/xlators/cluster/ec/src/ec-helpers.c b/xlators/cluster/ec/src/ec-helpers.c
index dec39b9d2aa..48f54475e01 100644
--- a/xlators/cluster/ec/src/ec-helpers.c
+++ b/xlators/cluster/ec/src/ec-helpers.c
@@ -10,7 +10,7 @@
 
 #include <libgen.h>
 
-#include "byte-order.h"
+#include <glusterfs/byte-order.h>
 
 #include "ec.h"
 #include "ec-mem-types.h"
@@ -476,7 +476,7 @@ out:
 int32_t
 ec_loc_setup_path(xlator_t *xl, loc_t *loc)
 {
-    uuid_t root = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1};
+    static uuid_t root = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1};
     char *name;
     int32_t ret = -EINVAL;
 
@@ -717,6 +717,7 @@ __ec_inode_get(inode_t *inode, xlator_t *xl)
             memset(ctx, 0, sizeof(*ctx));
             INIT_LIST_HEAD(&ctx->heal);
             INIT_LIST_HEAD(&ctx->stripe_cache.lru);
+            ctx->heal_count = 0;
             value = (uint64_t)(uintptr_t)ctx;
             if (__inode_ctx_set(inode, xl, &value) != 0) {
                 GF_FREE(ctx);
@@ -752,6 +753,7 @@ __ec_fd_get(fd_t *fd, xlator_t *xl)
 {
     int i = 0;
     ec_fd_t *ctx = NULL;
+    ec_inode_t *ictx = NULL;
     uint64_t value = 0;
     ec_t *ec = xl->private;
 
@@ -774,6 +776,12 @@ __ec_fd_get(fd_t *fd, xlator_t *xl)
                 GF_FREE(ctx);
                 return NULL;
             }
+            /* Only refering bad-version so no need for lock
+             * */
+            ictx = __ec_inode_get(fd->inode, xl);
+            if (ictx) {
+                ctx->bad_version = ictx->bad_version;
+            }
         }
     } else {
         ctx = (ec_fd_t *)(uintptr_t)value;
diff --git a/xlators/cluster/ec/src/ec-inode-read.c b/xlators/cluster/ec/src/ec-inode-read.c
index c02d5401439..dad5f4d7018 100644
--- a/xlators/cluster/ec/src/ec-inode-read.c
+++ b/xlators/cluster/ec/src/ec-inode-read.c
@@ -8,9 +8,6 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-
 #include "ec.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
@@ -135,7 +132,7 @@ ec_manager_access(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_access(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_access_cbk_t func, void *data, loc_t *loc,
+          uint32_t fop_flags, fop_access_cbk_t func, void *data, loc_t *loc,
           int32_t mask, dict_t *xdata)
 {
     ec_cbk_t callback = {.access = func};
@@ -149,7 +146,7 @@ ec_access(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_ACCESS, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_access,
+                               target, fop_flags, ec_wind_access,
                                ec_manager_access, callback, data);
     if (fop == NULL) {
         goto out;
@@ -393,15 +390,34 @@ ec_manager_getxattr(ec_fop_data_t *fop, int32_t state)
 int32_t
 ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
                      int32_t op_ret, int32_t op_errno, uintptr_t mask,
-                     uintptr_t good, uintptr_t bad, dict_t *xdata)
+                     uintptr_t good, uintptr_t bad, uint32_t pending,
+                     dict_t *xdata)
 {
-    ec_fop_data_t *fop = cookie;
-    fop_getxattr_cbk_t func = fop->data;
+    fop_getxattr_cbk_t func = cookie;
     ec_t *ec = xl->private;
     dict_t *dict = NULL;
     char *str;
     char bin1[65], bin2[65];
 
+    /* We try to return the 'pending' information in xdata, but if this cannot
+     * be set, we will ignore it silently. We prefer to report the success or
+     * failure of the heal itself. */
+    if (xdata == NULL) {
+        xdata = dict_new();
+    } else {
+        dict_ref(xdata);
+    }
+    if (xdata != NULL) {
+        if (dict_set_uint32(xdata, EC_XATTR_HEAL_NEW, pending) != 0) {
+            /* dict_set_uint32() is marked as 'warn_unused_result' and gcc
+             * enforces to check the result in this case. However we don't
+             * really care if it succeeded or not. We'll just do the same.
+             *
+             * This empty 'if' avoids the warning, and it will be removed by
+             * the optimizer. */
+        }
+    }
+
     if (op_ret >= 0) {
         dict = dict_new();
         if (dict == NULL) {
@@ -435,18 +451,21 @@ ec_getxattr_heal_cbk(call_frame_t *frame, void *cookie, xlator_t *xl,
     }
 
 out:
-    func(frame, NULL, xl, op_ret, op_errno, dict, NULL);
+    func(frame, NULL, xl, op_ret, op_errno, dict, xdata);
 
     if (dict != NULL) {
         dict_unref(dict);
     }
+    if (xdata != NULL) {
+        dict_unref(xdata);
+    }
 
     return 0;
 }
 
 void
 ec_getxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_getxattr_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_getxattr_cbk_t func, void *data, loc_t *loc,
             const char *name, dict_t *xdata)
 {
     ec_cbk_t callback = {.getxattr = func};
@@ -468,7 +487,7 @@ ec_getxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     }
 
     fop = ec_fop_data_allocate(
-        frame, this, GF_FOP_GETXATTR, EC_FLAG_LOCK_SHARED, target, minimum,
+        frame, this, GF_FOP_GETXATTR, EC_FLAG_LOCK_SHARED, target, fop_flags,
         ec_wind_getxattr, ec_manager_getxattr, callback, data);
     if (fop == NULL) {
         goto out;
@@ -588,7 +607,7 @@ ec_wind_fgetxattr(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fgetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fgetxattr_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fgetxattr_cbk_t func, void *data, fd_t *fd,
              const char *name, dict_t *xdata)
 {
     ec_cbk_t callback = {.fgetxattr = func};
@@ -602,7 +621,7 @@ ec_fgetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(
-        frame, this, GF_FOP_FGETXATTR, EC_FLAG_LOCK_SHARED, target, minimum,
+        frame, this, GF_FOP_FGETXATTR, EC_FLAG_LOCK_SHARED, target, fop_flags,
         ec_wind_fgetxattr, ec_manager_getxattr, callback, data);
     if (fop == NULL) {
         goto out;
@@ -774,13 +793,15 @@ ec_manager_open(ec_fop_data_t *fop, int32_t state)
 
                 return EC_STATE_REPORT;
             }
-            err = ec_loc_from_loc(fop->xl, &ctx->loc, &fop->loc[0]);
-            if (err != 0) {
-                UNLOCK(&fop->fd->lock);
+            if (!ctx->loc.inode) {
+                err = ec_loc_from_loc(fop->xl, &ctx->loc, &fop->loc[0]);
+                if (err != 0) {
+                    UNLOCK(&fop->fd->lock);
 
-                fop->error = -err;
+                    fop->error = -err;
 
-                return EC_STATE_REPORT;
+                    return EC_STATE_REPORT;
+                }
             }
 
             ctx->flags = fop->int32;
@@ -869,9 +890,9 @@ ec_manager_open(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_open(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_open_cbk_t func, void *data, loc_t *loc, int32_t flags, fd_t *fd,
-        dict_t *xdata)
+ec_open(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_open_cbk_t func, void *data, loc_t *loc,
+        int32_t flags, fd_t *fd, dict_t *xdata)
 {
     ec_cbk_t callback = {.open = func};
     ec_fop_data_t *fop = NULL;
@@ -884,7 +905,7 @@ ec_open(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_OPEN, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_open, ec_manager_open,
+                               target, fop_flags, ec_wind_open, ec_manager_open,
                                callback, data);
     if (fop == NULL) {
         goto out;
@@ -1071,7 +1092,7 @@ ec_manager_readlink(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_readlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_readlink_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_readlink_cbk_t func, void *data, loc_t *loc,
             size_t size, dict_t *xdata)
 {
     ec_cbk_t callback = {.readlink = func};
@@ -1085,7 +1106,7 @@ ec_readlink(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(
-        frame, this, GF_FOP_READLINK, EC_FLAG_LOCK_SHARED, target, minimum,
+        frame, this, GF_FOP_READLINK, EC_FLAG_LOCK_SHARED, target, fop_flags,
         ec_wind_readlink, ec_manager_readlink, callback, data);
     if (fop == NULL) {
         goto out;
@@ -1331,6 +1352,7 @@ int32_t
 ec_manager_readv(ec_fop_data_t *fop, int32_t state)
 {
     ec_cbk_data_t *cbk;
+    ec_t *ec = fop->xl->private;
 
     switch (state) {
         case EC_STATE_INIT:
@@ -1350,6 +1372,9 @@ ec_manager_readv(ec_fop_data_t *fop, int32_t state)
             return EC_STATE_DISPATCH;
 
         case EC_STATE_DISPATCH:
+            if (ec->read_mask) {
+                fop->mask &= ec->read_mask;
+            }
             ec_dispatch_min(fop);
 
             return EC_STATE_PREPARE_ANSWER;
@@ -1417,9 +1442,9 @@ ec_manager_readv(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_readv(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_readv_cbk_t func, void *data, fd_t *fd, size_t size, off_t offset,
-         uint32_t flags, dict_t *xdata)
+ec_readv(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_readv_cbk_t func, void *data, fd_t *fd,
+         size_t size, off_t offset, uint32_t flags, dict_t *xdata)
 {
     ec_cbk_t callback = {.readv = func};
     ec_fop_data_t *fop = NULL;
@@ -1432,8 +1457,8 @@ ec_readv(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_READ, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_readv, ec_manager_readv,
-                               callback, data);
+                               target, fop_flags, ec_wind_readv,
+                               ec_manager_readv, callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1637,9 +1662,9 @@ ec_manager_seek(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_seek(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_seek_cbk_t func, void *data, fd_t *fd, off_t offset,
-        gf_seek_what_t what, dict_t *xdata)
+ec_seek(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_seek_cbk_t func, void *data, fd_t *fd,
+        off_t offset, gf_seek_what_t what, dict_t *xdata)
 {
     ec_cbk_t callback = {.seek = func};
     ec_fop_data_t *fop = NULL;
@@ -1652,7 +1677,7 @@ ec_seek(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_SEEK, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_seek, ec_manager_seek,
+                               target, fop_flags, ec_wind_seek, ec_manager_seek,
                                callback, data);
     if (fop == NULL) {
         goto out;
@@ -1855,8 +1880,9 @@ ec_manager_stat(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_stat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-        fop_stat_cbk_t func, void *data, loc_t *loc, dict_t *xdata)
+ec_stat(call_frame_t *frame, xlator_t *this, uintptr_t target,
+        uint32_t fop_flags, fop_stat_cbk_t func, void *data, loc_t *loc,
+        dict_t *xdata)
 {
     ec_cbk_t callback = {.stat = func};
     ec_fop_data_t *fop = NULL;
@@ -1869,7 +1895,7 @@ ec_stat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_STAT, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_stat, ec_manager_stat,
+                               target, fop_flags, ec_wind_stat, ec_manager_stat,
                                callback, data);
     if (fop == NULL) {
         goto out;
@@ -1965,8 +1991,9 @@ ec_wind_fstat(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 }
 
 void
-ec_fstat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
-         fop_fstat_cbk_t func, void *data, fd_t *fd, dict_t *xdata)
+ec_fstat(call_frame_t *frame, xlator_t *this, uintptr_t target,
+         uint32_t fop_flags, fop_fstat_cbk_t func, void *data, fd_t *fd,
+         dict_t *xdata)
 {
     ec_cbk_t callback = {.fstat = func};
     ec_fop_data_t *fop = NULL;
@@ -1979,8 +2006,8 @@ ec_fstat(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_FSTAT, EC_FLAG_LOCK_SHARED,
-                               target, minimum, ec_wind_fstat, ec_manager_stat,
-                               callback, data);
+                               target, fop_flags, ec_wind_fstat,
+                               ec_manager_stat, callback, data);
     if (fop == NULL) {
         goto out;
     }
diff --git a/xlators/cluster/ec/src/ec-inode-write.c b/xlators/cluster/ec/src/ec-inode-write.c
index b915a992a00..9b5fe2a7fdc 100644
--- a/xlators/cluster/ec/src/ec-inode-write.c
+++ b/xlators/cluster/ec/src/ec-inode-write.c
@@ -8,10 +8,6 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-
-#include "ec.h"
 #include "ec-messages.h"
 #include "ec-helpers.h"
 #include "ec-common.h"
@@ -89,6 +85,8 @@ ec_update_write(ec_fop_data_t *fop, uintptr_t mask, off_t offset, uint64_t size)
         goto out;
     }
 
+    if (fop->locks[0].lock)
+        ec_lock_update_good(fop->locks[0].lock, fop);
     vector.iov_base = iobuf->ptr;
     vector.iov_len = size;
     memset(vector.iov_base, 0, vector.iov_len);
@@ -183,26 +181,26 @@ ec_xattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
     switch (fop->id) {
         case GF_FOP_SETXATTR:
             if (fop->cbks.setxattr) {
-                fop->cbks.setxattr(frame, cookie, this, op_ret, op_errno,
-                                   xdata);
+                QUORUM_CBK(fop->cbks.setxattr, fop, frame, cookie, this, op_ret,
+                           op_errno, xdata);
             }
             break;
         case GF_FOP_REMOVEXATTR:
             if (fop->cbks.removexattr) {
-                fop->cbks.removexattr(frame, cookie, this, op_ret, op_errno,
-                                      xdata);
+                QUORUM_CBK(fop->cbks.removexattr, fop, frame, cookie, this,
+                           op_ret, op_errno, xdata);
             }
             break;
         case GF_FOP_FSETXATTR:
             if (fop->cbks.fsetxattr) {
-                fop->cbks.fsetxattr(frame, cookie, this, op_ret, op_errno,
-                                    xdata);
+                QUORUM_CBK(fop->cbks.fsetxattr, fop, frame, cookie, this,
+                           op_ret, op_errno, xdata);
             }
             break;
         case GF_FOP_FREMOVEXATTR:
             if (fop->cbks.fremovexattr) {
-                fop->cbks.fremovexattr(frame, cookie, this, op_ret, op_errno,
-                                       xdata);
+                QUORUM_CBK(fop->cbks.fremovexattr, fop, frame, cookie, this,
+                           op_ret, op_errno, xdata);
             }
             break;
     }
@@ -281,7 +279,7 @@ ec_manager_xattr(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_removexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-               int32_t minimum, fop_removexattr_cbk_t func, void *data,
+               uint32_t fop_flags, fop_removexattr_cbk_t func, void *data,
                loc_t *loc, const char *name, dict_t *xdata)
 {
     ec_cbk_t callback = {.removexattr = func};
@@ -295,7 +293,7 @@ ec_removexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_REMOVEXATTR, 0, target,
-                               minimum, ec_wind_removexattr, ec_manager_xattr,
+                               fop_flags, ec_wind_removexattr, ec_manager_xattr,
                                callback, data);
     if (fop == NULL) {
         goto out;
@@ -361,7 +359,7 @@ ec_wind_fremovexattr(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fremovexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-                int32_t minimum, fop_fremovexattr_cbk_t func, void *data,
+                uint32_t fop_flags, fop_fremovexattr_cbk_t func, void *data,
                 fd_t *fd, const char *name, dict_t *xdata)
 {
     ec_cbk_t callback = {.fremovexattr = func};
@@ -375,8 +373,8 @@ ec_fremovexattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_FREMOVEXATTR, 0, target,
-                               minimum, ec_wind_fremovexattr, ec_manager_xattr,
-                               callback, data);
+                               fop_flags, ec_wind_fremovexattr,
+                               ec_manager_xattr, callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -492,16 +490,15 @@ ec_manager_setattr(ec_fop_data_t *fop, int32_t state)
 
             if (fop->id == GF_FOP_SETATTR) {
                 if (fop->cbks.setattr != NULL) {
-                    fop->cbks.setattr(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                      cbk->op_errno, &cbk->iatt[0],
-                                      &cbk->iatt[1], cbk->xdata);
+                    QUORUM_CBK(fop->cbks.setattr, fop, fop->req_frame, fop,
+                               fop->xl, cbk->op_ret, cbk->op_errno,
+                               &cbk->iatt[0], &cbk->iatt[1], cbk->xdata);
                 }
             } else {
                 if (fop->cbks.fsetattr != NULL) {
-                    fop->cbks.fsetattr(fop->req_frame, fop, fop->xl,
-                                       cbk->op_ret, cbk->op_errno,
-                                       &cbk->iatt[0], &cbk->iatt[1],
-                                       cbk->xdata);
+                    QUORUM_CBK(fop->cbks.fsetattr, fop, fop->req_frame, fop,
+                               fop->xl, cbk->op_ret, cbk->op_errno,
+                               &cbk->iatt[0], &cbk->iatt[1], cbk->xdata);
                 }
             }
 
@@ -550,7 +547,7 @@ ec_manager_setattr(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_setattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_setattr_cbk_t func, void *data, loc_t *loc,
+           uint32_t fop_flags, fop_setattr_cbk_t func, void *data, loc_t *loc,
            struct iatt *stbuf, int32_t valid, dict_t *xdata)
 {
     ec_cbk_t callback = {.setattr = func};
@@ -563,9 +560,9 @@ ec_setattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_SETATTR, 0, target, minimum,
-                               ec_wind_setattr, ec_manager_setattr, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_SETATTR, 0, target,
+                               fop_flags, ec_wind_setattr, ec_manager_setattr,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -627,7 +624,7 @@ ec_wind_fsetattr(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fsetattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fsetattr_cbk_t func, void *data, fd_t *fd,
+            uint32_t fop_flags, fop_fsetattr_cbk_t func, void *data, fd_t *fd,
             struct iatt *stbuf, int32_t valid, dict_t *xdata)
 {
     ec_cbk_t callback = {.fsetattr = func};
@@ -640,9 +637,9 @@ ec_fsetattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSETATTR, 0, target, minimum,
-                               ec_wind_fsetattr, ec_manager_setattr, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FSETATTR, 0, target,
+                               fop_flags, ec_wind_fsetattr, ec_manager_setattr,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -707,7 +704,7 @@ ec_wind_setxattr(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_setxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_setxattr_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_setxattr_cbk_t func, void *data, loc_t *loc,
             dict_t *dict, int32_t flags, dict_t *xdata)
 {
     ec_cbk_t callback = {.setxattr = func};
@@ -720,9 +717,9 @@ ec_setxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_SETXATTR, 0, target, minimum,
-                               ec_wind_setxattr, ec_manager_xattr, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_SETXATTR, 0, target,
+                               fop_flags, ec_wind_setxattr, ec_manager_xattr,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -825,7 +822,7 @@ ec_wind_fsetxattr(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fsetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fsetxattr_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fsetxattr_cbk_t func, void *data, fd_t *fd,
              dict_t *dict, int32_t flags, dict_t *xdata)
 {
     ec_cbk_t callback = {.fsetxattr = func};
@@ -839,7 +836,7 @@ ec_fsetxattr(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_FSETXATTR, 0, target,
-                               minimum, ec_wind_fsetxattr, ec_manager_xattr,
+                               fop_flags, ec_wind_fsetxattr, ec_manager_xattr,
                                callback, data);
     if (fop == NULL) {
         goto out;
@@ -992,9 +989,9 @@ ec_manager_fallocate(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.fallocate != NULL) {
-                fop->cbks.fallocate(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                    cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                    cbk->xdata);
+                QUORUM_CBK(fop->cbks.fallocate, fop, fop->req_frame, fop,
+                           fop->xl, cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -1035,7 +1032,7 @@ ec_manager_fallocate(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_fallocate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_fallocate_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_fallocate_cbk_t func, void *data, fd_t *fd,
              int32_t mode, off_t offset, size_t len, dict_t *xdata)
 {
     ec_cbk_t callback = {.fallocate = func};
@@ -1049,8 +1046,8 @@ ec_fallocate(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_FALLOCATE, 0, target,
-                               minimum, ec_wind_fallocate, ec_manager_fallocate,
-                               callback, data);
+                               fop_flags, ec_wind_fallocate,
+                               ec_manager_fallocate, callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1209,8 +1206,8 @@ ec_manager_discard(ec_fop_data_t *fop, int32_t state)
                 ec_dispatch_all(fop);
                 return EC_STATE_DELAYED_START;
             } else {
-                /*Assume discard to have succeeded on mask*/
-                fop->good = fop->mask;
+                /* Assume discard to have succeeded on all bricks */
+                ec_succeed_all(fop);
             }
 
             /* Fall through */
@@ -1245,9 +1242,9 @@ ec_manager_discard(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.discard != NULL) {
-                fop->cbks.discard(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                  cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                  cbk->xdata);
+                QUORUM_CBK(fop->cbks.discard, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -1289,7 +1286,7 @@ ec_manager_discard(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_discard(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_discard_cbk_t func, void *data, fd_t *fd,
+           uint32_t fop_flags, fop_discard_cbk_t func, void *data, fd_t *fd,
            off_t offset, size_t len, dict_t *xdata)
 {
     ec_cbk_t callback = {.discard = func};
@@ -1302,9 +1299,9 @@ ec_discard(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_DISCARD, 0, target, minimum,
-                               ec_wind_discard, ec_manager_discard, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_DISCARD, 0, target,
+                               fop_flags, ec_wind_discard, ec_manager_discard,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1405,6 +1402,7 @@ int32_t
 ec_manager_truncate(ec_fop_data_t *fop, int32_t state)
 {
     ec_cbk_data_t *cbk;
+    off_t offset_down;
 
     switch (state) {
         case EC_STATE_INIT:
@@ -1416,16 +1414,19 @@ ec_manager_truncate(ec_fop_data_t *fop, int32_t state)
             /* Fall through */
 
         case EC_STATE_LOCK:
+            offset_down = fop->user_size;
+            ec_adjust_offset_down(fop->xl->private, &offset_down, _gf_true);
+
             if (fop->id == GF_FOP_TRUNCATE) {
                 ec_lock_prepare_inode(
                     fop, &fop->loc[0],
                     EC_UPDATE_DATA | EC_UPDATE_META | EC_QUERY_INFO,
-                    fop->offset, EC_RANGE_FULL);
+                    offset_down, EC_RANGE_FULL);
             } else {
                 ec_lock_prepare_fd(
                     fop, fop->fd,
                     EC_UPDATE_DATA | EC_UPDATE_META | EC_QUERY_INFO,
-                    fop->offset, EC_RANGE_FULL);
+                    offset_down, EC_RANGE_FULL);
             }
             ec_lock(fop);
 
@@ -1471,17 +1472,15 @@ ec_manager_truncate(ec_fop_data_t *fop, int32_t state)
 
             if (fop->id == GF_FOP_TRUNCATE) {
                 if (fop->cbks.truncate != NULL) {
-                    fop->cbks.truncate(fop->req_frame, fop, fop->xl,
-                                       cbk->op_ret, cbk->op_errno,
-                                       &cbk->iatt[0], &cbk->iatt[1],
-                                       cbk->xdata);
+                    QUORUM_CBK(fop->cbks.truncate, fop, fop->req_frame, fop,
+                               fop->xl, cbk->op_ret, cbk->op_errno,
+                               &cbk->iatt[0], &cbk->iatt[1], cbk->xdata);
                 }
             } else {
                 if (fop->cbks.ftruncate != NULL) {
-                    fop->cbks.ftruncate(fop->req_frame, fop, fop->xl,
-                                        cbk->op_ret, cbk->op_errno,
-                                        &cbk->iatt[0], &cbk->iatt[1],
-                                        cbk->xdata);
+                    QUORUM_CBK(fop->cbks.ftruncate, fop, fop->req_frame, fop,
+                               fop->xl, cbk->op_ret, cbk->op_errno,
+                               &cbk->iatt[0], &cbk->iatt[1], cbk->xdata);
                 }
             }
 
@@ -1530,7 +1529,7 @@ ec_manager_truncate(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_truncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_truncate_cbk_t func, void *data, loc_t *loc,
+            uint32_t fop_flags, fop_truncate_cbk_t func, void *data, loc_t *loc,
             off_t offset, dict_t *xdata)
 {
     ec_cbk_t callback = {.truncate = func};
@@ -1543,9 +1542,9 @@ ec_truncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_TRUNCATE, 0, target, minimum,
-                               ec_wind_truncate, ec_manager_truncate, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_TRUNCATE, 0, target,
+                               fop_flags, ec_wind_truncate, ec_manager_truncate,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1604,7 +1603,7 @@ ec_wind_ftruncate(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_ftruncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
-             int32_t minimum, fop_ftruncate_cbk_t func, void *data, fd_t *fd,
+             uint32_t fop_flags, fop_ftruncate_cbk_t func, void *data, fd_t *fd,
              off_t offset, dict_t *xdata)
 {
     ec_cbk_t callback = {.ftruncate = func};
@@ -1618,8 +1617,8 @@ ec_ftruncate(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
     fop = ec_fop_data_allocate(frame, this, GF_FOP_FTRUNCATE, 0, target,
-                               minimum, ec_wind_ftruncate, ec_manager_truncate,
-                               callback, data);
+                               fop_flags, ec_wind_ftruncate,
+                               ec_manager_truncate, callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1973,6 +1972,23 @@ ec_get_and_merge_stripe(ec_t *ec, ec_fop_data_t *fop, ec_stripe_part_t which)
     return found;
 }
 
+static uintptr_t
+ec_get_lock_good_mask(inode_t *inode, xlator_t *xl)
+{
+    ec_lock_t *lock = NULL;
+    ec_inode_t *ictx = NULL;
+    LOCK(&inode->lock);
+    {
+        ictx = __ec_inode_get(inode, xl);
+        if (ictx)
+            lock = ictx->inode_lock;
+    }
+    UNLOCK(&inode->lock);
+    if (lock)
+        return lock->good_mask;
+    return 0;
+}
+
 void
 ec_writev_start(ec_fop_data_t *fop)
 {
@@ -2009,20 +2025,29 @@ ec_writev_start(ec_fop_data_t *fop)
     if (err != 0) {
         goto failed_fd;
     }
+    tail = fop->size - fop->user_size - fop->head;
     if (fop->head > 0) {
-        found_stripe = ec_get_and_merge_stripe(ec, fop, EC_STRIPE_HEAD);
-        if (!found_stripe) {
-            if (ec_make_internal_fop_xdata(&xdata)) {
-                err = -ENOMEM;
-                goto failed_xdata;
+        if (current > fop->offset) {
+            found_stripe = ec_get_and_merge_stripe(ec, fop, EC_STRIPE_HEAD);
+            if (!found_stripe) {
+                if (ec_make_internal_fop_xdata(&xdata)) {
+                    err = -ENOMEM;
+                    goto failed_xdata;
+                }
+                ec_readv(fop->frame, fop->xl,
+                         ec_get_lock_good_mask(fop->fd->inode, fop->xl),
+                         EC_MINIMUM_MIN, ec_writev_merge_head, NULL, fd,
+                         ec->stripe_size, fop->offset, 0, xdata);
+            }
+        } else {
+            memset(fop->vector[0].iov_base, 0, fop->head);
+            memset(fop->vector[0].iov_base + fop->size - tail, 0, tail);
+            if (ec->stripe_cache && (fop->size <= ec->stripe_size)) {
+                ec_add_stripe_in_cache(ec, fop);
             }
-            ec_readv(fop->frame, fop->xl, -1, EC_MINIMUM_MIN,
-                     ec_writev_merge_head, NULL, fd, ec->stripe_size,
-                     fop->offset, 0, xdata);
         }
     }
 
-    tail = fop->size - fop->user_size - fop->head;
     if ((tail > 0) && ((fop->head == 0) || (fop->size > ec->stripe_size))) {
         /* Current locking scheme will make sure the 'current' below will
          * never decrease while the fop is in progress, so the checks will
@@ -2035,8 +2060,10 @@ ec_writev_start(ec_fop_data_t *fop)
                     err = -ENOMEM;
                     goto failed_xdata;
                 }
-                ec_readv(fop->frame, fop->xl, -1, EC_MINIMUM_MIN,
-                         ec_writev_merge_tail, NULL, fd, ec->stripe_size,
+                ec_readv(fop->frame, fop->xl,
+                         ec_get_lock_good_mask(fop->fd->inode, fop->xl),
+                         EC_MINIMUM_MIN, ec_writev_merge_tail, NULL, fd,
+                         ec->stripe_size,
                          fop->offset + fop->size - ec->stripe_size, 0, xdata);
             }
         } else {
@@ -2211,9 +2238,9 @@ ec_manager_writev(ec_fop_data_t *fop, int32_t state)
             GF_ASSERT(cbk != NULL);
 
             if (fop->cbks.writev != NULL) {
-                fop->cbks.writev(fop->req_frame, fop, fop->xl, cbk->op_ret,
-                                 cbk->op_errno, &cbk->iatt[0], &cbk->iatt[1],
-                                 cbk->xdata);
+                QUORUM_CBK(fop->cbks.writev, fop, fop->req_frame, fop, fop->xl,
+                           cbk->op_ret, cbk->op_errno, &cbk->iatt[0],
+                           &cbk->iatt[1], cbk->xdata);
             }
 
             return EC_STATE_LOCK_REUSE;
@@ -2262,7 +2289,7 @@ ec_manager_writev(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_writev(call_frame_t *frame, xlator_t *this, uintptr_t target,
-          int32_t minimum, fop_writev_cbk_t func, void *data, fd_t *fd,
+          uint32_t fop_flags, fop_writev_cbk_t func, void *data, fd_t *fd,
           struct iovec *vector, int32_t count, off_t offset, uint32_t flags,
           struct iobref *iobref, dict_t *xdata)
 {
@@ -2276,7 +2303,7 @@ ec_writev(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_WRITE, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_WRITE, 0, target, fop_flags,
                                ec_wind_writev, ec_manager_writev, callback,
                                data);
     if (fop == NULL) {
diff --git a/xlators/cluster/ec/src/ec-locks.c b/xlators/cluster/ec/src/ec-locks.c
index 58c9acb6ded..601960d6154 100644
--- a/xlators/cluster/ec/src/ec-locks.c
+++ b/xlators/cluster/ec/src/ec-locks.c
@@ -8,13 +8,9 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
-
 #include "ec-helpers.h"
 #include "ec-common.h"
 #include "ec-combine.h"
-#include "ec-method.h"
 #include "ec-fops.h"
 #include "ec-messages.h"
 
@@ -28,9 +24,36 @@ ec_lock_check(ec_fop_data_t *fop, uintptr_t *mask)
     ec_t *ec = fop->xl->private;
     ec_cbk_data_t *ans = NULL;
     ec_cbk_data_t *cbk = NULL;
-    uintptr_t locked = 0, notlocked = 0;
+    uintptr_t locked = 0;
+    int32_t good = 0;
+    int32_t eagain = 0;
+    int32_t estale = 0;
     int32_t error = -1;
 
+    /* There are some errors that we'll handle in an special way while trying
+     * to acquire a lock.
+     *
+     *   EAGAIN:  If it's found during a parallel non-blocking lock request, we
+     *            consider that there's contention on the inode, so we consider
+     *            the acquisition a failure and try again with a sequential
+     *            blocking lock request. This will ensure that we get a lock on
+     *            as many bricks as possible (ignoring EAGAIN here would cause
+     *            unnecessary triggers of self-healing).
+     *
+     *            If it's found during a sequential blocking lock request, it's
+     *            considered an error. Lock will only succeed if there are
+     *            enough other bricks locked.
+     *
+     *   ESTALE:  This can appear during parallel or sequential lock request if
+     *            the inode has just been unlinked. We consider this error is
+     *            not recoverable, but we also don't consider it as fatal. So,
+     *            if it happens during parallel lock, we won't attempt a
+     *            sequential one unless there are EAGAIN errors on other
+     *            bricks (and are enough to form a quorum), but if we reach
+     *            quorum counting the ESTALE bricks, we consider the whole
+     *            result of the operation is ESTALE instead of EIO.
+     */
+
     list_for_each_entry(ans, &fop->cbk_list, list)
     {
         if (ans->op_ret >= 0) {
@@ -38,24 +61,23 @@ ec_lock_check(ec_fop_data_t *fop, uintptr_t *mask)
                 error = EIO;
             }
             locked |= ans->mask;
+            good = ans->count;
             cbk = ans;
-        } else {
-            if (ans->op_errno == EAGAIN) {
-                switch (fop->uint32) {
-                    case EC_LOCK_MODE_NONE:
-                    case EC_LOCK_MODE_ALL:
-                        /* Goal is to treat non-blocking lock as failure
-                         * even if there is a single EAGAIN*/
-                        notlocked |= ans->mask;
-                        break;
-                }
-            }
+        } else if (ans->op_errno == ESTALE) {
+            estale += ans->count;
+        } else if ((ans->op_errno == EAGAIN) &&
+                   (fop->uint32 != EC_LOCK_MODE_INC)) {
+            eagain += ans->count;
         }
     }
 
     if (error == -1) {
-        if (gf_bits_count(locked | notlocked) >= ec->fragments) {
-            if (notlocked == 0) {
+        /* If we have enough quorum with succeeded and EAGAIN answers, we
+         * ignore for now any ESTALE answer. If there are EAGAIN answers,
+         * we retry with a sequential blocking lock request if needed.
+         * Otherwise we succeed. */
+        if ((good + eagain) >= ec->fragments) {
+            if (eagain == 0) {
                 if (fop->answer == NULL) {
                     fop->answer = cbk;
                 }
@@ -68,21 +90,28 @@ ec_lock_check(ec_fop_data_t *fop, uintptr_t *mask)
                     case EC_LOCK_MODE_NONE:
                         error = EAGAIN;
                         break;
-
                     case EC_LOCK_MODE_ALL:
                         fop->uint32 = EC_LOCK_MODE_INC;
                         break;
-
                     default:
+                        /* This shouldn't happen because eagain cannot be > 0
+                         * when fop->uint32 is EC_LOCK_MODE_INC. */
                         error = EIO;
                         break;
                 }
             }
         } else {
-            if (fop->answer && fop->answer->op_ret < 0)
+            /* We have been unable to find enough candidates that will be able
+             * to take the lock. If we have quorum on some answer, we return
+             * it. Otherwise we check if ESTALE answers allow us to reach
+             * quorum. If so, we return ESTALE. */
+            if (fop->answer && fop->answer->op_ret < 0) {
                 error = fop->answer->op_errno;
-            else
+            } else if ((good + eagain + estale) >= ec->fragments) {
+                error = ESTALE;
+            } else {
                 error = EIO;
+            }
         }
     }
 
@@ -275,7 +304,7 @@ ec_manager_entrylk(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_entrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
-           int32_t minimum, fop_entrylk_cbk_t func, void *data,
+           uint32_t fop_flags, fop_entrylk_cbk_t func, void *data,
            const char *volume, loc_t *loc, const char *basename,
            entrylk_cmd cmd, entrylk_type type, dict_t *xdata)
 {
@@ -285,13 +314,12 @@ ec_entrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
 
     gf_msg_trace("ec", 0, "EC(ENTRYLK) %p", frame);
 
-    VALIDATE_OR_GOTO(this, out);
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_ENTRYLK, 0, target, minimum,
-                               ec_wind_entrylk, ec_manager_entrylk, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_ENTRYLK, 0, target,
+                               fop_flags, ec_wind_entrylk, ec_manager_entrylk,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -404,7 +432,7 @@ ec_wind_fentrylk(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_fentrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
-            int32_t minimum, fop_fentrylk_cbk_t func, void *data,
+            uint32_t fop_flags, fop_fentrylk_cbk_t func, void *data,
             const char *volume, fd_t *fd, const char *basename, entrylk_cmd cmd,
             entrylk_type type, dict_t *xdata)
 {
@@ -417,9 +445,9 @@ ec_fentrylk(call_frame_t *frame, xlator_t *this, uintptr_t target,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FENTRYLK, 0, target, minimum,
-                               ec_wind_fentrylk, ec_manager_entrylk, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FENTRYLK, 0, target,
+                               fop_flags, ec_wind_fentrylk, ec_manager_entrylk,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -651,7 +679,7 @@ ec_manager_inodelk(ec_fop_data_t *fop, int32_t state)
 
 void
 ec_inodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
-           uintptr_t target, int32_t minimum, fop_inodelk_cbk_t func,
+           uintptr_t target, uint32_t fop_flags, fop_inodelk_cbk_t func,
            void *data, const char *volume, loc_t *loc, int32_t cmd,
            struct gf_flock *flock, dict_t *xdata)
 {
@@ -665,9 +693,9 @@ ec_inodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_INODELK, 0, target, minimum,
-                               ec_wind_inodelk, ec_manager_inodelk, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_INODELK, 0, target,
+                               fop_flags, ec_wind_inodelk, ec_manager_inodelk,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -783,7 +811,7 @@ ec_wind_finodelk(ec_t *ec, ec_fop_data_t *fop, int32_t idx)
 
 void
 ec_finodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
-            uintptr_t target, int32_t minimum, fop_finodelk_cbk_t func,
+            uintptr_t target, uint32_t fop_flags, fop_finodelk_cbk_t func,
             void *data, const char *volume, fd_t *fd, int32_t cmd,
             struct gf_flock *flock, dict_t *xdata)
 {
@@ -797,9 +825,9 @@ ec_finodelk(call_frame_t *frame, xlator_t *this, gf_lkowner_t *owner,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_FINODELK, 0, target, minimum,
-                               ec_wind_finodelk, ec_manager_inodelk, callback,
-                               data);
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_FINODELK, 0, target,
+                               fop_flags, ec_wind_finodelk, ec_manager_inodelk,
+                               callback, data);
     if (fop == NULL) {
         goto out;
     }
@@ -1033,7 +1061,7 @@ ec_manager_lk(ec_fop_data_t *fop, int32_t state)
 }
 
 void
-ec_lk(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
+ec_lk(call_frame_t *frame, xlator_t *this, uintptr_t target, uint32_t fop_flags,
       fop_lk_cbk_t func, void *data, fd_t *fd, int32_t cmd,
       struct gf_flock *flock, dict_t *xdata)
 {
@@ -1046,7 +1074,7 @@ ec_lk(call_frame_t *frame, xlator_t *this, uintptr_t target, int32_t minimum,
     GF_VALIDATE_OR_GOTO(this->name, frame, out);
     GF_VALIDATE_OR_GOTO(this->name, this->private, out);
 
-    fop = ec_fop_data_allocate(frame, this, GF_FOP_LK, 0, target, minimum,
+    fop = ec_fop_data_allocate(frame, this, GF_FOP_LK, 0, target, fop_flags,
                                ec_wind_lk, ec_manager_lk, callback, data);
     if (fop == NULL) {
         goto out;
diff --git a/xlators/cluster/ec/src/ec-mem-types.h b/xlators/cluster/ec/src/ec-mem-types.h
index 7af8fe2ef49..3252c4c1c58 100644
--- a/xlators/cluster/ec/src/ec-mem-types.h
+++ b/xlators/cluster/ec/src/ec-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __EC_MEM_TYPES_H__
 #define __EC_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_ec_mem_types_ {
     ec_mt_ec_t = gf_common_mt_end + 1,
diff --git a/xlators/cluster/ec/src/ec-messages.h b/xlators/cluster/ec/src/ec-messages.h
index 5f673d69aa4..72e98f11286 100644
--- a/xlators/cluster/ec/src/ec-messages.h
+++ b/xlators/cluster/ec/src/ec-messages.h
@@ -11,7 +11,7 @@
 #ifndef _EC_MESSAGES_H_
 #define _EC_MESSAGES_H_
 
-#include "glfs-message-id.h"
+#include <glusterfs/glfs-message-id.h>
 
 /* To add new message IDs, append new identifiers at the end of the list.
  *
@@ -55,6 +55,7 @@ GLFS_MSGID(EC, EC_MSG_INVALID_CONFIG, EC_MSG_HEAL_FAIL,
            EC_MSG_CONFIG_XATTR_INVALID, EC_MSG_EXTENSION, EC_MSG_EXTENSION_NONE,
            EC_MSG_EXTENSION_UNKNOWN, EC_MSG_EXTENSION_UNSUPPORTED,
            EC_MSG_EXTENSION_FAILED, EC_MSG_NO_GF, EC_MSG_MATRIX_FAILED,
-           EC_MSG_DYN_CREATE_FAILED, EC_MSG_DYN_CODEGEN_FAILED);
+           EC_MSG_DYN_CREATE_FAILED, EC_MSG_DYN_CODEGEN_FAILED,
+           EC_MSG_THREAD_CLEANUP_FAILED, EC_MSG_FD_BAD);
 
 #endif /* !_EC_MESSAGES_H_ */
diff --git a/xlators/cluster/ec/src/ec-method.h b/xlators/cluster/ec/src/ec-method.h
index ca33f4ffdce..f91233b2f88 100644
--- a/xlators/cluster/ec/src/ec-method.h
+++ b/xlators/cluster/ec/src/ec-method.h
@@ -11,8 +11,6 @@
 #ifndef __EC_METHOD_H__
 #define __EC_METHOD_H__
 
-#include "xlator.h"
-
 #include "ec-types.h"
 #include "ec-galois.h"
 
diff --git a/xlators/cluster/ec/src/ec-types.h b/xlators/cluster/ec/src/ec-types.h
index 80d9c0d4014..de9b89bb2c9 100644
--- a/xlators/cluster/ec/src/ec-types.h
+++ b/xlators/cluster/ec/src/ec-types.h
@@ -11,10 +11,10 @@
 #ifndef __EC_TYPES_H__
 #define __EC_TYPES_H__
 
-#include "xlator.h"
-#include "timer.h"
+#include <glusterfs/xlator.h>
+#include <glusterfs/timer.h>
 #include "libxlator.h"
-#include "atomic.h"
+#include <glusterfs/atomic.h>
 
 #define EC_GF_MAX_REGS 16
 
@@ -130,7 +130,12 @@ typedef void (*ec_resume_f)(ec_fop_data_t *, int32_t);
 
 enum _ec_read_policy { EC_ROUND_ROBIN, EC_GFID_HASH, EC_READ_POLICY_MAX };
 
-enum _ec_heal_need { EC_HEAL_NONEED, EC_HEAL_MAYBE, EC_HEAL_MUST };
+enum _ec_heal_need {
+    EC_HEAL_NONEED,
+    EC_HEAL_MAYBE,
+    EC_HEAL_MUST,
+    EC_HEAL_PURGE_INDEX
+};
 
 enum _ec_stripe_part { EC_STRIPE_HEAD, EC_STRIPE_TAIL };
 
@@ -150,6 +155,7 @@ struct _ec_fd {
     loc_t loc;
     uintptr_t open;
     int32_t flags;
+    uint64_t bad_version;
     ec_fd_status_t fd_status[0];
 };
 
@@ -171,6 +177,7 @@ struct _ec_inode {
     gf_boolean_t have_config;
     gf_boolean_t have_version;
     gf_boolean_t have_size;
+    int32_t heal_count;
     ec_config_t config;
     uint64_t pre_version[2];
     uint64_t post_version[2];
@@ -179,14 +186,15 @@ struct _ec_inode {
     uint64_t dirty[2];
     struct list_head heal;
     ec_stripe_list_t stripe_cache;
+    uint64_t bad_version;
 };
 
 typedef int32_t (*fop_heal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
                                   int32_t, uintptr_t, uintptr_t, uintptr_t,
-                                  dict_t *);
+                                  uint32_t, dict_t *);
 typedef int32_t (*fop_fheal_cbk_t)(call_frame_t *, void *, xlator_t *, int32_t,
                                    int32_t, uintptr_t, uintptr_t, uintptr_t,
-                                   dict_t *);
+                                   uint32_t, dict_t *);
 
 union _ec_cbk {
     fop_access_cbk_t access;
@@ -264,6 +272,7 @@ struct _ec_lock {
     uint32_t refs_pending;  /* Refs assigned to fops being prepared */
     uint32_t waiting_flags; /*Track xattrop/dirty marking*/
     gf_boolean_t acquired;
+    gf_boolean_t contention;
     gf_boolean_t unlock_now;
     gf_boolean_t release;
     gf_boolean_t query;
@@ -307,9 +316,9 @@ struct _ec_fop_data {
     int32_t id; /* ID of the file operation */
     int32_t refs;
     int32_t state;
-    int32_t minimum; /* Minimum number of successful
-                        operation required to conclude a
-                        fop as successful */
+    uint32_t minimum; /* Minimum number of successful
+                         operation required to conclude a
+                         fop as successful */
     int32_t expected;
     int32_t winds;
     int32_t jobs;
@@ -324,11 +333,12 @@ struct _ec_fop_data {
     ec_cbk_data_t *answer;         /* accepted answer */
     int32_t lock_count;
     int32_t locked;
+    gf_lock_t lock;
     ec_lock_link_t locks[2];
     int32_t first_lock;
-    gf_lock_t lock;
 
-    uint32_t flags;
+    uint32_t fop_flags; /* Flags passed by the caller. */
+    uint32_t flags;     /* Internal flags. */
     uint32_t first;
     uintptr_t mask;
     uintptr_t healing; /*Dispatch is done but call is successful only
@@ -616,6 +626,11 @@ struct _ec_statistics {
                                 requests. (Basically memory allocation
                                 errors). */
     } stripe_cache;
+    struct {
+        gf_atomic_t attempted; /*Number of heals attempted on
+                                files/directories*/
+        gf_atomic_t completed; /*Number of heals complted on files/directories*/
+    } shd;
 };
 
 struct _ec {
@@ -641,6 +656,8 @@ struct _ec {
     uintptr_t xl_notify;      /* Bit flag representing
                                  notification for bricks. */
     uintptr_t node_mask;
+    uintptr_t read_mask;         /*Stores user defined read-mask*/
+    gf_atomic_t async_fop_count; /* Number of on going asynchronous fops. */
     xlator_t **xl_list;
     gf_lock_t lock;
     gf_timer_t *timer;
@@ -650,6 +667,7 @@ struct _ec {
     gf_boolean_t optimistic_changelog;
     gf_boolean_t parallel_writes;
     uint32_t stripe_cache;
+    uint32_t quorum_count;
     uint32_t background_heals;
     uint32_t heal_wait_qlen;
     uint32_t self_heal_window_size; /* max size of read/writes */
diff --git a/xlators/cluster/ec/src/ec.c b/xlators/cluster/ec/src/ec.c
index b56dd9ada3e..7344be4968d 100644
--- a/xlators/cluster/ec/src/ec.c
+++ b/xlators/cluster/ec/src/ec.c
@@ -8,10 +8,10 @@
   cases as published by the Free Software Foundation.
 */
 
-#include "defaults.h"
-#include "statedump.h"
-#include "compat-errno.h"
-#include "upcall-utils.h"
+#include <glusterfs/defaults.h>
+#include <glusterfs/statedump.h>
+#include <glusterfs/compat-errno.h>
+#include <glusterfs/upcall-utils.h>
 
 #include "ec.h"
 #include "ec-messages.h"
@@ -23,7 +23,7 @@
 #include "ec-method.h"
 #include "ec-code.h"
 #include "ec-heald.h"
-#include "events.h"
+#include <glusterfs/events.h>
 
 static char *ec_read_policies[EC_READ_POLICY_MAX + 1] = {
     [EC_ROUND_ROBIN] = "round-robin",
@@ -285,6 +285,7 @@ reconfigure(xlator_t *this, dict_t *options)
     GF_OPTION_RECONF("parallel-writes", ec->parallel_writes, options, bool,
                      failed);
     GF_OPTION_RECONF("stripe-cache", ec->stripe_cache, options, uint32, failed);
+    GF_OPTION_RECONF("quorum-count", ec->quorum_count, options, uint32, failed);
     ret = 0;
     if (ec_assign_read_policy(ec, read_policy)) {
         ret = -1;
@@ -324,13 +325,18 @@ ec_get_event_from_state(ec_t *ec)
 void
 ec_up(xlator_t *this, ec_t *ec)
 {
+    char str1[32], str2[32];
+
     if (ec->timer != NULL) {
         gf_timer_call_cancel(this->ctx, ec->timer);
         ec->timer = NULL;
     }
 
     ec->up = 1;
-    gf_msg(this->name, GF_LOG_INFO, 0, EC_MSG_EC_UP, "Going UP");
+    gf_msg(this->name, GF_LOG_INFO, 0, EC_MSG_EC_UP,
+           "Going UP : Child UP = %s Child Notify = %s",
+           ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes),
+           ec_bin(str2, sizeof(str2), ec->xl_notify, ec->nodes));
 
     gf_event(EVENT_EC_MIN_BRICKS_UP, "subvol=%s", this->name);
 }
@@ -338,13 +344,18 @@ ec_up(xlator_t *this, ec_t *ec)
 void
 ec_down(xlator_t *this, ec_t *ec)
 {
+    char str1[32], str2[32];
+
     if (ec->timer != NULL) {
         gf_timer_call_cancel(this->ctx, ec->timer);
         ec->timer = NULL;
     }
 
     ec->up = 0;
-    gf_msg(this->name, GF_LOG_INFO, 0, EC_MSG_EC_DOWN, "Going DOWN");
+    gf_msg(this->name, GF_LOG_INFO, 0, EC_MSG_EC_DOWN,
+           "Going DOWN : Child UP = %s Child Notify = %s",
+           ec_bin(str1, sizeof(str1), ec->xl_up, ec->nodes),
+           ec_bin(str2, sizeof(str2), ec->xl_notify, ec->nodes));
 
     gf_event(EVENT_EC_MIN_BRICKS_NOT_UP, "subvol=%s", this->name);
 }
@@ -355,6 +366,7 @@ ec_notify_cbk(void *data)
     ec_t *ec = data;
     glusterfs_event_t event = GF_EVENT_MAXVAL;
     gf_boolean_t propagate = _gf_false;
+    gf_boolean_t launch_heal = _gf_false;
 
     LOCK(&ec->lock);
     {
@@ -384,6 +396,11 @@ ec_notify_cbk(void *data)
              * still bricks DOWN, they will be healed when they
              * come up. */
             ec_up(ec->xl, ec);
+
+            if (ec->shd.iamshd && !ec->shutdown) {
+                launch_heal = _gf_true;
+                GF_ATOMIC_INC(ec->async_fop_count);
+            }
         }
 
         propagate = _gf_true;
@@ -391,13 +408,12 @@ ec_notify_cbk(void *data)
 unlock:
     UNLOCK(&ec->lock);
 
+    if (launch_heal) {
+        /* We have just brought the volume UP, so we trigger
+         * a self-heal check on the root directory. */
+        ec_launch_replace_heal(ec);
+    }
     if (propagate) {
-        if ((event == GF_EVENT_CHILD_UP) && ec->shd.iamshd) {
-            /* We have just brought the volume UP, so we trigger
-             * a self-heal check on the root directory. */
-            ec_launch_replace_heal(ec);
-        }
-
         default_notify(ec->xl, event, NULL);
     }
 }
@@ -425,10 +441,55 @@ ec_disable_delays(ec_t *ec)
 {
     ec->shutdown = _gf_true;
 
-    return list_empty(&ec->pending_fops);
+    return __ec_is_last_fop(ec);
 }
 
 void
+ec_cleanup_healer_object(ec_t *ec)
+{
+    struct subvol_healer *healer = NULL;
+    ec_self_heald_t *shd = NULL;
+    void *res = NULL;
+    int i = 0;
+    gf_boolean_t is_join = _gf_false;
+
+    shd = &ec->shd;
+    if (!shd->iamshd)
+        return;
+
+    for (i = 0; i < ec->nodes; i++) {
+        healer = &shd->index_healers[i];
+        pthread_mutex_lock(&healer->mutex);
+        {
+            healer->rerun = 1;
+            if (healer->running) {
+                pthread_cond_signal(&healer->cond);
+                is_join = _gf_true;
+            }
+        }
+        pthread_mutex_unlock(&healer->mutex);
+        if (is_join) {
+            pthread_join(healer->thread, &res);
+            is_join = _gf_false;
+        }
+
+        healer = &shd->full_healers[i];
+        pthread_mutex_lock(&healer->mutex);
+        {
+            healer->rerun = 1;
+            if (healer->running) {
+                pthread_cond_signal(&healer->cond);
+                is_join = _gf_true;
+            }
+        }
+        pthread_mutex_unlock(&healer->mutex);
+        if (is_join) {
+            pthread_join(healer->thread, &res);
+            is_join = _gf_false;
+        }
+    }
+}
+void
 ec_pending_fops_completed(ec_t *ec)
 {
     if (ec->shutdown) {
@@ -441,6 +502,9 @@ ec_set_up_state(ec_t *ec, uintptr_t index_mask, uintptr_t new_state)
 {
     uintptr_t current_state = 0;
 
+    if (xlator_is_cleanup_starting(ec->xl))
+        return _gf_false;
+
     if ((ec->xl_notify & index_mask) == 0) {
         ec->xl_notify |= index_mask;
         ec->xl_notify_count++;
@@ -462,6 +526,7 @@ ec_upcall(ec_t *ec, struct gf_upcall *upcall)
     struct gf_upcall_cache_invalidation *ci = NULL;
     struct gf_upcall_inodelk_contention *lc = NULL;
     inode_t *inode;
+    inode_table_t *table;
 
     switch (upcall->event_type) {
         case GF_UPCALL_CACHE_INVALIDATION:
@@ -475,8 +540,18 @@ ec_upcall(ec_t *ec, struct gf_upcall *upcall)
                 /* The lock is not owned by EC, ignore it. */
                 return _gf_true;
             }
-            inode = inode_find(((xlator_t *)ec->xl->graph->top)->itable,
-                               upcall->gfid);
+            table = ((xlator_t *)ec->xl->graph->top)->itable;
+            if (table == NULL) {
+                /* Self-heal daemon doesn't have an inode table on the top
+                 * xlator because it doesn't need it. In this case we should
+                 * use the inode table managed by EC itself where all inodes
+                 * being healed should be present. However self-heal doesn't
+                 * use eager-locking and inodelk's are already released as
+                 * soon as possible. In this case we can safely ignore these
+                 * notifications. */
+                return _gf_false;
+            }
+            inode = inode_find(table, upcall->gfid);
             /* If inode is not found, it means that it's already released,
              * so we can ignore it. Probably it has been released and
              * destroyed while the contention notification was being sent.
@@ -544,6 +619,7 @@ ec_notify(xlator_t *this, int32_t event, void *data, void *data2)
         /* If there aren't pending fops running after we have waken up
          * them, we immediately propagate the notification. */
         propagate = ec_disable_delays(ec);
+        ec_cleanup_healer_object(ec);
         goto unlock;
     }
 
@@ -554,7 +630,10 @@ ec_notify(xlator_t *this, int32_t event, void *data, void *data2)
         if (event == GF_EVENT_CHILD_UP) {
             /* We need to trigger a selfheal if a brick changes
              * to UP state. */
-            needs_shd_check = ec_set_up_state(ec, mask, mask);
+            if (ec_set_up_state(ec, mask, mask) && ec->shd.iamshd &&
+                !ec->shutdown) {
+                needs_shd_check = _gf_true;
+            }
         } else if (event == GF_EVENT_CHILD_DOWN) {
             ec_set_up_state(ec, mask, 0);
         }
@@ -584,17 +663,21 @@ ec_notify(xlator_t *this, int32_t event, void *data, void *data2)
             }
         } else {
             propagate = _gf_false;
+            needs_shd_check = _gf_false;
+        }
+
+        if (needs_shd_check) {
+            GF_ATOMIC_INC(ec->async_fop_count);
         }
     }
 unlock:
     UNLOCK(&ec->lock);
 
 done:
+    if (needs_shd_check) {
+        ec_launch_replace_heal(ec);
+    }
     if (propagate) {
-        if (needs_shd_check && ec->shd.iamshd) {
-            ec_launch_replace_heal(ec);
-        }
-
         error = default_notify(this, event, data);
     }
 
@@ -627,6 +710,69 @@ ec_statistics_init(ec_t *ec)
     GF_ATOMIC_INIT(ec->stats.stripe_cache.evicts, 0);
     GF_ATOMIC_INIT(ec->stats.stripe_cache.allocs, 0);
     GF_ATOMIC_INIT(ec->stats.stripe_cache.errors, 0);
+    GF_ATOMIC_INIT(ec->stats.shd.attempted, 0);
+    GF_ATOMIC_INIT(ec->stats.shd.completed, 0);
+}
+
+static int
+ec_assign_read_mask(ec_t *ec, char *read_mask_str)
+{
+    char *mask = NULL;
+    char *maskptr = NULL;
+    char *saveptr = NULL;
+    char *id_str = NULL;
+    int id = 0;
+    int ret = 0;
+    uintptr_t read_mask = 0;
+
+    if (!read_mask_str) {
+        ec->read_mask = 0;
+        ret = 0;
+        goto out;
+    }
+
+    mask = gf_strdup(read_mask_str);
+    if (!mask) {
+        ret = -1;
+        goto out;
+    }
+    maskptr = mask;
+
+    for (;;) {
+        id_str = strtok_r(maskptr, ":", &saveptr);
+        if (id_str == NULL)
+            break;
+        if (gf_string2int(id_str, &id)) {
+            gf_msg(ec->xl->name, GF_LOG_ERROR, 0, EC_MSG_XLATOR_INIT_FAIL,
+                   "In read-mask \"%s\" id %s is not a valid integer",
+                   read_mask_str, id_str);
+            ret = -1;
+            goto out;
+        }
+
+        if ((id < 0) || (id >= ec->nodes)) {
+            gf_msg(ec->xl->name, GF_LOG_ERROR, 0, EC_MSG_XLATOR_INIT_FAIL,
+                   "In read-mask \"%s\" id %d is not in range [0 - %d]",
+                   read_mask_str, id, ec->nodes - 1);
+            ret = -1;
+            goto out;
+        }
+        read_mask |= (1UL << id);
+        maskptr = NULL;
+    }
+
+    if (gf_bits_count(read_mask) < ec->fragments) {
+        gf_msg(ec->xl->name, GF_LOG_ERROR, 0, EC_MSG_XLATOR_INIT_FAIL,
+               "read-mask \"%s\" should contain at least %d ids", read_mask_str,
+               ec->fragments);
+        ret = -1;
+        goto out;
+    }
+    ec->read_mask = read_mask;
+    ret = 0;
+out:
+    GF_FREE(mask);
+    return ret;
 }
 
 int32_t
@@ -636,6 +782,7 @@ init(xlator_t *this)
     char *read_policy = NULL;
     char *extensions = NULL;
     int32_t err;
+    char *read_mask_str = NULL;
 
     if (this->parents == NULL) {
         gf_msg(this->name, GF_LOG_WARNING, 0, EC_MSG_NO_PARENTS,
@@ -656,6 +803,7 @@ init(xlator_t *this)
     ec->xl = this;
     LOCK_INIT(&ec->lock);
 
+    GF_ATOMIC_INIT(ec->async_fop_count, 0);
     INIT_LIST_HEAD(&ec->pending_fops);
     INIT_LIST_HEAD(&ec->heal_waiting);
     INIT_LIST_HEAD(&ec->healing);
@@ -714,12 +862,18 @@ init(xlator_t *this)
     if (ec_assign_read_policy(ec, read_policy))
         goto failed;
 
+    GF_OPTION_INIT("heal-timeout", ec->shd.timeout, int32, failed);
     GF_OPTION_INIT("shd-max-threads", ec->shd.max_threads, uint32, failed);
     GF_OPTION_INIT("shd-wait-qlength", ec->shd.wait_qlength, uint32, failed);
     GF_OPTION_INIT("optimistic-change-log", ec->optimistic_changelog, bool,
                    failed);
     GF_OPTION_INIT("parallel-writes", ec->parallel_writes, bool, failed);
     GF_OPTION_INIT("stripe-cache", ec->stripe_cache, uint32, failed);
+    GF_OPTION_INIT("quorum-count", ec->quorum_count, uint32, failed);
+    GF_OPTION_INIT("ec-read-mask", read_mask_str, str, failed);
+
+    if (ec_assign_read_mask(ec, read_mask_str))
+        goto failed;
 
     this->itable = inode_table_new(EC_SHD_INODE_LRU_LIMIT, this);
     if (!this->itable)
@@ -759,6 +913,7 @@ failed:
 void
 fini(xlator_t *this)
 {
+    ec_selfheal_daemon_fini(this);
     __ec_destroy_private(this);
 }
 
@@ -797,11 +952,12 @@ ec_gf_entrylk(call_frame_t *frame, xlator_t *this, const char *volume,
               loc_t *loc, const char *basename, entrylk_cmd cmd,
               entrylk_type type, dict_t *xdata)
 {
-    int32_t minimum = EC_MINIMUM_ALL;
+    uint32_t fop_flags = EC_MINIMUM_ALL;
+
     if (cmd == ENTRYLK_UNLOCK)
-        minimum = EC_MINIMUM_ONE;
-    ec_entrylk(frame, this, -1, minimum, default_entrylk_cbk, NULL, volume, loc,
-               basename, cmd, type, xdata);
+        fop_flags = EC_MINIMUM_ONE;
+    ec_entrylk(frame, this, -1, fop_flags, default_entrylk_cbk, NULL, volume,
+               loc, basename, cmd, type, xdata);
 
     return 0;
 }
@@ -811,10 +967,11 @@ ec_gf_fentrylk(call_frame_t *frame, xlator_t *this, const char *volume,
                fd_t *fd, const char *basename, entrylk_cmd cmd,
                entrylk_type type, dict_t *xdata)
 {
-    int32_t minimum = EC_MINIMUM_ALL;
+    uint32_t fop_flags = EC_MINIMUM_ALL;
+
     if (cmd == ENTRYLK_UNLOCK)
-        minimum = EC_MINIMUM_ONE;
-    ec_fentrylk(frame, this, -1, minimum, default_fentrylk_cbk, NULL, volume,
+        fop_flags = EC_MINIMUM_ONE;
+    ec_fentrylk(frame, this, -1, fop_flags, default_fentrylk_cbk, NULL, volume,
                 fd, basename, cmd, type, xdata);
 
     return 0;
@@ -905,7 +1062,7 @@ ec_gf_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
 {
     int error = 0;
     ec_t *ec = this->private;
-    int32_t minimum = EC_MINIMUM_ONE;
+    int32_t fop_flags = EC_MINIMUM_ONE;
 
     if (name && strcmp(name, EC_XATTR_HEAL) != 0) {
         EC_INTERNAL_XATTR_OR_GOTO(name, NULL, error, out);
@@ -920,11 +1077,11 @@ ec_gf_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
 
     if (name && ((fnmatch(GF_XATTR_STIME_PATTERN, name, 0) == 0) ||
                  XATTR_IS_NODE_UUID(name) || XATTR_IS_NODE_UUID_LIST(name))) {
-        minimum = EC_MINIMUM_ALL;
+        fop_flags = EC_MINIMUM_ALL;
     }
 
-    ec_getxattr(frame, this, -1, minimum, default_getxattr_cbk, NULL, loc, name,
-                xdata);
+    ec_getxattr(frame, this, -1, fop_flags, default_getxattr_cbk, NULL, loc,
+                name, xdata);
 
     return 0;
 out:
@@ -954,11 +1111,12 @@ int32_t
 ec_gf_inodelk(call_frame_t *frame, xlator_t *this, const char *volume,
               loc_t *loc, int32_t cmd, struct gf_flock *flock, dict_t *xdata)
 {
-    int32_t minimum = EC_MINIMUM_ALL;
+    int32_t fop_flags = EC_MINIMUM_ALL;
+
     if (flock->l_type == F_UNLCK)
-        minimum = EC_MINIMUM_ONE;
+        fop_flags = EC_MINIMUM_ONE;
 
-    ec_inodelk(frame, this, &frame->root->lk_owner, -1, minimum,
+    ec_inodelk(frame, this, &frame->root->lk_owner, -1, fop_flags,
                default_inodelk_cbk, NULL, volume, loc, cmd, flock, xdata);
 
     return 0;
@@ -968,10 +1126,11 @@ int32_t
 ec_gf_finodelk(call_frame_t *frame, xlator_t *this, const char *volume,
                fd_t *fd, int32_t cmd, struct gf_flock *flock, dict_t *xdata)
 {
-    int32_t minimum = EC_MINIMUM_ALL;
+    int32_t fop_flags = EC_MINIMUM_ALL;
+
     if (flock->l_type == F_UNLCK)
-        minimum = EC_MINIMUM_ONE;
-    ec_finodelk(frame, this, &frame->root->lk_owner, -1, minimum,
+        fop_flags = EC_MINIMUM_ONE;
+    ec_finodelk(frame, this, &frame->root->lk_owner, -1, fop_flags,
                 default_finodelk_cbk, NULL, volume, fd, cmd, flock, xdata);
 
     return 0;
@@ -991,10 +1150,11 @@ int32_t
 ec_gf_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
          struct gf_flock *flock, dict_t *xdata)
 {
-    int32_t minimum = EC_MINIMUM_ALL;
+    int32_t fop_flags = EC_MINIMUM_ALL;
+
     if (flock->l_type == F_UNLCK)
-        minimum = EC_MINIMUM_ONE;
-    ec_lk(frame, this, -1, minimum, default_lk_cbk, NULL, fd, cmd, flock,
+        fop_flags = EC_MINIMUM_ONE;
+    ec_lk(frame, this, -1, fop_flags, default_lk_cbk, NULL, fd, cmd, flock,
           xdata);
 
     return 0;
@@ -1389,6 +1549,10 @@ ec_dump_private(xlator_t *this)
     gf_proc_dump_write("childs_up", "%u", ec->xl_up_count);
     gf_proc_dump_write("childs_up_mask", "%s",
                        ec_bin(tmp, sizeof(tmp), ec->xl_up, ec->nodes));
+    if (ec->read_mask) {
+        gf_proc_dump_write("read-mask", "%s",
+                           ec_bin(tmp, sizeof(tmp), ec->read_mask, ec->nodes));
+    }
     gf_proc_dump_write("background-heals", "%d", ec->background_heals);
     gf_proc_dump_write("heal-wait-qlength", "%d", ec->heal_wait_qlen);
     gf_proc_dump_write("self-heal-window-size", "%" PRIu32,
@@ -1397,6 +1561,7 @@ ec_dump_private(xlator_t *this)
     gf_proc_dump_write("heal-waiters", "%d", ec->heal_waiters);
     gf_proc_dump_write("read-policy", "%s", ec_read_policies[ec->read_policy]);
     gf_proc_dump_write("parallel-writes", "%d", ec->parallel_writes);
+    gf_proc_dump_write("quorum-count", "%u", ec->quorum_count);
 
     snprintf(key_prefix, GF_DUMP_MAX_BUF_LEN, "%s.%s.stats.stripe_cache",
              this->type, this->name);
@@ -1416,6 +1581,10 @@ ec_dump_private(xlator_t *this)
                        GF_ATOMIC_GET(ec->stats.stripe_cache.allocs));
     gf_proc_dump_write("errors", "%" GF_PRI_ATOMIC,
                        GF_ATOMIC_GET(ec->stats.stripe_cache.errors));
+    gf_proc_dump_write("heals-attempted", "%" GF_PRI_ATOMIC,
+                       GF_ATOMIC_GET(ec->stats.shd.attempted));
+    gf_proc_dump_write("heals-completed", "%" GF_PRI_ATOMIC,
+                       GF_ATOMIC_GET(ec->stats.shd.completed));
 
     return 0;
 }
@@ -1666,4 +1835,39 @@ struct volume_options options[] = {
                     "specially for sequential writes. However, this will also"
                     "lead to extra memory consumption, maximum "
                     "(cache size * stripe size) Bytes per open file."},
-    {.key = {NULL}}};
+    {
+        .key = {"quorum-count"},
+        .type = GF_OPTION_TYPE_INT,
+        .default_value = "0",
+        .description =
+            "This option can be used to define how many successes on"
+            "the bricks constitute a success to the application. This"
+            " count should be in the range"
+            "[disperse-data-count,  disperse-count] (inclusive)",
+    },
+    {
+        .key = {"ec-read-mask"},
+        .type = GF_OPTION_TYPE_STR,
+        .default_value = NULL,
+        .description = "This option can be used to choose which bricks can be"
+                       " used for reading data/metadata of a file/directory",
+    },
+    {
+        .key = {NULL},
+    },
+};
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .notify = notify,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1},
+    .dumpops = &dumpops,
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "disperse",
+    .category = GF_MAINTAINED,
+};
diff --git a/xlators/cluster/ec/src/ec.h b/xlators/cluster/ec/src/ec.h
index 1b210d9adc1..6f6de6d5981 100644
--- a/xlators/cluster/ec/src/ec.h
+++ b/xlators/cluster/ec/src/ec.h
@@ -18,6 +18,7 @@
 #define EC_XATTR_SIZE EC_XATTR_PREFIX "size"
 #define EC_XATTR_VERSION EC_XATTR_PREFIX "version"
 #define EC_XATTR_HEAL EC_XATTR_PREFIX "heal"
+#define EC_XATTR_HEAL_NEW EC_XATTR_PREFIX "heal-new"
 #define EC_XATTR_DIRTY EC_XATTR_PREFIX "dirty"
 #define EC_STRIPE_CACHE_MAX_SIZE 10
 #define EC_VERSION_SIZE 2
diff --git a/xlators/cluster/stripe/src/Makefile.am b/xlators/cluster/stripe/src/Makefile.am
deleted file mode 100644
index 2b594567db1..00000000000
--- a/xlators/cluster/stripe/src/Makefile.am
+++ /dev/null
@@ -1,22 +0,0 @@
-xlator_LTLIBRARIES = stripe.la
-xlatordir = $(libdir)/glusterfs/$(PACKAGE_VERSION)/xlator/cluster
-
-stripe_la_LDFLAGS = -module $(GF_XLATOR_DEFAULT_LDFLAGS)
-
-
-stripe_la_SOURCES = stripe.c stripe-helpers.c \
-	$(top_builddir)/xlators/lib/src/libxlator.c
-
-stripe_la_LIBADD = $(top_builddir)/libglusterfs/src/libglusterfs.la
-
-noinst_HEADERS = stripe.h stripe-mem-types.h \
-	$(top_builddir)/xlators/lib/src/libxlator.h
-
-AM_CPPFLAGS = $(GF_CPPFLAGS) -I$(top_srcdir)/libglusterfs/src \
-	-I$(top_srcdir)/xlators/lib/src \
-	-I$(top_srcdir)/rpc/xdr/src -I$(top_builddir)/rpc/xdr/src
-
-AM_CFLAGS = -Wall $(GF_CFLAGS)
-
-CLEANFILES = 
-
diff --git a/xlators/cluster/stripe/src/stripe-helpers.c b/xlators/cluster/stripe/src/stripe-helpers.c
deleted file mode 100644
index c3743723d1d..00000000000
--- a/xlators/cluster/stripe/src/stripe-helpers.c
+++ /dev/null
@@ -1,658 +0,0 @@
-/*
-  Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#include <fnmatch.h>
-
-#include "stripe.h"
-#include "byte-order.h"
-#include "mem-types.h"
-#include "logging.h"
-
-void
-stripe_local_wipe(stripe_local_t *local)
-{
-    if (!local)
-        goto out;
-
-    loc_wipe(&local->loc);
-    loc_wipe(&local->loc2);
-
-    if (local->fd)
-        fd_unref(local->fd);
-
-    if (local->inode)
-        inode_unref(local->inode);
-
-    if (local->xattr)
-        dict_unref(local->xattr);
-
-    if (local->xdata)
-        dict_unref(local->xdata);
-
-out:
-    return;
-}
-
-int
-stripe_aggregate(dict_t *this, char *key, data_t *value, void *data)
-{
-    dict_t *dst = NULL;
-    int64_t *ptr = 0, *size = NULL;
-    int32_t ret = -1;
-
-    dst = data;
-
-    if (strcmp(key, QUOTA_SIZE_KEY) == 0) {
-        ret = dict_get_bin(dst, key, (void **)&size);
-        if (ret < 0) {
-            size = GF_CALLOC(1, sizeof(int64_t), gf_common_mt_char);
-            if (size == NULL) {
-                gf_log("stripe", GF_LOG_WARNING, "memory allocation failed");
-                goto out;
-            }
-            ret = dict_set_bin(dst, key, size, sizeof(int64_t));
-            if (ret < 0) {
-                gf_log("stripe", GF_LOG_WARNING,
-                       "stripe aggregate dict set failed");
-                GF_FREE(size);
-                goto out;
-            }
-        }
-
-        ptr = data_to_bin(value);
-        if (ptr == NULL) {
-            gf_log("stripe", GF_LOG_WARNING, "data to bin failed");
-            goto out;
-        }
-
-        *size = hton64(ntoh64(*size) + ntoh64(*ptr));
-    } else if (strcmp(key, GF_CONTENT_KEY)) {
-        /* No need to aggregate 'CONTENT' data */
-        ret = dict_set(dst, key, value);
-        if (ret)
-            gf_log("stripe", GF_LOG_WARNING, "xattr dict set failed");
-    }
-
-out:
-    return 0;
-}
-
-void
-stripe_aggregate_xattr(dict_t *dst, dict_t *src)
-{
-    if ((dst == NULL) || (src == NULL)) {
-        goto out;
-    }
-
-    dict_foreach(src, stripe_aggregate, dst);
-out:
-    return;
-}
-
-int32_t
-stripe_xattr_aggregate(char *buffer, stripe_local_t *local, int32_t *total)
-{
-    int32_t i = 0;
-    int32_t ret = -1;
-    int32_t len = 0;
-    char *sbuf = NULL;
-    stripe_xattr_sort_t *xattr = NULL;
-
-    if (!buffer || !local || !local->xattr_list)
-        goto out;
-
-    sbuf = buffer;
-
-    for (i = 0; i < local->nallocs; i++) {
-        xattr = local->xattr_list + i;
-        len = xattr->xattr_len - 1; /* length includes \0 */
-
-        if (len && xattr && xattr->xattr_value) {
-            memcpy(buffer, xattr->xattr_value, len);
-            buffer += len;
-            *buffer++ = ' ';
-        }
-    }
-
-    *--buffer = '\0';
-    if (total)
-        *total = buffer - sbuf;
-    ret = 0;
-
-out:
-    return ret;
-}
-
-int32_t
-stripe_free_xattr_str(stripe_local_t *local)
-{
-    int32_t i = 0;
-    int32_t ret = -1;
-    stripe_xattr_sort_t *xattr = NULL;
-
-    if (!local || !local->xattr_list)
-        goto out;
-
-    for (i = 0; i < local->nallocs; i++) {
-        xattr = local->xattr_list + i;
-
-        if (xattr && xattr->xattr_value)
-            GF_FREE(xattr->xattr_value);
-    }
-
-    ret = 0;
-out:
-    return ret;
-}
-
-int32_t
-stripe_fill_lockinfo_xattr(xlator_t *this, stripe_local_t *local,
-                           void **xattr_serz)
-{
-    int32_t ret = -1, i = 0, len = 0;
-    dict_t *tmp1 = NULL, *tmp2 = NULL;
-    char *buf = NULL;
-    stripe_xattr_sort_t *xattr = NULL;
-
-    if (xattr_serz == NULL) {
-        goto out;
-    }
-
-    tmp2 = dict_new();
-
-    if (tmp2 == NULL) {
-        goto out;
-    }
-
-    for (i = 0; i < local->nallocs; i++) {
-        xattr = local->xattr_list + i;
-        len = xattr->xattr_len;
-
-        if (len && xattr && xattr->xattr_value) {
-            ret = dict_reset(tmp2);
-            if (ret < 0) {
-                gf_log(this->name, GF_LOG_DEBUG, "dict_reset failed (%s)",
-                       strerror(-ret));
-            }
-
-            ret = dict_unserialize(xattr->xattr_value, xattr->xattr_len, &tmp2);
-            if (ret < 0) {
-                gf_log(this->name, GF_LOG_WARNING,
-                       "dict_unserialize failed (%s)", strerror(-ret));
-                ret = -1;
-                goto out;
-            }
-
-            tmp1 = dict_copy(tmp2, tmp1);
-            if (tmp1 == NULL) {
-                gf_log(this->name, GF_LOG_WARNING, "dict_copy failed (%s)",
-                       strerror(-ret));
-                ret = -1;
-                goto out;
-            }
-        }
-    }
-
-    len = dict_serialized_length(tmp1);
-    if (len > 0) {
-        buf = GF_CALLOC(1, len, gf_common_mt_dict_t);
-        if (buf == NULL) {
-            ret = -1;
-            goto out;
-        }
-
-        ret = dict_serialize(tmp1, buf);
-        if (ret < 0) {
-            gf_log(this->name, GF_LOG_WARNING, "dict_serialize failed (%s)",
-                   strerror(-ret));
-            GF_FREE(buf);
-            ret = -1;
-            goto out;
-        }
-
-        *xattr_serz = buf;
-    }
-
-    ret = 0;
-out:
-    if (tmp1 != NULL) {
-        dict_unref(tmp1);
-    }
-
-    if (tmp2 != NULL) {
-        dict_unref(tmp2);
-    }
-
-    return ret;
-}
-
-int32_t
-stripe_fill_pathinfo_xattr(xlator_t *this, stripe_local_t *local,
-                           char **xattr_serz)
-{
-    int ret = -1;
-    int32_t padding = 0;
-    int32_t tlen = 0;
-    int len = 0;
-    char stripe_size_str[20] = {
-        0,
-    };
-    char *pathinfo_serz = NULL;
-
-    if (!local) {
-        gf_log(this->name, GF_LOG_ERROR, "Possible NULL deref");
-        goto out;
-    }
-
-    len = snprintf(stripe_size_str, sizeof(stripe_size_str), "%" PRId64,
-                   local->fctx ? local->fctx->stripe_size : 0);
-    if (len < 0 || len >= sizeof(stripe_size_str))
-        goto out;
-    /* extra bytes for decorations (brackets and <>'s) */
-    padding = strlen(this->name) + SLEN(STRIPE_PATHINFO_HEADER) + len + 7;
-    local->xattr_total_len += (padding + 2);
-
-    pathinfo_serz = GF_MALLOC(local->xattr_total_len, gf_common_mt_char);
-    if (!pathinfo_serz)
-        goto out;
-
-    /* xlator info */
-    (void)sprintf(pathinfo_serz, "(<" STRIPE_PATHINFO_HEADER "%s:[%s]> ",
-                  this->name, stripe_size_str);
-
-    ret = stripe_xattr_aggregate(pathinfo_serz + padding, local, &tlen);
-    if (ret) {
-        gf_log(this->name, GF_LOG_ERROR, "Cannot aggregate pathinfo list");
-        GF_FREE(pathinfo_serz);
-        goto out;
-    }
-
-    *(pathinfo_serz + padding + tlen) = ')';
-    *(pathinfo_serz + padding + tlen + 1) = '\0';
-
-    *xattr_serz = pathinfo_serz;
-
-    ret = 0;
-out:
-    return ret;
-}
-
-/**
- * stripe_get_matching_bs - Get the matching block size for the given path.
- */
-int32_t
-stripe_get_matching_bs(const char *path, stripe_private_t *priv)
-{
-    struct stripe_options *trav = NULL;
-    uint64_t block_size = 0;
-
-    GF_VALIDATE_OR_GOTO("stripe", priv, out);
-    GF_VALIDATE_OR_GOTO("stripe", path, out);
-
-    LOCK(&priv->lock);
-    {
-        block_size = priv->block_size;
-        trav = priv->pattern;
-        while (trav) {
-            if (!fnmatch(trav->path_pattern, path, FNM_NOESCAPE)) {
-                block_size = trav->block_size;
-                break;
-            }
-            trav = trav->next;
-        }
-    }
-    UNLOCK(&priv->lock);
-
-out:
-    return block_size;
-}
-
-int32_t
-stripe_ctx_handle(xlator_t *this, call_frame_t *prev, stripe_local_t *local,
-                  dict_t *dict)
-{
-    char key[256] = {
-        0,
-    };
-    data_t *data = NULL;
-    int32_t index = 0;
-    stripe_private_t *priv = NULL;
-
-    priv = this->private;
-
-    if (!local->fctx) {
-        local->fctx = GF_CALLOC(1, sizeof(stripe_fd_ctx_t),
-                                gf_stripe_mt_stripe_fd_ctx_t);
-        if (!local->fctx) {
-            local->op_errno = ENOMEM;
-            local->op_ret = -1;
-            goto out;
-        }
-
-        local->fctx->static_array = 0;
-    }
-    /* Stripe block size */
-    sprintf(key, "trusted.%s.stripe-size", this->name);
-    data = dict_get(dict, key);
-    if (!data) {
-        local->xattr_self_heal_needed = 1;
-        gf_log(this->name, GF_LOG_ERROR, "Failed to get stripe-size");
-        goto out;
-    } else {
-        if (!local->fctx->stripe_size) {
-            local->fctx->stripe_size = data_to_int64(data);
-        }
-
-        if (local->fctx->stripe_size != data_to_int64(data)) {
-            gf_log(this->name, GF_LOG_WARNING,
-                   "stripe-size mismatch in blocks");
-            local->xattr_self_heal_needed = 1;
-        }
-    }
-
-    /* Stripe count */
-    sprintf(key, "trusted.%s.stripe-count", this->name);
-    data = dict_get(dict, key);
-
-    if (!data) {
-        local->xattr_self_heal_needed = 1;
-        gf_log(this->name, GF_LOG_ERROR, "Failed to get stripe-count");
-        goto out;
-    }
-    if (!local->fctx->xl_array) {
-        local->fctx->stripe_count = data_to_int32(data);
-        if (!local->fctx->stripe_count) {
-            gf_log(this->name, GF_LOG_ERROR, "error with stripe-count xattr");
-            local->op_ret = -1;
-            local->op_errno = EIO;
-            goto out;
-        }
-
-        local->fctx->xl_array = GF_CALLOC(local->fctx->stripe_count,
-                                          sizeof(xlator_t *),
-                                          gf_stripe_mt_xlator_t);
-
-        if (!local->fctx->xl_array) {
-            local->op_errno = ENOMEM;
-            local->op_ret = -1;
-            goto out;
-        }
-    }
-    if (local->fctx->stripe_count != data_to_int32(data)) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "error with stripe-count xattr (%d != %d)",
-               local->fctx->stripe_count, data_to_int32(data));
-        local->op_ret = -1;
-        local->op_errno = EIO;
-        goto out;
-    }
-
-    /* index */
-    sprintf(key, "trusted.%s.stripe-index", this->name);
-    data = dict_get(dict, key);
-    if (!data) {
-        local->xattr_self_heal_needed = 1;
-        gf_log(this->name, GF_LOG_ERROR, "Failed to get stripe-index");
-        goto out;
-    }
-    index = data_to_int32(data);
-    if (index > priv->child_count) {
-        gf_log(this->name, GF_LOG_ERROR, "error with stripe-index xattr (%d)",
-               index);
-        local->op_ret = -1;
-        local->op_errno = EIO;
-        goto out;
-    }
-    if (local->fctx->xl_array) {
-        if (!local->fctx->xl_array[index])
-            local->fctx->xl_array[index] = prev->this;
-    }
-
-    sprintf(key, "trusted.%s.stripe-coalesce", this->name);
-    data = dict_get(dict, key);
-    if (!data) {
-        /*
-         * The file was probably created prior to coalesce support.
-         * Assume non-coalesce mode for this file to maintain backwards
-         * compatibility.
-         */
-        gf_log(this->name, GF_LOG_DEBUG,
-               "missing stripe-coalesce "
-               "attr, assume non-coalesce mode");
-        local->fctx->stripe_coalesce = 0;
-    } else {
-        local->fctx->stripe_coalesce = data_to_int32(data);
-    }
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_xattr_request_build(xlator_t *this, dict_t *dict, uint64_t stripe_size,
-                           uint32_t stripe_count, uint32_t stripe_index,
-                           uint32_t stripe_coalesce)
-{
-    char key[256] = {
-        0,
-    };
-    int32_t ret = -1;
-
-    sprintf(key, "trusted.%s.stripe-size", this->name);
-    ret = dict_set_int64(dict, key, stripe_size);
-    if (ret) {
-        gf_log(this->name, GF_LOG_WARNING, "failed to set %s in xattr_req dict",
-               key);
-        goto out;
-    }
-
-    sprintf(key, "trusted.%s.stripe-count", this->name);
-    ret = dict_set_int32(dict, key, stripe_count);
-    if (ret) {
-        gf_log(this->name, GF_LOG_WARNING, "failed to set %s in xattr_req dict",
-               key);
-        goto out;
-    }
-
-    sprintf(key, "trusted.%s.stripe-index", this->name);
-    ret = dict_set_int32(dict, key, stripe_index);
-    if (ret) {
-        gf_log(this->name, GF_LOG_WARNING, "failed to set %s in xattr_req dict",
-               key);
-        goto out;
-    }
-
-    sprintf(key, "trusted.%s.stripe-coalesce", this->name);
-    ret = dict_set_int32(dict, key, stripe_coalesce);
-    if (ret) {
-        gf_log(this->name, GF_LOG_WARNING, "failed to set %s in xattr_req_dict",
-               key);
-        goto out;
-    }
-out:
-    return ret;
-}
-
-static int
-set_default_block_size(stripe_private_t *priv, char *num)
-{
-    int ret = -1;
-    GF_VALIDATE_OR_GOTO("stripe", THIS, out);
-    GF_VALIDATE_OR_GOTO(THIS->name, priv, out);
-    GF_VALIDATE_OR_GOTO(THIS->name, num, out);
-
-    if (gf_string2bytesize_uint64(num, &priv->block_size) != 0) {
-        gf_log(THIS->name, GF_LOG_ERROR, "invalid number format \"%s\"", num);
-        goto out;
-    }
-
-    ret = 0;
-
-out:
-    return ret;
-}
-
-int
-set_stripe_block_size(xlator_t *this, stripe_private_t *priv, char *data)
-{
-    int ret = -1;
-    char *tmp_str = NULL;
-    char *tmp_str1 = NULL;
-    char *dup_str = NULL;
-    char *stripe_str = NULL;
-    char *pattern = NULL;
-    char *num = NULL;
-    struct stripe_options *temp_stripeopt = NULL;
-    struct stripe_options *stripe_opt = NULL;
-
-    if (!this || !priv || !data)
-        goto out;
-
-    /* Get the pattern for striping.
-       "option block-size *avi:10MB" etc */
-    stripe_str = strtok_r(data, ",", &tmp_str);
-    while (stripe_str) {
-        dup_str = gf_strdup(stripe_str);
-        stripe_opt = GF_CALLOC(1, sizeof(struct stripe_options),
-                               gf_stripe_mt_stripe_options);
-        if (!stripe_opt) {
-            goto out;
-        }
-
-        pattern = strtok_r(dup_str, ":", &tmp_str1);
-        num = strtok_r(NULL, ":", &tmp_str1);
-        if (!num) {
-            num = pattern;
-            pattern = "*";
-            ret = set_default_block_size(priv, num);
-            if (ret)
-                goto out;
-        }
-        if (gf_string2bytesize_uint64(num, &stripe_opt->block_size) != 0) {
-            gf_log(this->name, GF_LOG_ERROR, "invalid number format \"%s\"",
-                   num);
-            goto out;
-        }
-
-        if (stripe_opt->block_size < STRIPE_MIN_BLOCK_SIZE) {
-            gf_log(this->name, GF_LOG_ERROR,
-                   "Invalid Block-size: "
-                   "%s. Should be at least %llu bytes",
-                   num, STRIPE_MIN_BLOCK_SIZE);
-            goto out;
-        }
-        if (stripe_opt->block_size % 512) {
-            gf_log(this->name, GF_LOG_ERROR,
-                   "Block-size: %s should"
-                   " be a multiple of 512 bytes",
-                   num);
-            goto out;
-        }
-
-        memcpy(stripe_opt->path_pattern, pattern, strlen(pattern));
-
-        gf_log(this->name, GF_LOG_DEBUG,
-               "block-size : pattern %s : size %" PRId64,
-               stripe_opt->path_pattern, stripe_opt->block_size);
-
-        if (priv->pattern)
-            temp_stripeopt = NULL;
-        else
-            temp_stripeopt = priv->pattern;
-
-        stripe_opt->next = temp_stripeopt;
-
-        priv->pattern = stripe_opt;
-        stripe_opt = NULL;
-
-        GF_FREE(dup_str);
-        dup_str = NULL;
-
-        stripe_str = strtok_r(NULL, ",", &tmp_str);
-    }
-
-    ret = 0;
-out:
-
-    GF_FREE(dup_str);
-
-    GF_FREE(stripe_opt);
-
-    return ret;
-}
-
-int32_t
-stripe_iatt_merge(struct iatt *from, struct iatt *to)
-{
-    if (to->ia_size < from->ia_size)
-        to->ia_size = from->ia_size;
-    if (to->ia_mtime < from->ia_mtime)
-        to->ia_mtime = from->ia_mtime;
-    if (to->ia_ctime < from->ia_ctime)
-        to->ia_ctime = from->ia_ctime;
-    if (to->ia_atime < from->ia_atime)
-        to->ia_atime = from->ia_atime;
-    return 0;
-}
-
-off_t
-coalesced_offset(off_t offset, uint64_t stripe_size, int stripe_count)
-{
-    size_t line_size = 0;
-    uint64_t stripe_num = 0;
-    off_t coalesced_offset = 0;
-
-    line_size = stripe_size * stripe_count;
-    stripe_num = offset / line_size;
-
-    coalesced_offset = (stripe_num * stripe_size) + (offset % stripe_size);
-
-    return coalesced_offset;
-}
-
-off_t
-uncoalesced_size(off_t size, uint64_t stripe_size, int stripe_count,
-                 int stripe_index)
-{
-    uint64_t nr_full_stripe_chunks = 0, mod = 0;
-
-    if (!size)
-        return size;
-
-    /*
-     * Estimate the number of fully written stripes from the
-     * local file size. Each stripe_size chunk corresponds to
-     * a stripe.
-     */
-    nr_full_stripe_chunks = (size / stripe_size) * stripe_count;
-    mod = size % stripe_size;
-
-    if (!mod) {
-        /*
-         * There is no remainder, thus we could have overestimated
-         * the size of the file in terms of chunks. Trim the number
-         * of chunks by the following stripe members and leave it
-         * up to those nodes to respond with a larger size (if
-         * necessary).
-         */
-        nr_full_stripe_chunks -= stripe_count - (stripe_index + 1);
-        size = nr_full_stripe_chunks * stripe_size;
-    } else {
-        /*
-         * There is a remainder and thus we own the last chunk of the
-         * file. Add the preceding stripe members of the final stripe
-         * along with the remainder to calculate the exact size.
-         */
-        nr_full_stripe_chunks += stripe_index;
-        size = nr_full_stripe_chunks * stripe_size + mod;
-    }
-
-    return size;
-}
diff --git a/xlators/cluster/stripe/src/stripe-mem-types.h b/xlators/cluster/stripe/src/stripe-mem-types.h
deleted file mode 100644
index dcbef31212b..00000000000
--- a/xlators/cluster/stripe/src/stripe-mem-types.h
+++ /dev/null
@@ -1,29 +0,0 @@
-/*
-  Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#ifndef __STRIPE_MEM_TYPES_H__
-#define __STRIPE_MEM_TYPES_H__
-
-#include "mem-types.h"
-
-enum gf_stripe_mem_types_ {
-    gf_stripe_mt_iovec = gf_common_mt_end + 1,
-    gf_stripe_mt_stripe_replies,
-    gf_stripe_mt_stripe_fd_ctx_t,
-    gf_stripe_mt_char,
-    gf_stripe_mt_int8_t,
-    gf_stripe_mt_int32_t,
-    gf_stripe_mt_xlator_t,
-    gf_stripe_mt_stripe_private_t,
-    gf_stripe_mt_stripe_options,
-    gf_stripe_mt_xattr_sort_t,
-    gf_stripe_mt_end
-};
-#endif
diff --git a/xlators/cluster/stripe/src/stripe.c b/xlators/cluster/stripe/src/stripe.c
deleted file mode 100644
index 6010c1ed6c3..00000000000
--- a/xlators/cluster/stripe/src/stripe.c
+++ /dev/null
@@ -1,5612 +0,0 @@
-/*
-  Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-/**
- * xlators/cluster/stripe:
- *    Stripe translator, stripes the data across its child nodes,
- *    as per the options given in the volfile. The striping works
- *    fairly simple. It writes files at different offset as per
- *    calculation. So, 'ls -l' output at the real posix level will
- *    show file size bigger than the actual size. But when one does
- *    'df' or 'du <file>', real size of the file on the server is shown.
- *
- * WARNING:
- *  Stripe translator can't regenerate data if a child node gets disconnected.
- *  So, no 'self-heal' for stripe. Hence the advice, use stripe only when its
- *  very much necessary, or else, use it in combination with AFR, to have a
- *  backup copy.
- */
-#include <fnmatch.h>
-#include "stripe.h"
-#include "libxlator.h"
-#include "byte-order.h"
-#include "statedump.h"
-
-struct volume_options options[];
-
-int32_t
-stripe_sh_chown_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int32_t op_ret, int32_t op_errno, struct iatt *preop,
-                    struct iatt *postop, dict_t *xdata)
-{
-    int callcnt = -1;
-    stripe_local_t *local = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        STRIPE_STACK_DESTROY(frame);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_sh_make_entry_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                         int32_t op_ret, int32_t op_errno, inode_t *inode,
-                         struct iatt *buf, struct iatt *preparent,
-                         struct iatt *postparent, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!frame || !frame->local || !cookie || !this) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    STACK_WIND(frame, stripe_sh_chown_cbk, prev->this,
-               prev->this->fops->setattr, &local->loc, &local->stbuf,
-               (GF_SET_ATTR_UID | GF_SET_ATTR_GID), NULL);
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_entry_self_heal(call_frame_t *frame, xlator_t *this,
-                       stripe_local_t *local)
-{
-    xlator_list_t *trav = NULL;
-    call_frame_t *rframe = NULL;
-    stripe_local_t *rlocal = NULL;
-    stripe_private_t *priv = NULL;
-    dict_t *xdata = NULL;
-    int ret = 0;
-
-    if (!local || !this || !frame) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    if (!(IA_ISREG(local->stbuf.ia_type) || IA_ISDIR(local->stbuf.ia_type)))
-        return 0;
-
-    priv = this->private;
-    trav = this->children;
-    rframe = copy_frame(frame);
-    if (!rframe) {
-        goto out;
-    }
-    rlocal = mem_get0(this->local_pool);
-    if (!rlocal) {
-        goto out;
-    }
-    rframe->local = rlocal;
-    rlocal->call_count = priv->child_count;
-    loc_copy(&rlocal->loc, &local->loc);
-    memcpy(&rlocal->stbuf, &local->stbuf, sizeof(struct iatt));
-
-    xdata = dict_new();
-    if (!xdata)
-        goto out;
-
-    ret = dict_set_gfuuid(xdata, "gfid-req", local->stbuf.ia_gfid, true);
-    if (ret)
-        gf_log(this->name, GF_LOG_WARNING, "%s: failed to set gfid-req",
-               local->loc.path);
-
-    while (trav) {
-        if (IA_ISREG(local->stbuf.ia_type)) {
-            STACK_WIND(
-                rframe, stripe_sh_make_entry_cbk, trav->xlator,
-                trav->xlator->fops->mknod, &local->loc,
-                st_mode_from_ia(local->stbuf.ia_prot, local->stbuf.ia_type), 0,
-                0, xdata);
-        }
-        if (IA_ISDIR(local->stbuf.ia_type)) {
-            STACK_WIND(
-                rframe, stripe_sh_make_entry_cbk, trav->xlator,
-                trav->xlator->fops->mkdir, &local->loc,
-                st_mode_from_ia(local->stbuf.ia_prot, local->stbuf.ia_type), 0,
-                xdata);
-        }
-        trav = trav->next;
-    }
-
-    if (xdata)
-        dict_unref(xdata);
-    return 0;
-
-out:
-    if (rframe)
-        STRIPE_STACK_DESTROY(rframe);
-
-    return 0;
-}
-
-int32_t
-stripe_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, inode_t *inode,
-                  struct iatt *buf, dict_t *xdata, struct iatt *postparent)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-    int ret = 0;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            if ((op_errno != ENOENT) && (op_errno != ESTALE))
-                gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                       prev->this->name, strerror(op_errno));
-            if (local->op_errno != ESTALE)
-                local->op_errno = op_errno;
-            if (((op_errno != ENOENT) && (op_errno != ENOTCONN) &&
-                 (op_errno != ESTALE)) ||
-                (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-            if (op_errno == ENOENT)
-                local->entry_self_heal_needed = 1;
-        }
-
-        if (op_ret >= 0) {
-            local->op_ret = 0;
-            if (IA_ISREG(buf->ia_type)) {
-                ret = stripe_ctx_handle(this, prev, local, xdata);
-                if (ret)
-                    gf_log(this->name, GF_LOG_ERROR,
-                           "Error getting fctx info from"
-                           " dict");
-            }
-
-            if (FIRST_CHILD(this) == prev->this) {
-                local->stbuf = *buf;
-                local->postparent = *postparent;
-                local->inode = inode_ref(inode);
-                if (xdata)
-                    local->xdata = dict_ref(xdata);
-                if (local->xattr) {
-                    stripe_aggregate_xattr(local->xdata, local->xattr);
-                    dict_unref(local->xattr);
-                    local->xattr = NULL;
-                }
-            }
-
-            if (!local->xdata && !local->xattr) {
-                local->xattr = dict_ref(xdata);
-            } else if (local->xdata) {
-                stripe_aggregate_xattr(local->xdata, xdata);
-            } else if (local->xattr) {
-                stripe_aggregate_xattr(local->xattr, xdata);
-            }
-
-            local->stbuf_blocks += buf->ia_blocks;
-            local->postparent_blocks += postparent->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-            if (local->postparent_size < postparent->ia_size)
-                local->postparent_size = postparent->ia_size;
-
-            if (gf_uuid_is_null(local->ia_gfid))
-                gf_uuid_copy(local->ia_gfid, buf->ia_gfid);
-
-            /* Make sure the gfid on all the nodes are same */
-            if (gf_uuid_compare(local->ia_gfid, buf->ia_gfid)) {
-                gf_log(this->name, GF_LOG_WARNING,
-                       "%s: gfid different on subvolume %s", local->loc.path,
-                       prev->this->name);
-            }
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->op_ret == 0 && local->entry_self_heal_needed &&
-            !gf_uuid_is_null(local->loc.inode->gfid))
-            stripe_entry_self_heal(frame, this, local);
-
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-            local->stbuf.ia_size = local->stbuf_size;
-            local->postparent.ia_blocks = local->postparent_blocks;
-            local->postparent.ia_size = local->postparent_size;
-            inode_ctx_put(local->inode, this, (uint64_t)(long)local->fctx);
-        }
-
-        STRIPE_STACK_UNWIND(lookup, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, local->xdata,
-                            &local->postparent);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_lookup(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-    int64_t filesize = 0;
-    int ret = 0;
-    uint64_t tmpctx = 0;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    loc_copy(&local->loc, loc);
-
-    inode_ctx_get(local->inode, this, &tmpctx);
-    if (tmpctx)
-        local->fctx = (stripe_fd_ctx_t *)(long)tmpctx;
-
-    /* quick-read friendly changes */
-    if (xdata && dict_get(xdata, GF_CONTENT_KEY)) {
-        ret = dict_get_int64(xdata, GF_CONTENT_KEY, &filesize);
-        if (!ret && (filesize > priv->block_size))
-            dict_del(xdata, GF_CONTENT_KEY);
-    }
-
-    /* get stripe-size xattr on lookup. This would be required for
-     * open/read/write/pathinfo calls. Hence we send down the request
-     * even when type == IA_INVAL */
-
-    /*
-     * We aren't guaranteed to have xdata here. We need the format info for
-     * the file, so allocate xdata if necessary.
-     */
-    if (!xdata)
-        xdata = dict_new();
-    else
-        xdata = dict_ref(xdata);
-
-    if (xdata &&
-        (IA_ISREG(loc->inode->ia_type) || (loc->inode->ia_type == IA_INVAL))) {
-        ret = stripe_xattr_request_build(this, xdata, 8, 4, 4, 0);
-        if (ret)
-            gf_log(this->name, GF_LOG_ERROR,
-                   "Failed to build"
-                   " xattr request for %s",
-                   loc->path);
-    }
-
-    /* Every time in stripe lookup, all child nodes
-       should be looked up */
-    local->call_count = priv->child_count;
-    while (trav) {
-        STACK_WIND(frame, stripe_lookup_cbk, trav->xlator,
-                   trav->xlator->fops->lookup, loc, xdata);
-        trav = trav->next;
-    }
-
-    dict_unref(xdata);
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(lookup, frame, -1, op_errno, NULL, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_stat_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                int32_t op_ret, int32_t op_errno, struct iatt *buf,
-                dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret == 0) {
-            local->op_ret = 0;
-
-            if (FIRST_CHILD(this) == prev->this) {
-                local->stbuf = *buf;
-            }
-
-            local->stbuf_blocks += buf->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-        }
-
-        STRIPE_STACK_UNWIND(stat, frame, local->op_ret, local->op_errno,
-                            &local->stbuf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_stat(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    if (IA_ISREG(loc->inode->ia_type)) {
-        inode_ctx_get(loc->inode, this, (uint64_t *)&fctx);
-        if (!fctx)
-            goto err;
-        local->fctx = fctx;
-    }
-
-    while (trav) {
-        STACK_WIND(frame, stripe_stat_cbk, trav->xlator,
-                   trav->xlator->fops->stat, loc, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-
-err:
-    STRIPE_STACK_UNWIND(stat, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_statfs_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, struct statvfs *stbuf,
-                  dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    int32_t callcnt = 0;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret && (op_errno != ENOTCONN)) {
-            local->op_errno = op_errno;
-        }
-        if (op_ret == 0) {
-            struct statvfs *dict_buf = &local->statvfs_buf;
-            dict_buf->f_bsize = stbuf->f_bsize;
-            dict_buf->f_frsize = stbuf->f_frsize;
-            dict_buf->f_blocks += stbuf->f_blocks;
-            dict_buf->f_bfree += stbuf->f_bfree;
-            dict_buf->f_bavail += stbuf->f_bavail;
-            dict_buf->f_files += stbuf->f_files;
-            dict_buf->f_ffree += stbuf->f_ffree;
-            dict_buf->f_favail += stbuf->f_favail;
-            dict_buf->f_fsid = stbuf->f_fsid;
-            dict_buf->f_flag = stbuf->f_flag;
-            dict_buf->f_namemax = stbuf->f_namemax;
-            local->op_ret = 0;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        STRIPE_STACK_UNWIND(statfs, frame, local->op_ret, local->op_errno,
-                            &local->statvfs_buf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-
-    trav = this->children;
-    priv = this->private;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    local->op_errno = ENOTCONN;
-    frame->local = local;
-
-    local->call_count = priv->child_count;
-    while (trav) {
-        STACK_WIND(frame, stripe_statfs_cbk, trav->xlator,
-                   trav->xlator->fops->statfs, loc, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(statfs, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_truncate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                    struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret == 0) {
-            local->op_ret = 0;
-            if (FIRST_CHILD(this) == prev->this) {
-                local->pre_buf = *prebuf;
-                local->post_buf = *postbuf;
-            }
-
-            local->prebuf_blocks += prebuf->ia_blocks;
-            local->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, local->fctx, prev);
-            correct_file_size(postbuf, local->fctx, prev);
-
-            if (local->prebuf_size < prebuf->ia_size)
-                local->prebuf_size = prebuf->ia_size;
-
-            if (local->postbuf_size < postbuf->ia_size)
-                local->postbuf_size = postbuf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->pre_buf.ia_blocks = local->prebuf_blocks;
-            local->pre_buf.ia_size = local->prebuf_size;
-            local->post_buf.ia_blocks = local->postbuf_blocks;
-            local->post_buf.ia_size = local->postbuf_size;
-        }
-
-        STRIPE_STACK_UNWIND(truncate, frame, local->op_ret, local->op_errno,
-                            &local->pre_buf, &local->post_buf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_truncate(call_frame_t *frame, xlator_t *this, loc_t *loc, off_t offset,
-                dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = EINVAL;
-    int i, eof_idx;
-    off_t dest_offset, tmp_offset;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    inode_ctx_get(loc->inode, this, (uint64_t *)&fctx);
-    if (!fctx) {
-        gf_log(this->name, GF_LOG_ERROR, "no stripe context");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    local->fctx = fctx;
-    eof_idx = (offset / fctx->stripe_size) % fctx->stripe_count;
-
-    for (i = 0; i < fctx->stripe_count; i++) {
-        if (!fctx->xl_array[i]) {
-            gf_log(this->name, GF_LOG_ERROR, "no xlator at index %d", i);
-            op_errno = EINVAL;
-            goto err;
-        }
-
-        if (fctx->stripe_coalesce) {
-            /*
-             * The node that owns EOF is truncated to the exact
-             * coalesced offset. Nodes prior to this index should
-             * be rounded up to the size of the complete stripe,
-             * while nodes after this index should be rounded down
-             * to the size of the previous stripe.
-             */
-            if (i < eof_idx)
-                tmp_offset = gf_roof(offset,
-                                     fctx->stripe_size * fctx->stripe_count);
-            else if (i > eof_idx)
-                tmp_offset = gf_floor(offset,
-                                      fctx->stripe_size * fctx->stripe_count);
-            else
-                tmp_offset = offset;
-
-            dest_offset = coalesced_offset(tmp_offset, fctx->stripe_size,
-                                           fctx->stripe_count);
-        } else {
-            dest_offset = offset;
-        }
-
-        STACK_WIND(frame, stripe_truncate_cbk, fctx->xl_array[i],
-                   fctx->xl_array[i]->fops->truncate, loc, dest_offset, NULL);
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(truncate, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_setattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                   int32_t op_ret, int32_t op_errno, struct iatt *preop,
-                   struct iatt *postop, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret == 0) {
-            local->op_ret = 0;
-
-            if (FIRST_CHILD(this) == prev->this) {
-                local->pre_buf = *preop;
-                local->post_buf = *postop;
-            }
-
-            local->prebuf_blocks += preop->ia_blocks;
-            local->postbuf_blocks += postop->ia_blocks;
-
-            correct_file_size(preop, local->fctx, prev);
-            correct_file_size(postop, local->fctx, prev);
-
-            if (local->prebuf_size < preop->ia_size)
-                local->prebuf_size = preop->ia_size;
-            if (local->postbuf_size < postop->ia_size)
-                local->postbuf_size = postop->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->pre_buf.ia_blocks = local->prebuf_blocks;
-            local->pre_buf.ia_size = local->prebuf_size;
-            local->post_buf.ia_blocks = local->postbuf_blocks;
-            local->post_buf.ia_size = local->postbuf_size;
-        }
-
-        STRIPE_STACK_UNWIND(setattr, frame, local->op_ret, local->op_errno,
-                            &local->pre_buf, &local->post_buf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_setattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
-               struct iatt *stbuf, int32_t valid, dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    if (!IA_ISDIR(loc->inode->ia_type) && !IA_ISREG(loc->inode->ia_type)) {
-        local->call_count = 1;
-        STACK_WIND(frame, stripe_setattr_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->setattr, loc, stbuf, valid, NULL);
-        return 0;
-    }
-
-    if (IA_ISREG(loc->inode->ia_type)) {
-        inode_ctx_get(loc->inode, this, (uint64_t *)&fctx);
-        if (!fctx)
-            goto err;
-        local->fctx = fctx;
-    }
-
-    local->call_count = priv->child_count;
-    while (trav) {
-        STACK_WIND(frame, stripe_setattr_cbk, trav->xlator,
-                   trav->xlator->fops->setattr, loc, stbuf, valid, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(setattr, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_fsetattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                struct iatt *stbuf, int32_t valid, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_setattr_cbk, trav->xlator,
-                   trav->xlator->fops->fsetattr, fd, stbuf, valid, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fsetattr, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_stack_rename_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno, struct iatt *buf,
-                        struct iatt *preoldparent, struct iatt *postoldparent,
-                        struct iatt *prenewparent, struct iatt *postnewparent,
-                        dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret == 0) {
-            local->op_ret = 0;
-
-            local->stbuf.ia_blocks += buf->ia_blocks;
-            local->preparent.ia_blocks += preoldparent->ia_blocks;
-            local->postparent.ia_blocks += postoldparent->ia_blocks;
-            local->pre_buf.ia_blocks += prenewparent->ia_blocks;
-            local->post_buf.ia_blocks += postnewparent->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf.ia_size < buf->ia_size)
-                local->stbuf.ia_size = buf->ia_size;
-
-            if (local->preparent.ia_size < preoldparent->ia_size)
-                local->preparent.ia_size = preoldparent->ia_size;
-
-            if (local->postparent.ia_size < postoldparent->ia_size)
-                local->postparent.ia_size = postoldparent->ia_size;
-
-            if (local->pre_buf.ia_size < prenewparent->ia_size)
-                local->pre_buf.ia_size = prenewparent->ia_size;
-
-            if (local->post_buf.ia_size < postnewparent->ia_size)
-                local->post_buf.ia_size = postnewparent->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        STRIPE_STACK_UNWIND(rename, frame, local->op_ret, local->op_errno,
-                            &local->stbuf, &local->preparent,
-                            &local->postparent, &local->pre_buf,
-                            &local->post_buf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_first_rename_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno, struct iatt *buf,
-                        struct iatt *preoldparent, struct iatt *postoldparent,
-                        struct iatt *prenewparent, struct iatt *postnewparent,
-                        dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        op_errno = EINVAL;
-        goto unwind;
-    }
-
-    if (op_ret == -1) {
-        goto unwind;
-    }
-
-    local = frame->local;
-    trav = this->children;
-
-    local->stbuf = *buf;
-    local->preparent = *preoldparent;
-    local->postparent = *postoldparent;
-    local->pre_buf = *prenewparent;
-    local->post_buf = *postnewparent;
-
-    local->op_ret = 0;
-    local->call_count--;
-
-    trav = trav->next; /* Skip first child */
-    while (trav) {
-        STACK_WIND(frame, stripe_stack_rename_cbk, trav->xlator,
-                   trav->xlator->fops->rename, &local->loc, &local->loc2, NULL);
-        trav = trav->next;
-    }
-    return 0;
-
-unwind:
-    STRIPE_STACK_UNWIND(rename, frame, -1, op_errno, buf, preoldparent,
-                        postoldparent, prenewparent, postnewparent, NULL);
-    return 0;
-}
-
-int32_t
-stripe_rename(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
-              dict_t *xdata)
-{
-    stripe_private_t *priv = NULL;
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(oldloc, err);
-    VALIDATE_OR_GOTO(oldloc->path, err);
-    VALIDATE_OR_GOTO(oldloc->inode, err);
-    VALIDATE_OR_GOTO(newloc, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* If any one node is down, don't allow rename */
-    if (priv->nodes_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    frame->local = local;
-
-    local->op_ret = -1;
-    loc_copy(&local->loc, oldloc);
-    loc_copy(&local->loc2, newloc);
-
-    local->call_count = priv->child_count;
-
-    if (IA_ISREG(oldloc->inode->ia_type)) {
-        inode_ctx_get(oldloc->inode, this, (uint64_t *)&fctx);
-        if (!fctx)
-            goto err;
-        local->fctx = fctx;
-    }
-
-    STACK_WIND(frame, stripe_first_rename_cbk, trav->xlator,
-               trav->xlator->fops->rename, oldloc, newloc, NULL);
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(rename, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL, NULL);
-    return 0;
-}
-int32_t
-stripe_first_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno,
-                        struct iatt *preparent, struct iatt *postparent,
-                        dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    if (op_ret == -1) {
-        gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-               strerror(op_errno));
-        goto out;
-    }
-    local->op_ret = 0;
-    local->preparent = *preparent;
-    local->postparent = *postparent;
-    local->preparent_blocks += preparent->ia_blocks;
-    local->postparent_blocks += postparent->ia_blocks;
-
-    STRIPE_STACK_UNWIND(unlink, frame, local->op_ret, local->op_errno,
-                        &local->preparent, &local->postparent, xdata);
-    return 0;
-out:
-    STRIPE_STACK_UNWIND(unlink, frame, -1, op_errno, NULL, NULL, NULL);
-
-    return 0;
-}
-
-int32_t
-stripe_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, struct iatt *preparent,
-                  struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-                   strerror(op_errno));
-            local->op_errno = op_errno;
-            if (op_errno != ENOENT) {
-                local->failed = 1;
-                local->op_ret = op_ret;
-            }
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (callcnt == 1) {
-        if (local->failed) {
-            op_errno = local->op_errno;
-            goto out;
-        }
-        STACK_WIND(frame, stripe_first_unlink_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->unlink, &local->loc, local->xflag,
-                   local->xdata);
-    }
-    return 0;
-out:
-    STRIPE_STACK_UNWIND(unlink, frame, -1, op_errno, NULL, NULL, NULL);
-
-    return 0;
-}
-
-int32_t
-stripe_unlink(call_frame_t *frame, xlator_t *this, loc_t *loc, int xflag,
-              dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Don't unlink a file if a node is down */
-    if (priv->nodes_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    loc_copy(&local->loc, loc);
-    local->xflag = xflag;
-
-    if (xdata)
-        local->xdata = dict_ref(xdata);
-
-    frame->local = local;
-    local->call_count = priv->child_count;
-    trav = trav->next; /* Skip the first child */
-
-    while (trav) {
-        STACK_WIND(frame, stripe_unlink_cbk, trav->xlator,
-                   trav->xlator->fops->unlink, loc, xflag, xdata);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(unlink, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_first_rmdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, struct iatt *preparent,
-                       struct iatt *postparent, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    if (op_ret == -1) {
-        goto err;
-    }
-
-    local = frame->local;
-    local->op_ret = 0;
-
-    local->call_count--; /* First child successful */
-
-    local->preparent = *preparent;
-    local->postparent = *postparent;
-    local->preparent_size = preparent->ia_size;
-    local->postparent_size = postparent->ia_size;
-    local->preparent_blocks += preparent->ia_blocks;
-    local->postparent_blocks += postparent->ia_blocks;
-
-    STRIPE_STACK_UNWIND(rmdir, frame, local->op_ret, local->op_errno,
-                        &local->preparent, &local->postparent, xdata);
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(rmdir, frame, op_ret, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_rmdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, struct iatt *preparent,
-                 struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-                   strerror(op_errno));
-            if (op_errno != ENOENT)
-                local->failed = 1;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (callcnt == 1) {
-        if (local->failed)
-            goto out;
-        STACK_WIND(frame, stripe_first_rmdir_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->rmdir, &local->loc, local->flags,
-                   NULL);
-    }
-    return 0;
-out:
-    STRIPE_STACK_UNWIND(rmdir, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_rmdir(call_frame_t *frame, xlator_t *this, loc_t *loc, int flags,
-             dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* don't delete a directory if any of the subvolume is down */
-    if (priv->nodes_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    loc_copy(&local->loc, loc);
-    local->flags = flags;
-    local->call_count = priv->child_count;
-    trav = trav->next; /* skip the first child */
-
-    while (trav) {
-        STACK_WIND(frame, stripe_rmdir_cbk, trav->xlator,
-                   trav->xlator->fops->rmdir, loc, flags, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(rmdir, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_mknod_ifreg_fail_unlink_cbk(call_frame_t *frame, void *cookie,
-                                   xlator_t *this, int32_t op_ret,
-                                   int32_t op_errno, struct iatt *preparent,
-                                   struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        STRIPE_STACK_UNWIND(mknod, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, &local->preparent,
-                            &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-/**
- */
-int32_t
-stripe_mknod_ifreg_setxattr_cbk(call_frame_t *frame, void *cookie,
-                                xlator_t *this, int32_t op_ret,
-                                int32_t op_errno, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    priv = this->private;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_ret = -1;
-            local->op_errno = op_errno;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->op_ret == -1) {
-            local->call_count = priv->child_count;
-            while (trav) {
-                STACK_WIND(frame, stripe_mknod_ifreg_fail_unlink_cbk,
-                           trav->xlator, trav->xlator->fops->unlink,
-                           &local->loc, 0, NULL);
-                trav = trav->next;
-            }
-            return 0;
-        }
-
-        STRIPE_STACK_UNWIND(mknod, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, &local->preparent,
-                            &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_mknod_ifreg_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, inode_t *inode,
-                       struct iatt *buf, struct iatt *preparent,
-                       struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    call_frame_t *prev = NULL;
-    xlator_list_t *trav = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    priv = this->private;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-            local->op_errno = op_errno;
-        }
-        if (op_ret >= 0) {
-            local->op_ret = op_ret;
-
-            /* Can be used as a mechanism to understand if mknod
-               was successful in at least one place */
-            if (gf_uuid_is_null(local->ia_gfid))
-                gf_uuid_copy(local->ia_gfid, buf->ia_gfid);
-
-            if (stripe_ctx_handle(this, prev, local, xdata))
-                gf_log(this->name, GF_LOG_ERROR,
-                       "Error getting fctx info from dict");
-
-            local->stbuf_blocks += buf->ia_blocks;
-            local->preparent_blocks += preparent->ia_blocks;
-            local->postparent_blocks += postparent->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-            if (local->preparent_size < preparent->ia_size)
-                local->preparent_size = preparent->ia_size;
-            if (local->postparent_size < postparent->ia_size)
-                local->postparent_size = postparent->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if ((local->op_ret == -1) && !gf_uuid_is_null(local->ia_gfid)) {
-            /* ia_gfid set means, at least on one node 'mknod'
-               is successful */
-            local->call_count = priv->child_count;
-            trav = this->children;
-            while (trav) {
-                STACK_WIND(frame, stripe_mknod_ifreg_fail_unlink_cbk,
-                           trav->xlator, trav->xlator->fops->unlink,
-                           &local->loc, 0, NULL);
-                trav = trav->next;
-            }
-            return 0;
-        }
-
-        if (local->op_ret != -1) {
-            local->preparent.ia_blocks = local->preparent_blocks;
-            local->preparent.ia_size = local->preparent_size;
-            local->postparent.ia_blocks = local->postparent_blocks;
-            local->postparent.ia_size = local->postparent_size;
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-            inode_ctx_put(local->inode, this, (uint64_t)(long)local->fctx);
-        }
-        STRIPE_STACK_UNWIND(mknod, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, &local->preparent,
-                            &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_mknod_first_ifreg_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                             int32_t op_ret, int32_t op_errno, inode_t *inode,
-                             struct iatt *buf, struct iatt *preparent,
-                             struct iatt *postparent, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    call_frame_t *prev = NULL;
-    xlator_list_t *trav = NULL;
-    int i = 1;
-    dict_t *dict = NULL;
-    int ret = 0;
-    int need_unref = 0;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    priv = this->private;
-    local = frame->local;
-    trav = this->children;
-
-    local->call_count--;
-
-    if (op_ret == -1) {
-        gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-               prev->this->name, strerror(op_errno));
-        local->failed = 1;
-        local->op_errno = op_errno;
-        goto out;
-    }
-
-    local->op_ret = op_ret;
-
-    local->stbuf = *buf;
-    local->preparent = *preparent;
-    local->postparent = *postparent;
-
-    if (gf_uuid_is_null(local->ia_gfid))
-        gf_uuid_copy(local->ia_gfid, buf->ia_gfid);
-    local->preparent.ia_blocks = local->preparent_blocks;
-    local->preparent.ia_size = local->preparent_size;
-    local->postparent.ia_blocks = local->postparent_blocks;
-    local->postparent.ia_size = local->postparent_size;
-    local->stbuf.ia_size = local->stbuf_size;
-    local->stbuf.ia_blocks = local->stbuf_blocks;
-
-    trav = trav->next;
-    while (trav) {
-        if (priv->xattr_supported) {
-            dict = dict_new();
-            if (!dict) {
-                gf_log(this->name, GF_LOG_ERROR, "failed to allocate dict %s",
-                       local->loc.path);
-            }
-            need_unref = 1;
-
-            dict_copy(local->xattr, dict);
-
-            ret = stripe_xattr_request_build(this, dict, local->stripe_size,
-                                             priv->child_count, i,
-                                             priv->coalesce);
-            if (ret)
-                gf_log(this->name, GF_LOG_ERROR,
-                       "Failed to build xattr request");
-
-        } else {
-            dict = local->xattr;
-        }
-
-        STACK_WIND(frame, stripe_mknod_ifreg_cbk, trav->xlator,
-                   trav->xlator->fops->mknod, &local->loc, local->mode,
-                   local->rdev, 0, dict);
-        trav = trav->next;
-        i++;
-
-        if (dict && need_unref)
-            dict_unref(dict);
-    }
-
-    return 0;
-
-out:
-
-    STRIPE_STACK_UNWIND(mknod, frame, op_ret, op_errno, NULL, NULL, NULL, NULL,
-                        NULL);
-    return 0;
-}
-
-int32_t
-stripe_single_mknod_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno, inode_t *inode,
-                        struct iatt *buf, struct iatt *preparent,
-                        struct iatt *postparent, dict_t *xdata)
-{
-    STRIPE_STACK_UNWIND(mknod, frame, op_ret, op_errno, inode, buf, preparent,
-                        postparent, xdata);
-    return 0;
-}
-
-int
-stripe_mknod(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
-             dev_t rdev, mode_t umask, dict_t *xdata)
-{
-    stripe_private_t *priv = NULL;
-    stripe_local_t *local = NULL;
-    int32_t op_errno = EINVAL;
-    int32_t i = 0;
-    dict_t *dict = NULL;
-    int ret = 0;
-    int need_unref = 0;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    if (S_ISREG(mode)) {
-        /* NOTE: on older kernels (older than 2.6.9),
-           creat() fops is sent as mknod() + open(). Hence handling
-           S_IFREG files is necessary */
-        if (priv->nodes_down) {
-            gf_log(this->name, GF_LOG_WARNING, "Some node down, returning EIO");
-            op_errno = EIO;
-            goto err;
-        }
-
-        /* Initialization */
-        local = mem_get0(this->local_pool);
-        if (!local) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        local->op_ret = -1;
-        local->op_errno = ENOTCONN;
-        local->stripe_size = stripe_get_matching_bs(loc->path, priv);
-        frame->local = local;
-        local->inode = inode_ref(loc->inode);
-        loc_copy(&local->loc, loc);
-        local->xattr = dict_copy_with_ref(xdata, NULL);
-        local->mode = mode;
-        local->umask = umask;
-        local->rdev = rdev;
-
-        /* Every time in stripe lookup, all child nodes should
-           be looked up */
-        local->call_count = priv->child_count;
-
-        if (priv->xattr_supported) {
-            dict = dict_new();
-            if (!dict) {
-                gf_log(this->name, GF_LOG_ERROR, "failed to allocate dict %s",
-                       loc->path);
-            }
-            need_unref = 1;
-
-            dict_copy(xdata, dict);
-
-            ret = stripe_xattr_request_build(this, dict, local->stripe_size,
-                                             priv->child_count, i,
-                                             priv->coalesce);
-            if (ret)
-                gf_log(this->name, GF_LOG_ERROR,
-                       "failed to build xattr request");
-        } else {
-            dict = xdata;
-        }
-
-        STACK_WIND(frame, stripe_mknod_first_ifreg_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->mknod, loc, mode, rdev, umask,
-                   dict);
-
-        if (dict && need_unref)
-            dict_unref(dict);
-        return 0;
-    }
-
-    STACK_WIND(frame, stripe_single_mknod_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->mknod, loc, mode, rdev, umask, xdata);
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(mknod, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL);
-    return 0;
-}
-
-int32_t
-stripe_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, inode_t *inode,
-                 struct iatt *buf, struct iatt *preparent,
-                 struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret >= 0) {
-            local->op_ret = 0;
-
-            local->stbuf_blocks += buf->ia_blocks;
-            local->preparent_blocks += preparent->ia_blocks;
-            local->postparent_blocks += postparent->ia_blocks;
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-            if (local->preparent_size < preparent->ia_size)
-                local->preparent_size = preparent->ia_size;
-            if (local->postparent_size < postparent->ia_size)
-                local->postparent_size = postparent->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed != -1) {
-            local->preparent.ia_blocks = local->preparent_blocks;
-            local->preparent.ia_size = local->preparent_size;
-            local->postparent.ia_blocks = local->postparent_blocks;
-            local->postparent.ia_size = local->postparent_size;
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-        }
-        STRIPE_STACK_UNWIND(mkdir, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, &local->preparent,
-                            &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_first_mkdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, inode_t *inode,
-                       struct iatt *buf, struct iatt *preparent,
-                       struct iatt *postparent, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-    xlator_list_t *trav = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-    trav = this->children;
-
-    local->call_count--; /* first child is successful */
-    trav = trav->next;   /* skip first child */
-
-    if (op_ret == -1) {
-        gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-               prev->this->name, strerror(op_errno));
-        local->op_errno = op_errno;
-        goto out;
-    }
-
-    local->op_ret = 0;
-
-    local->inode = inode_ref(inode);
-    local->stbuf = *buf;
-    local->postparent = *postparent;
-    local->preparent = *preparent;
-
-    local->stbuf_blocks += buf->ia_blocks;
-    local->preparent_blocks += preparent->ia_blocks;
-    local->postparent_blocks += postparent->ia_blocks;
-
-    local->stbuf_size = buf->ia_size;
-    local->preparent_size = preparent->ia_size;
-    local->postparent_size = postparent->ia_size;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_mkdir_cbk, trav->xlator,
-                   trav->xlator->fops->mkdir, &local->loc, local->mode,
-                   local->umask, local->xdata);
-        trav = trav->next;
-    }
-    return 0;
-out:
-    STRIPE_STACK_UNWIND(mkdir, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL);
-
-    return 0;
-}
-
-int
-stripe_mkdir(call_frame_t *frame, xlator_t *this, loc_t *loc, mode_t mode,
-             mode_t umask, dict_t *xdata)
-{
-    stripe_private_t *priv = NULL;
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    local->call_count = priv->child_count;
-    if (xdata)
-        local->xdata = dict_ref(xdata);
-    local->mode = mode;
-    local->umask = umask;
-    loc_copy(&local->loc, loc);
-    frame->local = local;
-
-    /* Every time in stripe lookup, all child nodes should be looked up */
-    STACK_WIND(frame, stripe_first_mkdir_cbk, trav->xlator,
-               trav->xlator->fops->mkdir, loc, mode, umask, xdata);
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(mkdir, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL);
-    return 0;
-}
-
-int32_t
-stripe_link_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                int32_t op_ret, int32_t op_errno, inode_t *inode,
-                struct iatt *buf, struct iatt *preparent,
-                struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret >= 0) {
-            local->op_ret = 0;
-
-            if (IA_ISREG(inode->ia_type)) {
-                inode_ctx_get(inode, this, (uint64_t *)&fctx);
-                if (!fctx) {
-                    gf_log(this->name, GF_LOG_ERROR,
-                           "failed to get stripe context");
-                    op_ret = -1;
-                    op_errno = EINVAL;
-                }
-            }
-
-            if (FIRST_CHILD(this) == prev->this) {
-                local->inode = inode_ref(inode);
-                local->stbuf = *buf;
-                local->postparent = *postparent;
-                local->preparent = *preparent;
-            }
-            local->stbuf_blocks += buf->ia_blocks;
-            local->preparent_blocks += preparent->ia_blocks;
-            local->postparent_blocks += postparent->ia_blocks;
-
-            correct_file_size(buf, fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-            if (local->preparent_size < preparent->ia_size)
-                local->preparent_size = preparent->ia_size;
-            if (local->postparent_size < postparent->ia_size)
-                local->postparent_size = postparent->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->preparent.ia_blocks = local->preparent_blocks;
-            local->preparent.ia_size = local->preparent_size;
-            local->postparent.ia_blocks = local->postparent_blocks;
-            local->postparent.ia_size = local->postparent_size;
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-        }
-        STRIPE_STACK_UNWIND(link, frame, local->op_ret, local->op_errno,
-                            local->inode, &local->stbuf, &local->preparent,
-                            &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_link(call_frame_t *frame, xlator_t *this, loc_t *oldloc, loc_t *newloc,
-            dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(oldloc, err);
-    VALIDATE_OR_GOTO(oldloc->path, err);
-    VALIDATE_OR_GOTO(oldloc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* If any one node is down, don't allow link operation */
-    if (priv->nodes_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    /* Every time in stripe lookup, all child
-       nodes should be looked up */
-    while (trav) {
-        STACK_WIND(frame, stripe_link_cbk, trav->xlator,
-                   trav->xlator->fops->link, oldloc, newloc, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(link, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL);
-    return 0;
-}
-
-int32_t
-stripe_create_fail_unlink_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                              int32_t op_ret, int32_t op_errno,
-                              struct iatt *preparent, struct iatt *postparent,
-                              dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        STRIPE_STACK_UNWIND(create, frame, local->op_ret, local->op_errno,
-                            local->fd, local->inode, &local->stbuf,
-                            &local->preparent, &local->postparent, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, fd_t *fd, inode_t *inode,
-                  struct iatt *buf, struct iatt *preparent,
-                  struct iatt *postparent, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    call_frame_t *prev = NULL;
-    xlator_list_t *trav = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    priv = this->private;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->failed = 1;
-            local->op_errno = op_errno;
-        }
-
-        if (op_ret >= 0) {
-            if (IA_ISREG(buf->ia_type)) {
-                if (stripe_ctx_handle(this, prev, local, xdata))
-                    gf_log(this->name, GF_LOG_ERROR,
-                           "Error getting fctx info from "
-                           "dict");
-            }
-
-            local->op_ret = op_ret;
-
-            local->stbuf_blocks += buf->ia_blocks;
-            local->preparent_blocks += preparent->ia_blocks;
-            local->postparent_blocks += postparent->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-            if (local->preparent_size < preparent->ia_size)
-                local->preparent_size = preparent->ia_size;
-            if (local->postparent_size < postparent->ia_size)
-                local->postparent_size = postparent->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret == -1) {
-            local->call_count = priv->child_count;
-            trav = this->children;
-            while (trav) {
-                STACK_WIND(frame, stripe_create_fail_unlink_cbk, trav->xlator,
-                           trav->xlator->fops->unlink, &local->loc, 0, NULL);
-                trav = trav->next;
-            }
-
-            return 0;
-        }
-
-        if (local->op_ret >= 0) {
-            local->preparent.ia_blocks = local->preparent_blocks;
-            local->preparent.ia_size = local->preparent_size;
-            local->postparent.ia_blocks = local->postparent_blocks;
-            local->postparent.ia_size = local->postparent_size;
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-
-            stripe_copy_xl_array(local->fctx->xl_array, priv->xl_array,
-                                 local->fctx->stripe_count);
-            inode_ctx_put(local->inode, this, (uint64_t)(uintptr_t)local->fctx);
-        }
-
-        /* Create itself has failed.. so return
-           without setxattring */
-        STRIPE_STACK_UNWIND(create, frame, local->op_ret, local->op_errno,
-                            local->fd, local->inode, &local->stbuf,
-                            &local->preparent, &local->postparent, NULL);
-    }
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_first_create_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno, fd_t *fd,
-                        inode_t *inode, struct iatt *buf,
-                        struct iatt *preparent, struct iatt *postparent,
-                        dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    call_frame_t *prev = NULL;
-    xlator_list_t *trav = NULL;
-    int i = 1;
-    dict_t *dict = NULL;
-    loc_t *loc = NULL;
-    int32_t need_unref = 0;
-    int32_t ret = -1;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    priv = this->private;
-    local = frame->local;
-    trav = this->children;
-    loc = &local->loc;
-
-    --local->call_count;
-
-    if (op_ret == -1) {
-        gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-               prev->this->name, strerror(op_errno));
-        local->failed = 1;
-        local->op_errno = op_errno;
-    }
-
-    local->op_ret = 0;
-    /* Get the mapping in inode private */
-    /* Get the stat buf right */
-    local->stbuf = *buf;
-    local->preparent = *preparent;
-    local->postparent = *postparent;
-
-    local->stbuf_blocks += buf->ia_blocks;
-    local->preparent_blocks += preparent->ia_blocks;
-    local->postparent_blocks += postparent->ia_blocks;
-
-    if (local->stbuf_size < buf->ia_size)
-        local->stbuf_size = buf->ia_size;
-    if (local->preparent_size < preparent->ia_size)
-        local->preparent_size = preparent->ia_size;
-    if (local->postparent_size < postparent->ia_size)
-        local->postparent_size = postparent->ia_size;
-
-    if (local->failed)
-        local->op_ret = -1;
-
-    if (local->op_ret == -1) {
-        local->call_count = 1;
-        STACK_WIND(frame, stripe_create_fail_unlink_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->unlink, &local->loc, 0, NULL);
-        return 0;
-    }
-
-    if (local->op_ret >= 0) {
-        local->preparent.ia_blocks = local->preparent_blocks;
-        local->preparent.ia_size = local->preparent_size;
-        local->postparent.ia_blocks = local->postparent_blocks;
-        local->postparent.ia_size = local->postparent_size;
-        local->stbuf.ia_size = local->stbuf_size;
-        local->stbuf.ia_blocks = local->stbuf_blocks;
-    }
-
-    /* Send a setxattr request to nodes where the
-       files are created */
-    trav = trav->next;
-    while (trav) {
-        if (priv->xattr_supported) {
-            dict = dict_new();
-            if (!dict) {
-                gf_log(this->name, GF_LOG_ERROR, "failed to allocate dict %s",
-                       loc->path);
-            }
-            need_unref = 1;
-
-            dict_copy(local->xattr, dict);
-
-            ret = stripe_xattr_request_build(this, dict, local->stripe_size,
-                                             priv->child_count, i,
-                                             priv->coalesce);
-            if (ret)
-                gf_log(this->name, GF_LOG_ERROR,
-                       "failed to build xattr request");
-        } else {
-            dict = local->xattr;
-        }
-
-        STACK_WIND(frame, stripe_create_cbk, trav->xlator,
-                   trav->xlator->fops->create, &local->loc, local->flags,
-                   local->mode, local->umask, local->fd, dict);
-        trav = trav->next;
-        if (need_unref && dict)
-            dict_unref(dict);
-        i++;
-    }
-
-out:
-    return 0;
-}
-
-/**
- * stripe_create - If a block-size is specified for the 'name', create the
- *    file in all the child nodes. If not, create it in only first child.
- *
- * @name- complete path of the file to be created.
- */
-int32_t
-stripe_create(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
-              mode_t mode, mode_t umask, fd_t *fd, dict_t *xdata)
-{
-    stripe_private_t *priv = NULL;
-    stripe_local_t *local = NULL;
-    int32_t op_errno = EINVAL;
-    int ret = 0;
-    int need_unref = 0;
-    int i = 0;
-    dict_t *dict = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-
-    /* files created in O_APPEND mode does not allow lseek() on fd */
-    flags &= ~O_APPEND;
-
-    if (priv->first_child_down || priv->nodes_down) {
-        gf_log(this->name, GF_LOG_DEBUG, "First node down, returning EIO");
-        op_errno = EIO;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    local->op_errno = ENOTCONN;
-    local->stripe_size = stripe_get_matching_bs(loc->path, priv);
-    frame->local = local;
-    local->inode = inode_ref(loc->inode);
-    loc_copy(&local->loc, loc);
-    local->fd = fd_ref(fd);
-    local->flags = flags;
-    local->mode = mode;
-    local->umask = umask;
-    if (xdata)
-        local->xattr = dict_ref(xdata);
-
-    local->call_count = priv->child_count;
-    /* Send a setxattr request to nodes where the
-       files are created */
-
-    if (priv->xattr_supported) {
-        dict = dict_new();
-        if (!dict) {
-            gf_log(this->name, GF_LOG_ERROR, "failed to allocate dict %s",
-                   loc->path);
-        }
-        need_unref = 1;
-
-        dict_copy(xdata, dict);
-
-        ret = stripe_xattr_request_build(this, dict, local->stripe_size,
-                                         priv->child_count, i, priv->coalesce);
-        if (ret)
-            gf_log(this->name, GF_LOG_ERROR, "failed to build xattr request");
-    } else {
-        dict = xdata;
-    }
-
-    STACK_WIND(frame, stripe_first_create_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->create, loc, flags, mode, umask, fd,
-               dict);
-
-    if (need_unref && dict)
-        dict_unref(dict);
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(create, frame, -1, op_errno, NULL, NULL, NULL, NULL,
-                        NULL, xdata);
-    return 0;
-}
-
-int32_t
-stripe_open_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                int32_t op_ret, int32_t op_errno, fd_t *fd, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-            local->op_errno = op_errno;
-        }
-
-        if (op_ret >= 0)
-            local->op_ret = op_ret;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        STRIPE_STACK_UNWIND(open, frame, local->op_ret, local->op_errno,
-                            local->fd, xdata);
-    }
-out:
-    return 0;
-}
-
-/**
- * stripe_open -
- */
-int32_t
-stripe_open(call_frame_t *frame, xlator_t *this, loc_t *loc, int32_t flags,
-            fd_t *fd, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    /* files opened in O_APPEND mode does not allow lseek() on fd */
-    flags &= ~O_APPEND;
-
-    local->fd = fd_ref(fd);
-    frame->local = local;
-    loc_copy(&local->loc, loc);
-
-    /* Striped files */
-    local->flags = flags;
-    local->call_count = priv->child_count;
-    local->stripe_size = stripe_get_matching_bs(loc->path, priv);
-
-    while (trav) {
-        STACK_WIND(frame, stripe_open_cbk, trav->xlator,
-                   trav->xlator->fops->open, &local->loc, local->flags,
-                   local->fd, xdata);
-        trav = trav->next;
-    }
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(open, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_opendir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                   int32_t op_ret, int32_t op_errno, fd_t *fd, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_ret = -1;
-            local->op_errno = op_errno;
-        }
-
-        if (op_ret >= 0)
-            local->op_ret = op_ret;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        STRIPE_STACK_UNWIND(opendir, frame, local->op_ret, local->op_errno,
-                            local->fd, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_opendir(call_frame_t *frame, xlator_t *this, loc_t *loc, fd_t *fd,
-               dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-    local->call_count = priv->child_count;
-    local->fd = fd_ref(fd);
-
-    while (trav) {
-        STACK_WIND(frame, stripe_opendir_cbk, trav->xlator,
-                   trav->xlator->fops->opendir, loc, fd, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(opendir, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_lk_cbk(call_frame_t *frame, void *cookie, xlator_t *this, int32_t op_ret,
-              int32_t op_errno, struct gf_flock *lock, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-        if (op_ret >= 0) {
-            if (FIRST_CHILD(this) == prev->this) {
-                /* First successful call, copy the *lock */
-                local->op_ret = op_ret;
-                local->lock = *lock;
-            }
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-        STRIPE_STACK_UNWIND(lk, frame, local->op_ret, local->op_errno,
-                            &local->lock, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_lk(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t cmd,
-          struct gf_flock *lock, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    trav = this->children;
-    priv = this->private;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_lk_cbk, trav->xlator, trav->xlator->fops->lk,
-                   fd, cmd, lock, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(lk, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_flush_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-                   strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-        if (op_ret >= 0)
-            local->op_ret = op_ret;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        STRIPE_STACK_UNWIND(flush, frame, local->op_ret, local->op_errno, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_flush(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_flush_cbk, trav->xlator,
-                   trav->xlator->fops->flush, fd, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(flush, frame, -1, op_errno, NULL);
-    return 0;
-}
-
-int32_t
-stripe_fsync_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                 struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-                   strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-        if (op_ret >= 0) {
-            local->op_ret = op_ret;
-            if (FIRST_CHILD(this) == prev->this) {
-                local->pre_buf = *prebuf;
-                local->post_buf = *postbuf;
-            }
-            local->prebuf_blocks += prebuf->ia_blocks;
-            local->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, local->fctx, prev);
-            correct_file_size(postbuf, local->fctx, prev);
-
-            if (local->prebuf_size < prebuf->ia_size)
-                local->prebuf_size = prebuf->ia_size;
-
-            if (local->postbuf_size < postbuf->ia_size)
-                local->postbuf_size = postbuf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->pre_buf.ia_blocks = local->prebuf_blocks;
-            local->pre_buf.ia_size = local->prebuf_size;
-            local->post_buf.ia_blocks = local->postbuf_blocks;
-            local->post_buf.ia_size = local->postbuf_size;
-        }
-
-        STRIPE_STACK_UNWIND(fsync, frame, local->op_ret, local->op_errno,
-                            &local->pre_buf, &local->post_buf, NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_fsync(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
-             dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    frame->local = local;
-
-    inode_ctx_get(fd->inode, this, (uint64_t *)&fctx);
-    if (!fctx) {
-        op_errno = EINVAL;
-        goto err;
-    }
-    local->fctx = fctx;
-    local->op_ret = -1;
-    local->call_count = priv->child_count;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_fsync_cbk, trav->xlator,
-                   trav->xlator->fops->fsync, fd, flags, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fsync, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_fstat_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, struct iatt *buf,
-                 dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned error %s",
-                   prev->this->name, strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-
-        if (op_ret == 0) {
-            local->op_ret = 0;
-
-            if (FIRST_CHILD(this) == prev->this)
-                local->stbuf = *buf;
-
-            local->stbuf_blocks += buf->ia_blocks;
-
-            correct_file_size(buf, local->fctx, prev);
-
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        if (local->op_ret != -1) {
-            local->stbuf.ia_size = local->stbuf_size;
-            local->stbuf.ia_blocks = local->stbuf_blocks;
-        }
-
-        STRIPE_STACK_UNWIND(fstat, frame, local->op_ret, local->op_errno,
-                            &local->stbuf, NULL);
-    }
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_fstat(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    if (IA_ISREG(fd->inode->ia_type)) {
-        inode_ctx_get(fd->inode, this, (uint64_t *)&fctx);
-        if (!fctx)
-            goto err;
-        local->fctx = fctx;
-    }
-
-    while (trav) {
-        STACK_WIND(frame, stripe_fstat_cbk, trav->xlator,
-                   trav->xlator->fops->fstat, fd, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fstat, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_ftruncate(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
-                 dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int i, eof_idx;
-    off_t dest_offset, tmp_offset;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    inode_ctx_get(fd->inode, this, (uint64_t *)&fctx);
-    if (!fctx) {
-        gf_log(this->name, GF_LOG_ERROR, "no stripe context");
-        op_errno = EINVAL;
-        goto err;
-    }
-    if (!fctx->stripe_count) {
-        gf_log(this->name, GF_LOG_ERROR, "no stripe count");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    local->fctx = fctx;
-    eof_idx = (offset / fctx->stripe_size) % fctx->stripe_count;
-
-    for (i = 0; i < fctx->stripe_count; i++) {
-        if (!fctx->xl_array[i]) {
-            gf_log(this->name, GF_LOG_ERROR,
-                   "no xlator at index "
-                   "%d",
-                   i);
-            op_errno = EINVAL;
-            goto err;
-        }
-
-        if (fctx->stripe_coalesce) {
-            if (i < eof_idx)
-                tmp_offset = gf_roof(offset,
-                                     fctx->stripe_size * fctx->stripe_count);
-            else if (i > eof_idx)
-                tmp_offset = gf_floor(offset,
-                                      fctx->stripe_size * fctx->stripe_count);
-            else
-                tmp_offset = offset;
-
-            dest_offset = coalesced_offset(tmp_offset, fctx->stripe_size,
-                                           fctx->stripe_count);
-        } else {
-            dest_offset = offset;
-        }
-
-        STACK_WIND(frame, stripe_truncate_cbk, fctx->xl_array[i],
-                   fctx->xl_array[i]->fops->ftruncate, fd, dest_offset, NULL);
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(ftruncate, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_fsyncdir_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-
-        if (op_ret == -1) {
-            gf_log(this->name, GF_LOG_DEBUG, "%s returned %s", prev->this->name,
-                   strerror(op_errno));
-            local->op_errno = op_errno;
-            if ((op_errno != ENOENT) || (prev->this == FIRST_CHILD(this)))
-                local->failed = 1;
-        }
-        if (op_ret >= 0)
-            local->op_ret = op_ret;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (local->failed)
-            local->op_ret = -1;
-
-        STRIPE_STACK_UNWIND(fsyncdir, frame, local->op_ret, local->op_errno,
-                            NULL);
-    }
-out:
-    return 0;
-}
-
-int32_t
-stripe_fsyncdir(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t flags,
-                dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    int32_t op_errno = 1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    local->call_count = priv->child_count;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_fsyncdir_cbk, trav->xlator,
-                   trav->xlator->fops->fsyncdir, fd, flags, NULL);
-        trav = trav->next;
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fsyncdir, frame, -1, op_errno, NULL);
-    return 0;
-}
-
-int32_t
-stripe_readv_fstat_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, struct iatt *buf,
-                       dict_t *xdata)
-{
-    int32_t i = 0;
-    int32_t callcnt = 0;
-    int32_t count = 0;
-    stripe_local_t *local = NULL;
-    struct iovec *vec = NULL;
-    struct iatt tmp_stbuf = {
-        0,
-    };
-    struct iobref *tmp_iobref = NULL;
-    struct iobuf *iobuf = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    local = frame->local;
-    prev = cookie;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->call_count;
-        if (op_ret != -1) {
-            correct_file_size(buf, local->fctx, prev);
-            if (local->stbuf_size < buf->ia_size)
-                local->stbuf_size = buf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        op_ret = 0;
-
-        /* Keep extra space for filling in '\0's */
-        vec = GF_CALLOC((local->count * 2), sizeof(struct iovec),
-                        gf_stripe_mt_iovec);
-        if (!vec) {
-            op_ret = -1;
-            goto done;
-        }
-
-        for (i = 0; i < local->wind_count; i++) {
-            if (local->replies[i].op_ret) {
-                memcpy((vec + count), local->replies[i].vector,
-                       (local->replies[i].count * sizeof(struct iovec)));
-                count += local->replies[i].count;
-                op_ret += local->replies[i].op_ret;
-            }
-            if ((local->replies[i].op_ret < local->replies[i].requested_size) &&
-                (local->stbuf_size > (local->offset + op_ret))) {
-                /* Fill in 0s here */
-                vec[count].iov_len = (local->replies[i].requested_size -
-                                      local->replies[i].op_ret);
-                iobuf = iobuf_get2(this->ctx->iobuf_pool, vec[count].iov_len);
-                if (!iobuf) {
-                    gf_log(this->name, GF_LOG_ERROR, "Out of memory.");
-                    op_ret = -1;
-                    op_errno = ENOMEM;
-                    goto done;
-                }
-                memset(iobuf->ptr, 0, vec[count].iov_len);
-                vec[count].iov_base = iobuf->ptr;
-
-                iobref_add(local->iobref, iobuf);
-                iobuf_unref(iobuf);
-
-                op_ret += vec[count].iov_len;
-                count++;
-            }
-            GF_FREE(local->replies[i].vector);
-        }
-
-        /* ENOENT signals EOF to the NFS-server */
-        if (op_ret != -1 && op_ret < local->readv_size &&
-            (local->offset + op_ret == buf->ia_size))
-            op_errno = ENOENT;
-
-        /* FIXME: notice that st_ino, and st_dev (gen) will be
-         * different than what inode will have. Make sure this doesn't
-         * cause any bugs at higher levels */
-        memcpy(&tmp_stbuf, &local->replies[0].stbuf, sizeof(struct iatt));
-        tmp_stbuf.ia_size = local->stbuf_size;
-
-    done:
-        GF_FREE(local->replies);
-        tmp_iobref = local->iobref;
-        STRIPE_STACK_UNWIND(readv, frame, op_ret, op_errno, vec, count,
-                            &tmp_stbuf, tmp_iobref, NULL);
-
-        iobref_unref(tmp_iobref);
-        GF_FREE(vec);
-    }
-out:
-    return 0;
-}
-
-/**
- * stripe_readv_cbk - get all the striped reads, and order it properly, send it
- *        to above layer after putting it in a single vector.
- */
-int32_t
-stripe_readv_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                 int32_t op_ret, int32_t op_errno, struct iovec *vector,
-                 int32_t count, struct iatt *stbuf, struct iobref *iobref,
-                 dict_t *xdata)
-{
-    int32_t index = 0;
-    int32_t callcnt = 0;
-    int32_t final_count = 0;
-    int32_t need_to_check_proper_size = 0;
-    call_frame_t *mframe = NULL;
-    stripe_local_t *mlocal = NULL;
-    stripe_local_t *local = NULL;
-    struct iovec *final_vec = NULL;
-    struct iatt tmp_stbuf = {
-        0,
-    };
-    struct iatt *tmp_stbuf_p = NULL;  // need it for a warning
-    struct iobref *tmp_iobref = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    call_frame_t *prev = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto end;
-    }
-
-    local = frame->local;
-    index = local->node_index;
-    prev = cookie;
-    mframe = local->orig_frame;
-    if (!mframe)
-        goto out;
-
-    mlocal = mframe->local;
-    if (!mlocal)
-        goto out;
-
-    fctx = mlocal->fctx;
-
-    LOCK(&mframe->lock);
-    {
-        mlocal->replies[index].op_ret = op_ret;
-        mlocal->replies[index].op_errno = op_errno;
-        mlocal->replies[index].requested_size = local->readv_size;
-        if (op_ret >= 0) {
-            mlocal->replies[index].stbuf = *stbuf;
-            mlocal->replies[index].count = count;
-            mlocal->replies[index].vector = iov_dup(vector, count);
-
-            correct_file_size(stbuf, fctx, prev);
-
-            if (local->stbuf_size < stbuf->ia_size)
-                local->stbuf_size = stbuf->ia_size;
-            local->stbuf_blocks += stbuf->ia_blocks;
-
-            if (!mlocal->iobref)
-                mlocal->iobref = iobref_new();
-            iobref_merge(mlocal->iobref, iobref);
-        }
-        callcnt = ++mlocal->call_count;
-    }
-    UNLOCK(&mframe->lock);
-
-    if (callcnt == mlocal->wind_count) {
-        op_ret = 0;
-
-        for (index = 0; index < mlocal->wind_count; index++) {
-            /* check whether each stripe returned
-             * 'expected' number of bytes */
-            if (mlocal->replies[index].op_ret == -1) {
-                op_ret = -1;
-                op_errno = mlocal->replies[index].op_errno;
-                break;
-            }
-            /* TODO: handle the 'holes' within the read range
-               properly */
-            if (mlocal->replies[index].op_ret <
-                mlocal->replies[index].requested_size) {
-                need_to_check_proper_size = 1;
-            }
-
-            op_ret += mlocal->replies[index].op_ret;
-            mlocal->count += mlocal->replies[index].count;
-        }
-        if (op_ret == -1)
-            goto done;
-        if (need_to_check_proper_size)
-            goto check_size;
-
-        final_vec = GF_CALLOC(mlocal->count, sizeof(struct iovec),
-                              gf_stripe_mt_iovec);
-
-        if (!final_vec) {
-            op_ret = -1;
-            goto done;
-        }
-
-        for (index = 0; index < mlocal->wind_count; index++) {
-            memcpy((final_vec + final_count), mlocal->replies[index].vector,
-                   (mlocal->replies[index].count * sizeof(struct iovec)));
-            final_count += mlocal->replies[index].count;
-            GF_FREE(mlocal->replies[index].vector);
-        }
-
-        /* FIXME: notice that st_ino, and st_dev (gen) will be
-         * different than what inode will have. Make sure this doesn't
-         * cause any bugs at higher levels */
-        memcpy(&tmp_stbuf, &mlocal->replies[0].stbuf, sizeof(struct iatt));
-        tmp_stbuf.ia_size = local->stbuf_size;
-        tmp_stbuf.ia_blocks = local->stbuf_blocks;
-
-    done:
-        /* */
-        GF_FREE(mlocal->replies);
-        tmp_iobref = mlocal->iobref;
-        /* work around for nfs truncated read. Bug 3774 */
-        tmp_stbuf_p = &tmp_stbuf;
-        WIPE(tmp_stbuf_p);
-        STRIPE_STACK_UNWIND(readv, mframe, op_ret, op_errno, final_vec,
-                            final_count, &tmp_stbuf, tmp_iobref, NULL);
-
-        iobref_unref(tmp_iobref);
-        GF_FREE(final_vec);
-    }
-
-    goto out;
-
-check_size:
-    mlocal->call_count = fctx->stripe_count;
-
-    for (index = 0; index < fctx->stripe_count; index++) {
-        STACK_WIND(mframe, stripe_readv_fstat_cbk, (fctx->xl_array[index]),
-                   (fctx->xl_array[index])->fops->fstat, mlocal->fd, NULL);
-    }
-
-out:
-    STRIPE_STACK_DESTROY(frame);
-end:
-    return 0;
-}
-
-int32_t
-stripe_readv(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-             off_t offset, uint32_t flags, dict_t *xdata)
-{
-    int32_t op_errno = EINVAL;
-    int32_t idx = 0;
-    int32_t index = 0;
-    int32_t num_stripe = 0;
-    int32_t off_index = 0;
-    size_t frame_size = 0;
-    off_t rounded_end = 0;
-    uint64_t tmp_fctx = 0;
-    uint64_t stripe_size = 0;
-    off_t rounded_start = 0;
-    off_t frame_offset = offset;
-    off_t dest_offset = 0;
-    stripe_local_t *local = NULL;
-    call_frame_t *rframe = NULL;
-    stripe_local_t *rlocal = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    inode_ctx_get(fd->inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        op_errno = EBADFD;
-        goto err;
-    }
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-    stripe_size = fctx->stripe_size;
-
-    STRIPE_VALIDATE_FCTX(fctx, err);
-
-    if (!stripe_size) {
-        gf_log(this->name, GF_LOG_DEBUG, "Wrong stripe size for the file");
-        goto err;
-    }
-    /* The file is stripe across the child nodes. Send the read request
-     * to the child nodes appropriately after checking which region of
-     * the file is in which child node. Always '0-<stripe_size>' part of
-     * the file resides in the first child.
-     */
-    rounded_start = gf_floor(offset, stripe_size);
-    rounded_end = gf_roof(offset + size, stripe_size);
-    num_stripe = (rounded_end - rounded_start) / stripe_size;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-
-    /* This is where all the vectors should be copied. */
-    local->replies = GF_CALLOC(num_stripe, sizeof(struct stripe_replies),
-                               gf_stripe_mt_stripe_replies);
-    if (!local->replies) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    off_index = (offset / stripe_size) % fctx->stripe_count;
-    local->wind_count = num_stripe;
-    local->readv_size = size;
-    local->offset = offset;
-    local->fd = fd_ref(fd);
-    local->fctx = fctx;
-
-    for (index = off_index; index < (num_stripe + off_index); index++) {
-        rframe = copy_frame(frame);
-        rlocal = mem_get0(this->local_pool);
-        if (!rlocal) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-
-        frame_size = min(gf_roof(frame_offset + 1, stripe_size),
-                         (offset + size)) -
-                     frame_offset;
-
-        rlocal->node_index = index - off_index;
-        rlocal->orig_frame = frame;
-        rlocal->readv_size = frame_size;
-        rframe->local = rlocal;
-        idx = (index % fctx->stripe_count);
-
-        if (fctx->stripe_coalesce)
-            dest_offset = coalesced_offset(frame_offset, stripe_size,
-                                           fctx->stripe_count);
-        else
-            dest_offset = frame_offset;
-
-        STACK_WIND(rframe, stripe_readv_cbk, fctx->xl_array[idx],
-                   fctx->xl_array[idx]->fops->readv, fd, frame_size,
-                   dest_offset, flags, xdata);
-
-        frame_offset += frame_size;
-    }
-
-    return 0;
-err:
-    if (rframe)
-        STRIPE_STACK_DESTROY(rframe);
-
-    STRIPE_STACK_UNWIND(readv, frame, -1, op_errno, NULL, 0, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_writev_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                  struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_local_t *mlocal = NULL;
-    call_frame_t *prev = NULL;
-    call_frame_t *mframe = NULL;
-    struct stripe_replies *reply = NULL;
-    int32_t i = 0;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-    mframe = local->orig_frame;
-    mlocal = mframe->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = ++mlocal->call_count;
-
-        mlocal->replies[local->node_index].op_ret = op_ret;
-        mlocal->replies[local->node_index].op_errno = op_errno;
-
-        if (op_ret >= 0) {
-            mlocal->post_buf = *postbuf;
-            mlocal->pre_buf = *prebuf;
-
-            mlocal->prebuf_blocks += prebuf->ia_blocks;
-            mlocal->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, mlocal->fctx, prev);
-            correct_file_size(postbuf, mlocal->fctx, prev);
-
-            if (mlocal->prebuf_size < prebuf->ia_size)
-                mlocal->prebuf_size = prebuf->ia_size;
-            if (mlocal->postbuf_size < postbuf->ia_size)
-                mlocal->postbuf_size = postbuf->ia_size;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if ((callcnt == mlocal->wind_count) && mlocal->unwind) {
-        mlocal->pre_buf.ia_size = mlocal->prebuf_size;
-        mlocal->pre_buf.ia_blocks = mlocal->prebuf_blocks;
-        mlocal->post_buf.ia_size = mlocal->postbuf_size;
-        mlocal->post_buf.ia_blocks = mlocal->postbuf_blocks;
-
-        /*
-         * Only return the number of consecutively written bytes up until
-         * the first error. Only return an error if it occurs first.
-         *
-         * When a short write occurs, the application should retry at the
-         * appropriate offset, at which point we'll potentially pass back
-         * the error.
-         */
-        for (i = 0, reply = mlocal->replies; i < mlocal->wind_count;
-             i++, reply++) {
-            if (reply->op_ret == -1) {
-                gf_log(this->name, GF_LOG_DEBUG,
-                       "reply %d "
-                       "returned error %s",
-                       i, strerror(reply->op_errno));
-                if (!mlocal->op_ret) {
-                    mlocal->op_ret = -1;
-                    mlocal->op_errno = reply->op_errno;
-                }
-                break;
-            }
-
-            mlocal->op_ret += reply->op_ret;
-
-            if (reply->op_ret < reply->requested_size)
-                break;
-        }
-
-        GF_FREE(mlocal->replies);
-
-        STRIPE_STACK_UNWIND(writev, mframe, mlocal->op_ret, mlocal->op_errno,
-                            &mlocal->pre_buf, &mlocal->post_buf, NULL);
-    }
-out:
-    if (frame)
-        STRIPE_STACK_DESTROY(frame);
-    return 0;
-}
-
-int32_t
-stripe_writev(call_frame_t *frame, xlator_t *this, fd_t *fd,
-              struct iovec *vector, int32_t count, off_t offset, uint32_t flags,
-              struct iobref *iobref, dict_t *xdata)
-{
-    struct iovec *tmp_vec = NULL;
-    stripe_local_t *local = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-    int32_t idx = 0;
-    int32_t total_size = 0;
-    int32_t offset_offset = 0;
-    int32_t remaining_size = 0;
-    int32_t tmp_count = count;
-    off_t fill_size = 0;
-    uint64_t stripe_size = 0;
-    uint64_t tmp_fctx = 0;
-    off_t dest_offset = 0;
-    off_t rounded_start = 0;
-    off_t rounded_end = 0;
-    int32_t total_chunks = 0;
-    call_frame_t *wframe = NULL;
-    stripe_local_t *wlocal = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    inode_ctx_get(fd->inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        op_errno = EINVAL;
-        goto err;
-    }
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-    stripe_size = fctx->stripe_size;
-
-    STRIPE_VALIDATE_FCTX(fctx, err);
-
-    /* File has to be stripped across the child nodes */
-    total_size = iov_length(vector, count);
-    remaining_size = total_size;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-    local->stripe_size = stripe_size;
-    local->fctx = fctx;
-
-    if (!stripe_size) {
-        gf_log(this->name, GF_LOG_DEBUG, "Wrong stripe size for the file");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    rounded_start = gf_floor(offset, stripe_size);
-    rounded_end = gf_roof(offset + total_size, stripe_size);
-    total_chunks = (rounded_end - rounded_start) / stripe_size;
-    local->replies = GF_CALLOC(total_chunks, sizeof(struct stripe_replies),
-                               gf_stripe_mt_stripe_replies);
-    if (!local->replies) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    total_chunks = 0;
-    while (1) {
-        wframe = copy_frame(frame);
-        wlocal = mem_get0(this->local_pool);
-        if (!wlocal) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        wlocal->orig_frame = frame;
-        wframe->local = wlocal;
-
-        /* Send striped chunk of the vector to child
-           nodes appropriately. */
-        idx = (((offset + offset_offset) / local->stripe_size) %
-               fctx->stripe_count);
-
-        fill_size = (local->stripe_size -
-                     ((offset + offset_offset) % local->stripe_size));
-        if (fill_size > remaining_size)
-            fill_size = remaining_size;
-
-        remaining_size -= fill_size;
-
-        tmp_count = iov_subset(vector, count, offset_offset,
-                               offset_offset + fill_size, NULL);
-        tmp_vec = GF_CALLOC(tmp_count, sizeof(struct iovec),
-                            gf_stripe_mt_iovec);
-        if (!tmp_vec) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        tmp_count = iov_subset(vector, count, offset_offset,
-                               offset_offset + fill_size, tmp_vec);
-
-        local->wind_count++;
-        if (remaining_size == 0)
-            local->unwind = 1;
-
-        /*
-         * Store off the request index (with respect to the chunk of the
-         * initial offset) and the size of the request. This is required
-         * in the callback to calculate an appropriate return value in
-         * the event of a write failure in one or more requests.
-         */
-        wlocal->node_index = total_chunks;
-        local->replies[total_chunks].requested_size = fill_size;
-
-        dest_offset = offset + offset_offset;
-        if (fctx->stripe_coalesce)
-            dest_offset = coalesced_offset(dest_offset, local->stripe_size,
-                                           fctx->stripe_count);
-
-        STACK_WIND(wframe, stripe_writev_cbk, fctx->xl_array[idx],
-                   fctx->xl_array[idx]->fops->writev, fd, tmp_vec, tmp_count,
-                   dest_offset, flags, iobref, xdata);
-
-        GF_FREE(tmp_vec);
-        offset_offset += fill_size;
-        total_chunks++;
-        if (remaining_size == 0)
-            break;
-    }
-
-    return 0;
-err:
-    if (wframe)
-        STRIPE_STACK_DESTROY(wframe);
-
-    STRIPE_STACK_UNWIND(writev, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_fallocate_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                     int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                     struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_local_t *mlocal = NULL;
-    call_frame_t *prev = NULL;
-    call_frame_t *mframe = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-    mframe = local->orig_frame;
-    mlocal = mframe->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = ++mlocal->call_count;
-
-        if (op_ret == 0) {
-            mlocal->post_buf = *postbuf;
-            mlocal->pre_buf = *prebuf;
-
-            mlocal->prebuf_blocks += prebuf->ia_blocks;
-            mlocal->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, mlocal->fctx, prev);
-            correct_file_size(postbuf, mlocal->fctx, prev);
-
-            if (mlocal->prebuf_size < prebuf->ia_size)
-                mlocal->prebuf_size = prebuf->ia_size;
-            if (mlocal->postbuf_size < postbuf->ia_size)
-                mlocal->postbuf_size = postbuf->ia_size;
-        }
-
-        /* return the first failure */
-        if (mlocal->op_ret == 0) {
-            mlocal->op_ret = op_ret;
-            mlocal->op_errno = op_errno;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if ((callcnt == mlocal->wind_count) && mlocal->unwind) {
-        mlocal->pre_buf.ia_size = mlocal->prebuf_size;
-        mlocal->pre_buf.ia_blocks = mlocal->prebuf_blocks;
-        mlocal->post_buf.ia_size = mlocal->postbuf_size;
-        mlocal->post_buf.ia_blocks = mlocal->postbuf_blocks;
-
-        STRIPE_STACK_UNWIND(fallocate, mframe, mlocal->op_ret, mlocal->op_errno,
-                            &mlocal->pre_buf, &mlocal->post_buf, NULL);
-    }
-out:
-    if (frame)
-        STRIPE_STACK_DESTROY(frame);
-    return 0;
-}
-
-int32_t
-stripe_fallocate(call_frame_t *frame, xlator_t *this, fd_t *fd, int32_t mode,
-                 off_t offset, size_t len, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-    int32_t idx = 0;
-    int32_t offset_offset = 0;
-    int32_t remaining_size = 0;
-    off_t fill_size = 0;
-    uint64_t stripe_size = 0;
-    uint64_t tmp_fctx = 0;
-    off_t dest_offset = 0;
-    call_frame_t *fframe = NULL;
-    stripe_local_t *flocal = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    inode_ctx_get(fd->inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        op_errno = EINVAL;
-        goto err;
-    }
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-    stripe_size = fctx->stripe_size;
-
-    STRIPE_VALIDATE_FCTX(fctx, err);
-
-    remaining_size = len;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-    local->stripe_size = stripe_size;
-    local->fctx = fctx;
-
-    if (!stripe_size) {
-        gf_log(this->name, GF_LOG_DEBUG, "Wrong stripe size for the file");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    while (1) {
-        fframe = copy_frame(frame);
-        flocal = mem_get0(this->local_pool);
-        if (!flocal) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        flocal->orig_frame = frame;
-        fframe->local = flocal;
-
-        /* send fallocate request to the associated child node */
-        idx = (((offset + offset_offset) / local->stripe_size) %
-               fctx->stripe_count);
-
-        fill_size = (local->stripe_size -
-                     ((offset + offset_offset) % local->stripe_size));
-        if (fill_size > remaining_size)
-            fill_size = remaining_size;
-
-        remaining_size -= fill_size;
-
-        local->wind_count++;
-        if (remaining_size == 0)
-            local->unwind = 1;
-
-        dest_offset = offset + offset_offset;
-        if (fctx->stripe_coalesce)
-            dest_offset = coalesced_offset(dest_offset, local->stripe_size,
-                                           fctx->stripe_count);
-
-        /*
-         * TODO: Create a separate handler for coalesce mode that sends a
-         * single fallocate per-child (since the ranges are linear).
-         */
-        STACK_WIND(fframe, stripe_fallocate_cbk, fctx->xl_array[idx],
-                   fctx->xl_array[idx]->fops->fallocate, fd, mode, dest_offset,
-                   fill_size, xdata);
-
-        offset_offset += fill_size;
-        if (remaining_size == 0)
-            break;
-    }
-
-    return 0;
-err:
-    if (fframe)
-        STRIPE_STACK_DESTROY(fframe);
-
-    STRIPE_STACK_UNWIND(fallocate, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_discard_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                   int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                   struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_local_t *mlocal = NULL;
-    call_frame_t *prev = NULL;
-    call_frame_t *mframe = NULL;
-
-    if (!this || !frame || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-    mframe = local->orig_frame;
-    mlocal = mframe->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = ++mlocal->call_count;
-
-        if (op_ret == 0) {
-            mlocal->post_buf = *postbuf;
-            mlocal->pre_buf = *prebuf;
-
-            mlocal->prebuf_blocks += prebuf->ia_blocks;
-            mlocal->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, mlocal->fctx, prev);
-            correct_file_size(postbuf, mlocal->fctx, prev);
-
-            if (mlocal->prebuf_size < prebuf->ia_size)
-                mlocal->prebuf_size = prebuf->ia_size;
-            if (mlocal->postbuf_size < postbuf->ia_size)
-                mlocal->postbuf_size = postbuf->ia_size;
-        }
-
-        /* return the first failure */
-        if (mlocal->op_ret == 0) {
-            mlocal->op_ret = op_ret;
-            mlocal->op_errno = op_errno;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if ((callcnt == mlocal->wind_count) && mlocal->unwind) {
-        mlocal->pre_buf.ia_size = mlocal->prebuf_size;
-        mlocal->pre_buf.ia_blocks = mlocal->prebuf_blocks;
-        mlocal->post_buf.ia_size = mlocal->postbuf_size;
-        mlocal->post_buf.ia_blocks = mlocal->postbuf_blocks;
-
-        STRIPE_STACK_UNWIND(discard, mframe, mlocal->op_ret, mlocal->op_errno,
-                            &mlocal->pre_buf, &mlocal->post_buf, NULL);
-    }
-out:
-    if (frame)
-        STRIPE_STACK_DESTROY(frame);
-
-    return 0;
-}
-
-int32_t
-stripe_discard(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
-               size_t len, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-    int32_t idx = 0;
-    int32_t offset_offset = 0;
-    int32_t remaining_size = 0;
-    off_t fill_size = 0;
-    uint64_t stripe_size = 0;
-    uint64_t tmp_fctx = 0;
-    off_t dest_offset = 0;
-    call_frame_t *fframe = NULL;
-    stripe_local_t *flocal = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    inode_ctx_get(fd->inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        op_errno = EINVAL;
-        goto err;
-    }
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-    stripe_size = fctx->stripe_size;
-
-    STRIPE_VALIDATE_FCTX(fctx, err);
-
-    remaining_size = len;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-    local->stripe_size = stripe_size;
-    local->fctx = fctx;
-
-    if (!stripe_size) {
-        gf_log(this->name, GF_LOG_DEBUG, "Wrong stripe size for the file");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    while (1) {
-        fframe = copy_frame(frame);
-        flocal = mem_get0(this->local_pool);
-        if (!flocal) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        flocal->orig_frame = frame;
-        fframe->local = flocal;
-
-        /* send discard request to the associated child node */
-        idx = (((offset + offset_offset) / local->stripe_size) %
-               fctx->stripe_count);
-
-        fill_size = (local->stripe_size -
-                     ((offset + offset_offset) % local->stripe_size));
-        if (fill_size > remaining_size)
-            fill_size = remaining_size;
-
-        remaining_size -= fill_size;
-
-        local->wind_count++;
-        if (remaining_size == 0)
-            local->unwind = 1;
-
-        dest_offset = offset + offset_offset;
-        if (fctx->stripe_coalesce)
-            dest_offset = coalesced_offset(dest_offset, local->stripe_size,
-                                           fctx->stripe_count);
-
-        /*
-         * TODO: Create a separate handler for coalesce mode that sends a
-         * single discard per-child (since the ranges are linear).
-         */
-        STACK_WIND(fframe, stripe_discard_cbk, fctx->xl_array[idx],
-                   fctx->xl_array[idx]->fops->discard, fd, dest_offset,
-                   fill_size, xdata);
-
-        offset_offset += fill_size;
-        if (remaining_size == 0)
-            break;
-    }
-
-    return 0;
-err:
-    if (fframe)
-        STRIPE_STACK_DESTROY(fframe);
-
-    STRIPE_STACK_UNWIND(discard, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_zerofill_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int32_t op_ret, int32_t op_errno, struct iatt *prebuf,
-                    struct iatt *postbuf, dict_t *xdata)
-{
-    int32_t callcnt = 0;
-    stripe_local_t *local = NULL;
-    stripe_local_t *mlocal = NULL;
-    call_frame_t *prev = NULL;
-    call_frame_t *mframe = NULL;
-
-    GF_ASSERT(frame);
-
-    if (!this || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-
-    prev = cookie;
-    local = frame->local;
-    mframe = local->orig_frame;
-    mlocal = mframe->local;
-
-    LOCK(&frame->lock);
-    {
-        callcnt = ++mlocal->call_count;
-
-        if (op_ret == 0) {
-            mlocal->post_buf = *postbuf;
-            mlocal->pre_buf = *prebuf;
-
-            mlocal->prebuf_blocks += prebuf->ia_blocks;
-            mlocal->postbuf_blocks += postbuf->ia_blocks;
-
-            correct_file_size(prebuf, mlocal->fctx, prev);
-            correct_file_size(postbuf, mlocal->fctx, prev);
-
-            if (mlocal->prebuf_size < prebuf->ia_size)
-                mlocal->prebuf_size = prebuf->ia_size;
-            if (mlocal->postbuf_size < postbuf->ia_size)
-                mlocal->postbuf_size = postbuf->ia_size;
-        }
-
-        /* return the first failure */
-        if (mlocal->op_ret == 0) {
-            mlocal->op_ret = op_ret;
-            mlocal->op_errno = op_errno;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if ((callcnt == mlocal->wind_count) && mlocal->unwind) {
-        mlocal->pre_buf.ia_size = mlocal->prebuf_size;
-        mlocal->pre_buf.ia_blocks = mlocal->prebuf_blocks;
-        mlocal->post_buf.ia_size = mlocal->postbuf_size;
-        mlocal->post_buf.ia_blocks = mlocal->postbuf_blocks;
-
-        STRIPE_STACK_UNWIND(zerofill, mframe, mlocal->op_ret, mlocal->op_errno,
-                            &mlocal->pre_buf, &mlocal->post_buf, NULL);
-    }
-out:
-    STRIPE_STACK_DESTROY(frame);
-    return 0;
-}
-
-int32_t
-stripe_zerofill(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
-                off_t len, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_fd_ctx_t *fctx = NULL;
-    int32_t op_errno = 1;
-    int32_t idx = 0;
-    int32_t offset_offset = 0;
-    int32_t remaining_size = 0;
-    off_t fill_size = 0;
-    uint64_t stripe_size = 0;
-    uint64_t tmp_fctx = 0;
-    off_t dest_offset = 0;
-    call_frame_t *fframe = NULL;
-    stripe_local_t *flocal = NULL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-    VALIDATE_OR_GOTO(fd->inode, err);
-
-    inode_ctx_get(fd->inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        op_errno = EINVAL;
-        goto err;
-    }
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-    stripe_size = fctx->stripe_size;
-
-    STRIPE_VALIDATE_FCTX(fctx, err);
-
-    remaining_size = len;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    frame->local = local;
-    local->stripe_size = stripe_size;
-    local->fctx = fctx;
-
-    if (!stripe_size) {
-        gf_log(this->name, GF_LOG_DEBUG, "Wrong stripe size for the file");
-        op_errno = EINVAL;
-        goto err;
-    }
-
-    while (1) {
-        fframe = copy_frame(frame);
-        flocal = mem_get0(this->local_pool);
-        if (!flocal) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-        flocal->orig_frame = frame;
-        fframe->local = flocal;
-
-        idx = (((offset + offset_offset) / local->stripe_size) %
-               fctx->stripe_count);
-
-        fill_size = (local->stripe_size -
-                     ((offset + offset_offset) % local->stripe_size));
-        if (fill_size > remaining_size)
-            fill_size = remaining_size;
-
-        remaining_size -= fill_size;
-
-        local->wind_count++;
-        if (remaining_size == 0)
-            local->unwind = 1;
-
-        dest_offset = offset + offset_offset;
-        if (fctx->stripe_coalesce)
-            dest_offset = coalesced_offset(dest_offset, local->stripe_size,
-                                           fctx->stripe_count);
-
-        STACK_WIND(fframe, stripe_zerofill_cbk, fctx->xl_array[idx],
-                   fctx->xl_array[idx]->fops->zerofill, fd, dest_offset,
-                   fill_size, xdata);
-        offset_offset += fill_size;
-        if (remaining_size == 0)
-            break;
-    }
-
-    return 0;
-err:
-    if (fframe)
-        STRIPE_STACK_DESTROY(fframe);
-
-    STRIPE_STACK_UNWIND(zerofill, frame, -1, op_errno, NULL, NULL, NULL);
-    return 0;
-}
-
-int32_t
-stripe_seek(call_frame_t *frame, xlator_t *this, fd_t *fd, off_t offset,
-            gf_seek_what_t what, dict_t *xdata)
-{
-    /* TBD */
-    gf_log(this->name, GF_LOG_INFO, "seek called on %s.",
-           uuid_utoa(fd->inode->gfid));
-    STRIPE_STACK_UNWIND(seek, frame, -1, ENOTSUP, 0, NULL);
-    return 0;
-}
-
-int32_t
-stripe_release(xlator_t *this, fd_t *fd)
-{
-    return 0;
-}
-
-int
-stripe_forget(xlator_t *this, inode_t *inode)
-{
-    uint64_t tmp_fctx = 0;
-    stripe_fd_ctx_t *fctx = NULL;
-
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(inode, err);
-
-    (void)inode_ctx_del(inode, this, &tmp_fctx);
-    if (!tmp_fctx) {
-        goto err;
-    }
-
-    fctx = (stripe_fd_ctx_t *)(long)tmp_fctx;
-
-    if (!fctx->static_array)
-        GF_FREE(fctx->xl_array);
-
-    GF_FREE(fctx);
-err:
-    return 0;
-}
-
-int32_t
-notify(xlator_t *this, int32_t event, void *data, ...)
-{
-    stripe_private_t *priv = NULL;
-    int down_client = 0;
-    int i = 0;
-    gf_boolean_t heard_from_all_children = _gf_false;
-
-    if (!this)
-        return 0;
-
-    priv = this->private;
-    if (!priv)
-        return 0;
-
-    switch (event) {
-        case GF_EVENT_CHILD_UP: {
-            /* get an index number to set */
-            for (i = 0; i < priv->child_count; i++) {
-                if (data == priv->xl_array[i])
-                    break;
-            }
-
-            if (priv->child_count == i) {
-                gf_log(this->name, GF_LOG_ERROR,
-                       "got GF_EVENT_CHILD_UP bad subvolume %s",
-                       data ? ((xlator_t *)data)->name : NULL);
-                break;
-            }
-
-            LOCK(&priv->lock);
-            {
-                if (data == FIRST_CHILD(this))
-                    priv->first_child_down = 0;
-                priv->last_event[i] = event;
-            }
-            UNLOCK(&priv->lock);
-        } break;
-        case GF_EVENT_CHILD_CONNECTING: {
-            // 'CONNECTING' doesn't ensure its CHILD_UP, so do nothing
-            goto out;
-        }
-        case GF_EVENT_CHILD_DOWN: {
-            /* get an index number to set */
-            for (i = 0; i < priv->child_count; i++) {
-                if (data == priv->xl_array[i])
-                    break;
-            }
-
-            if (priv->child_count == i) {
-                gf_log(this->name, GF_LOG_ERROR,
-                       "got GF_EVENT_CHILD_DOWN bad subvolume %s",
-                       data ? ((xlator_t *)data)->name : NULL);
-                break;
-            }
-
-            LOCK(&priv->lock);
-            {
-                if (data == FIRST_CHILD(this))
-                    priv->first_child_down = 1;
-                priv->last_event[i] = event;
-            }
-            UNLOCK(&priv->lock);
-        } break;
-
-        default: {
-            /* */
-            default_notify(this, event, data);
-            goto out;
-        } break;
-    }
-
-    // Consider child as down if it's last_event is not CHILD_UP
-    for (i = 0, down_client = 0; i < priv->child_count; i++)
-        if (priv->last_event[i] != GF_EVENT_CHILD_UP)
-            down_client++;
-
-    LOCK(&priv->lock);
-    {
-        priv->nodes_down = down_client;
-    }
-    UNLOCK(&priv->lock);
-
-    heard_from_all_children = _gf_true;
-    for (i = 0; i < priv->child_count; i++)
-        if (!priv->last_event[i])
-            heard_from_all_children = _gf_false;
-
-    if (heard_from_all_children)
-        default_notify(this, event, data);
-out:
-    return 0;
-}
-
-int
-stripe_setxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int op_ret, int op_errno, dict_t *xdata)
-{
-    int ret = -1;
-    int call_cnt = 0;
-    stripe_local_t *local = NULL;
-
-    if (!frame || !frame->local || !this) {
-        gf_log("", GF_LOG_ERROR, "Possible NULL deref");
-        return ret;
-    }
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        call_cnt = --local->wind_count;
-
-        /**
-         * We overwrite ->op_* values here for subsequent failure
-         * conditions, hence we propagate the last errno down the
-         * stack.
-         */
-        if (op_ret < 0) {
-            local->op_ret = op_ret;
-            local->op_errno = op_errno;
-            goto unlock;
-        }
-    }
-
-unlock:
-    UNLOCK(&frame->lock);
-
-    if (!call_cnt) {
-        STRIPE_STACK_UNWIND(setxattr, frame, local->op_ret, local->op_errno,
-                            xdata);
-    }
-
-    return 0;
-}
-
-#ifdef HAVE_BD_XLATOR
-int
-stripe_is_bd(dict_t *this, char *key, data_t *value, void *data)
-{
-    gf_boolean_t *is_bd = data;
-
-    if (data == NULL)
-        return 0;
-
-    if (XATTR_IS_BD(key))
-        *is_bd = _gf_true;
-
-    return 0;
-}
-
-static gf_boolean_t
-stripe_setxattr_is_bd(dict_t *dict)
-{
-    gf_boolean_t is_bd = _gf_false;
-
-    if (dict == NULL)
-        goto out;
-
-    dict_foreach(dict, stripe_is_bd, &is_bd);
-out:
-    return is_bd;
-}
-#else
-#define stripe_setxattr_is_bd(dict) _gf_false
-#endif
-
-int
-stripe_setxattr(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *dict,
-                int flags, dict_t *xdata)
-{
-    int32_t op_errno = EINVAL;
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    stripe_local_t *local = NULL;
-    int i = 0;
-    gf_boolean_t is_bd = _gf_false;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    GF_IF_INTERNAL_XATTR_GOTO("trusted.*stripe*", dict, op_errno, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    frame->local = local;
-    local->wind_count = priv->child_count;
-    local->op_ret = local->op_errno = 0;
-
-    is_bd = stripe_setxattr_is_bd(dict);
-
-    /**
-     * Set xattrs for directories on all subvolumes. Additionally
-     * this power is only given to a special client. Bd xlator
-     * also needs xattrs for regular files (ie LVs)
-     */
-    if (((frame->root->pid == GF_CLIENT_PID_GSYNCD) &&
-         IA_ISDIR(loc->inode->ia_type)) ||
-        is_bd) {
-        for (i = 0; i < priv->child_count; i++, trav = trav->next) {
-            STACK_WIND(frame, stripe_setxattr_cbk, trav->xlator,
-                       trav->xlator->fops->setxattr, loc, dict, flags, xdata);
-        }
-    } else {
-        local->wind_count = 1;
-        STACK_WIND(frame, stripe_setxattr_cbk, FIRST_CHILD(this),
-                   FIRST_CHILD(this)->fops->setxattr, loc, dict, flags, xdata);
-    }
-
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(setxattr, frame, -1, op_errno, NULL);
-    return 0;
-}
-
-int
-stripe_fsetxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                     int op_ret, int op_errno, dict_t *xdata)
-{
-    STRIPE_STACK_UNWIND(fsetxattr, frame, op_ret, op_errno, xdata);
-    return 0;
-}
-
-int
-stripe_is_special_key(dict_t *this, char *key, data_t *value, void *data)
-{
-    gf_boolean_t *is_special = NULL;
-
-    if (data == NULL) {
-        goto out;
-    }
-
-    is_special = data;
-
-    if (XATTR_IS_LOCKINFO(key) || XATTR_IS_BD(key))
-        *is_special = _gf_true;
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_fsetxattr_everyone_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                              int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    int call_count = 0;
-    stripe_local_t *local = NULL;
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        call_count = --local->wind_count;
-
-        if (op_ret < 0) {
-            local->op_ret = op_ret;
-            local->op_errno = op_errno;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (call_count == 0) {
-        STRIPE_STACK_UNWIND(fsetxattr, frame, local->op_ret, local->op_errno,
-                            NULL);
-    }
-    return 0;
-}
-
-int
-stripe_fsetxattr_to_everyone(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                             dict_t *dict, int flags, dict_t *xdata)
-{
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    int ret = -1;
-    stripe_local_t *local = NULL;
-
-    priv = this->private;
-
-    local = mem_get0(this->local_pool);
-    if (local == NULL) {
-        goto out;
-    }
-
-    frame->local = local;
-
-    local->wind_count = priv->child_count;
-
-    trav = this->children;
-
-    while (trav) {
-        STACK_WIND(frame, stripe_fsetxattr_everyone_cbk, trav->xlator,
-                   trav->xlator->fops->fsetxattr, fd, dict, flags, xdata);
-        trav = trav->next;
-    }
-
-    ret = 0;
-out:
-    return ret;
-}
-
-static gf_boolean_t
-stripe_fsetxattr_is_special(dict_t *dict)
-{
-    gf_boolean_t is_spl = _gf_false;
-
-    if (dict == NULL) {
-        goto out;
-    }
-
-    dict_foreach(dict, stripe_is_special_key, &is_spl);
-
-out:
-    return is_spl;
-}
-
-int
-stripe_fsetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd, dict_t *dict,
-                 int flags, dict_t *xdata)
-{
-    int32_t op_ret = -1, ret = -1, op_errno = EINVAL;
-    gf_boolean_t is_spl = _gf_false;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-
-    GF_IF_INTERNAL_XATTR_GOTO("trusted.*stripe*", dict, op_errno, err);
-
-    is_spl = stripe_fsetxattr_is_special(dict);
-    if (is_spl) {
-        ret = stripe_fsetxattr_to_everyone(frame, this, fd, dict, flags, xdata);
-        if (ret < 0) {
-            op_errno = ENOMEM;
-            goto err;
-        }
-
-        goto out;
-    }
-
-    STACK_WIND(frame, stripe_fsetxattr_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->fsetxattr, fd, dict, flags, xdata);
-out:
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fsetxattr, frame, op_ret, op_errno, NULL);
-    return 0;
-}
-
-int
-stripe_removexattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                       int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    STRIPE_STACK_UNWIND(removexattr, frame, op_ret, op_errno, xdata);
-    return 0;
-}
-
-int
-stripe_removexattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
-                   const char *name, dict_t *xdata)
-{
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(this, err);
-
-    GF_IF_NATIVE_XATTR_GOTO("trusted.*stripe*", name, op_errno, err);
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(loc, err);
-
-    STACK_WIND(frame, stripe_removexattr_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->removexattr, loc, name, xdata);
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(removexattr, frame, -1, op_errno, NULL);
-    return 0;
-}
-
-int
-stripe_fremovexattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                        int32_t op_ret, int32_t op_errno, dict_t *xdata)
-{
-    STRIPE_STACK_UNWIND(fremovexattr, frame, op_ret, op_errno, xdata);
-    return 0;
-}
-
-int
-stripe_fremovexattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                    const char *name, dict_t *xdata)
-{
-    int32_t op_ret = -1;
-    int32_t op_errno = EINVAL;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-
-    GF_IF_NATIVE_XATTR_GOTO("trusted.*stripe*", name, op_errno, err);
-
-    STACK_WIND(frame, stripe_fremovexattr_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->fremovexattr, fd, name, xdata);
-    return 0;
-err:
-    STRIPE_STACK_UNWIND(fremovexattr, frame, op_ret, op_errno, xdata);
-    return 0;
-}
-
-int32_t
-stripe_readdirp_lookup_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                           int op_ret, int op_errno, inode_t *inode,
-                           struct iatt *stbuf, dict_t *xattr,
-                           struct iatt *parent)
-{
-    stripe_local_t *local = NULL;
-    call_frame_t *main_frame = NULL;
-    stripe_local_t *main_local = NULL;
-    gf_dirent_t *entry = NULL;
-    call_frame_t *prev = NULL;
-    int done = 0;
-
-    local = frame->local;
-    prev = cookie;
-
-    entry = local->dirent;
-
-    main_frame = local->orig_frame;
-    main_local = main_frame->local;
-    LOCK(&frame->lock);
-    {
-        local->call_count--;
-        if (!local->call_count)
-            done = 1;
-        if (op_ret == -1) {
-            local->op_errno = op_errno;
-            local->op_ret = op_ret;
-            goto unlock;
-        }
-
-        if (stripe_ctx_handle(this, prev, local, xattr))
-            gf_log(this->name, GF_LOG_ERROR,
-                   "Error getting fctx info from dict.");
-
-        correct_file_size(stbuf, local->fctx, prev);
-
-        stripe_iatt_merge(stbuf, &entry->d_stat);
-        local->stbuf_blocks += stbuf->ia_blocks;
-    }
-unlock:
-    UNLOCK(&frame->lock);
-
-    if (done) {
-        inode_ctx_put(entry->inode, this, (uint64_t)(long)local->fctx);
-
-        done = 0;
-        LOCK(&main_frame->lock);
-        {
-            main_local->wind_count--;
-            if (!main_local->wind_count)
-                done = 1;
-            if (local->op_ret == -1) {
-                main_local->op_errno = local->op_errno;
-                main_local->op_ret = local->op_ret;
-            }
-            entry->d_stat.ia_blocks = local->stbuf_blocks;
-        }
-        UNLOCK(&main_frame->lock);
-        if (done) {
-            main_frame->local = NULL;
-            STRIPE_STACK_UNWIND(readdir, main_frame, main_local->op_ret,
-                                main_local->op_errno, &main_local->entries,
-                                NULL);
-            gf_dirent_free(&main_local->entries);
-            stripe_local_wipe(main_local);
-            mem_put(main_local);
-        }
-        frame->local = NULL;
-        stripe_local_wipe(local);
-        mem_put(local);
-        STRIPE_STACK_DESTROY(frame);
-    }
-
-    return 0;
-}
-
-int32_t
-stripe_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int32_t op_ret, int32_t op_errno, gf_dirent_t *orig_entries,
-                    dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    call_frame_t *prev = NULL;
-    gf_dirent_t *local_entry = NULL;
-    gf_dirent_t *tmp_entry = NULL;
-    xlator_list_t *trav = NULL;
-    loc_t loc = {
-        0,
-    };
-    int32_t count = 0;
-    stripe_private_t *priv = NULL;
-    int32_t subvols = 0;
-    dict_t *xattrs = NULL;
-    call_frame_t *local_frame = NULL;
-    stripe_local_t *local_ent = NULL;
-
-    if (!this || !frame->local || !cookie) {
-        gf_log("stripe", GF_LOG_DEBUG, "possible NULL deref");
-        goto out;
-    }
-    prev = cookie;
-    local = frame->local;
-    trav = this->children;
-    priv = this->private;
-
-    subvols = priv->child_count;
-
-    LOCK(&frame->lock);
-    {
-        local->op_errno = op_errno;
-        local->op_ret = op_ret;
-
-        if (op_ret != -1) {
-            list_splice_init(&orig_entries->list, &local->entries.list);
-            local->wind_count = op_ret;
-        }
-    }
-    UNLOCK(&frame->lock);
-
-    if (op_ret == -1) {
-        gf_log(this->name, GF_LOG_WARNING, "%s returned error %s",
-               prev->this->name, strerror(op_errno));
-        goto out;
-    }
-
-    xattrs = dict_new();
-    if (xattrs)
-        (void)stripe_xattr_request_build(this, xattrs, 0, 0, 0, 0);
-    count = op_ret;
-    list_for_each_entry_safe(local_entry, tmp_entry, (&local->entries.list),
-                             list)
-    {
-        if (!local_entry)
-            break;
-        if (!IA_ISREG(local_entry->d_stat.ia_type) || !local_entry->inode) {
-            LOCK(&frame->lock);
-            {
-                local->wind_count--;
-                count = local->wind_count;
-            }
-            UNLOCK(&frame->lock);
-            continue;
-        }
-
-        local_frame = copy_frame(frame);
-
-        if (!local_frame) {
-            op_errno = ENOMEM;
-            op_ret = -1;
-            goto out;
-        }
-
-        local_ent = mem_get0(this->local_pool);
-        if (!local_ent) {
-            op_errno = ENOMEM;
-            op_ret = -1;
-            goto out;
-        }
-
-        loc.inode = inode_ref(local_entry->inode);
-
-        gf_uuid_copy(loc.gfid, local_entry->d_stat.ia_gfid);
-
-        local_ent->orig_frame = frame;
-
-        local_ent->call_count = subvols;
-
-        local_ent->dirent = local_entry;
-
-        local_frame->local = local_ent;
-
-        trav = this->children;
-        while (trav) {
-            STACK_WIND(local_frame, stripe_readdirp_lookup_cbk, trav->xlator,
-                       trav->xlator->fops->lookup, &loc, xattrs);
-            trav = trav->next;
-        }
-        loc_wipe(&loc);
-    }
-out:
-    if (!count) {
-        /* all entries are directories */
-        frame->local = NULL;
-        STRIPE_STACK_UNWIND(readdir, frame, (local ? local->op_ret : -1),
-                            (local ? local->op_errno : EINVAL),
-                            (local ? &local->entries : NULL), NULL);
-        gf_dirent_free(&local->entries);
-        stripe_local_wipe(local);
-        mem_put(local);
-    }
-    if (xattrs)
-        dict_unref(xattrs);
-    return 0;
-}
-int32_t
-stripe_readdirp(call_frame_t *frame, xlator_t *this, fd_t *fd, size_t size,
-                off_t off, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    xlator_list_t *trav = NULL;
-    int op_errno = -1;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(fd, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    if (priv->first_child_down) {
-        op_errno = ENOTCONN;
-        goto err;
-    }
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    frame->local = local;
-
-    local->fd = fd_ref(fd);
-
-    local->wind_count = 0;
-
-    local->count = 0;
-    local->op_ret = -1;
-    INIT_LIST_HEAD(&local->entries);
-
-    if (!trav)
-        goto err;
-
-    STACK_WIND(frame, stripe_readdirp_cbk, trav->xlator,
-               trav->xlator->fops->readdirp, fd, size, off, xdata);
-    return 0;
-err:
-    op_errno = (op_errno == -1) ? errno : op_errno;
-    STRIPE_STACK_UNWIND(readdir, frame, -1, op_errno, NULL, NULL);
-
-    return 0;
-}
-
-int32_t
-mem_acct_init(xlator_t *this)
-{
-    int ret = -1;
-
-    if (!this)
-        goto out;
-
-    ret = xlator_mem_acct_init(this, gf_stripe_mt_end + 1);
-
-    if (ret != 0) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "Memory accounting init"
-               "failed");
-        goto out;
-    }
-
-out:
-    return ret;
-}
-
-static int
-clear_pattern_list(stripe_private_t *priv)
-{
-    struct stripe_options *prev = NULL;
-    struct stripe_options *trav = NULL;
-    int ret = -1;
-
-    GF_VALIDATE_OR_GOTO("stripe", priv, out);
-
-    trav = priv->pattern;
-    priv->pattern = NULL;
-    while (trav) {
-        prev = trav;
-        trav = trav->next;
-        GF_FREE(prev);
-    }
-
-    ret = 0;
-out:
-    return ret;
-}
-
-int
-reconfigure(xlator_t *this, dict_t *options)
-{
-    stripe_private_t *priv = NULL;
-    data_t *data = NULL;
-    int ret = -1;
-    volume_option_t *opt = NULL;
-
-    GF_ASSERT(this);
-    GF_ASSERT(this->private);
-
-    priv = this->private;
-
-    ret = 0;
-    LOCK(&priv->lock);
-    {
-        ret = clear_pattern_list(priv);
-        if (ret)
-            goto unlock;
-
-        data = dict_get(options, "block-size");
-        if (data) {
-            ret = set_stripe_block_size(this, priv, data->data);
-            if (ret)
-                goto unlock;
-        } else {
-            opt = xlator_volume_option_get(this, "block-size");
-            if (!opt) {
-                gf_log(this->name, GF_LOG_WARNING,
-                       "option 'block-size' not found");
-                ret = -1;
-                goto unlock;
-            }
-
-            if (gf_string2bytesize_uint64(opt->default_value,
-                                          &priv->block_size)) {
-                gf_log(this->name, GF_LOG_ERROR,
-                       "Unable to set default block-size ");
-                ret = -1;
-                goto unlock;
-            }
-        }
-
-        GF_OPTION_RECONF("coalesce", priv->coalesce, options, bool, unlock);
-    }
-unlock:
-    UNLOCK(&priv->lock);
-    if (ret)
-        goto out;
-
-    ret = 0;
-out:
-    return ret;
-}
-
-/**
- * init - This function is called when xlator-graph gets initialized.
- *     The option given in volfiles are parsed here.
- * @this -
- */
-int32_t
-init(xlator_t *this)
-{
-    stripe_private_t *priv = NULL;
-    volume_option_t *opt = NULL;
-    xlator_list_t *trav = NULL;
-    data_t *data = NULL;
-    int32_t count = 0;
-    int ret = -1;
-
-    if (!this)
-        goto out;
-
-    trav = this->children;
-    while (trav) {
-        count++;
-        trav = trav->next;
-    }
-
-    if (!count) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "stripe configured without \"subvolumes\" option. "
-               "exiting");
-        goto out;
-    }
-
-    if (!this->parents) {
-        gf_log(this->name, GF_LOG_WARNING, "dangling volume. check volfile ");
-    }
-
-    if (count == 1) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "stripe configured with only one \"subvolumes\" option."
-               " please check the volume. exiting");
-        goto out;
-    }
-
-    priv = GF_CALLOC(1, sizeof(stripe_private_t),
-                     gf_stripe_mt_stripe_private_t);
-
-    if (!priv)
-        goto out;
-    priv->xl_array = GF_CALLOC(count, sizeof(xlator_t *),
-                               gf_stripe_mt_xlator_t);
-    if (!priv->xl_array)
-        goto out;
-
-    priv->last_event = GF_CALLOC(count, sizeof(int), gf_stripe_mt_int32_t);
-    if (!priv->last_event)
-        goto out;
-
-    priv->child_count = count;
-    LOCK_INIT(&priv->lock);
-
-    trav = this->children;
-    count = 0;
-    while (trav) {
-        priv->xl_array[count++] = trav->xlator;
-        trav = trav->next;
-    }
-
-    if (count > 256) {
-        gf_log(this->name, GF_LOG_ERROR,
-               "maximum number of stripe subvolumes supported "
-               "is 256");
-        goto out;
-    }
-
-    ret = 0;
-    LOCK(&priv->lock);
-    {
-        opt = xlator_volume_option_get(this, "block-size");
-        if (!opt) {
-            gf_log(this->name, GF_LOG_WARNING, "option 'block-size' not found");
-            ret = -1;
-            goto unlock;
-        }
-        if (gf_string2bytesize_uint64(opt->default_value, &priv->block_size)) {
-            gf_log(this->name, GF_LOG_ERROR,
-                   "Unable to set default block-size ");
-            ret = -1;
-            goto unlock;
-        }
-        /* option stripe-pattern *avi:1GB,*pdf:16K */
-        data = dict_get(this->options, "block-size");
-        if (data) {
-            ret = set_stripe_block_size(this, priv, data->data);
-            if (ret)
-                goto unlock;
-        }
-    }
-unlock:
-    UNLOCK(&priv->lock);
-    if (ret)
-        goto out;
-
-    GF_OPTION_INIT("use-xattr", priv->xattr_supported, bool, out);
-    /* notify related */
-    priv->nodes_down = priv->child_count;
-
-    GF_OPTION_INIT("coalesce", priv->coalesce, bool, out);
-
-    this->local_pool = mem_pool_new(stripe_local_t, 128);
-    if (!this->local_pool) {
-        ret = -1;
-        gf_log(this->name, GF_LOG_ERROR,
-               "failed to create local_t's memory pool");
-        goto out;
-    }
-
-    this->private = priv;
-
-    ret = 0;
-out:
-    if (ret) {
-        if (priv) {
-            GF_FREE(priv->xl_array);
-            GF_FREE(priv);
-        }
-    }
-    return ret;
-}
-
-/**
- * fini -   Free all the private variables
- * @this -
- */
-void
-fini(xlator_t *this)
-{
-    stripe_private_t *priv = NULL;
-    struct stripe_options *prev = NULL;
-    struct stripe_options *trav = NULL;
-
-    if (!this)
-        goto out;
-
-    priv = this->private;
-    if (priv) {
-        this->private = NULL;
-        GF_FREE(priv->xl_array);
-
-        trav = priv->pattern;
-        while (trav) {
-            prev = trav;
-            trav = trav->next;
-            GF_FREE(prev);
-        }
-        GF_FREE(priv->last_event);
-        LOCK_DESTROY(&priv->lock);
-        GF_FREE(priv);
-    }
-
-out:
-    return;
-}
-
-int32_t
-stripe_getxattr_unwind(call_frame_t *frame, int op_ret, int op_errno,
-                       dict_t *dict, dict_t *xdata)
-
-{
-    STRIPE_STACK_UNWIND(getxattr, frame, op_ret, op_errno, dict, xdata);
-    return 0;
-}
-
-int
-stripe_internal_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                             int op_ret, int op_errno, dict_t *xattr,
-                             dict_t *xdata)
-{
-    char size_key[256] = {
-        0,
-    };
-    char index_key[256] = {
-        0,
-    };
-    char count_key[256] = {
-        0,
-    };
-    char coalesce_key[256] = {
-        0,
-    };
-
-    VALIDATE_OR_GOTO(frame, out);
-    VALIDATE_OR_GOTO(frame->local, out);
-
-    if (!xattr || (op_ret == -1))
-        goto out;
-
-    sprintf(size_key, "trusted.%s.stripe-size", this->name);
-    sprintf(count_key, "trusted.%s.stripe-count", this->name);
-    sprintf(index_key, "trusted.%s.stripe-index", this->name);
-    sprintf(coalesce_key, "trusted.%s.stripe-coalesce", this->name);
-
-    dict_del(xattr, size_key);
-    dict_del(xattr, count_key);
-    dict_del(xattr, index_key);
-    dict_del(xattr, coalesce_key);
-
-out:
-    STRIPE_STACK_UNWIND(getxattr, frame, op_ret, op_errno, xattr, xdata);
-
-    return 0;
-}
-
-int
-stripe_getxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                    int op_ret, int op_errno, dict_t *xattr, dict_t *xdata)
-{
-    int call_cnt = 0;
-    stripe_local_t *local = NULL;
-
-    VALIDATE_OR_GOTO(frame, out);
-    VALIDATE_OR_GOTO(frame->local, out);
-
-    local = frame->local;
-
-    LOCK(&frame->lock);
-    {
-        call_cnt = --local->wind_count;
-    }
-    UNLOCK(&frame->lock);
-
-    if (!xattr || (op_ret < 0))
-        goto out;
-
-    local->op_ret = 0;
-
-    if (!local->xattr) {
-        local->xattr = dict_ref(xattr);
-    } else {
-        stripe_aggregate_xattr(local->xattr, xattr);
-    }
-
-out:
-    if (!call_cnt) {
-        STRIPE_STACK_UNWIND(getxattr, frame, (local ? local->op_ret : -1),
-                            op_errno, (local ? local->xattr : NULL), xdata);
-    }
-
-    return 0;
-}
-
-int32_t
-stripe_vgetxattr_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
-                     int32_t op_ret, int32_t op_errno, dict_t *dict,
-                     dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    int32_t callcnt = 0;
-    int32_t ret = -1;
-    long cky = 0;
-    void *xattr_val = NULL;
-    void *xattr_serz = NULL;
-    stripe_xattr_sort_t *xattr = NULL;
-    dict_t *stripe_xattr = NULL;
-
-    if (!frame || !frame->local || !this) {
-        gf_log("", GF_LOG_ERROR, "Possible NULL deref");
-        return ret;
-    }
-
-    local = frame->local;
-    cky = (long)cookie;
-
-    if (local->xsel[0] == '\0') {
-        gf_log(this->name, GF_LOG_ERROR, "Empty xattr in cbk");
-        return ret;
-    }
-
-    LOCK(&frame->lock);
-    {
-        callcnt = --local->wind_count;
-
-        if (!dict || (op_ret < 0))
-            goto out;
-
-        if (!local->xattr_list)
-            local->xattr_list = (stripe_xattr_sort_t *)GF_CALLOC(
-                local->nallocs, sizeof(stripe_xattr_sort_t),
-                gf_stripe_mt_xattr_sort_t);
-
-        if (local->xattr_list) {
-            xattr = local->xattr_list + (int32_t)cky;
-
-            ret = dict_get_ptr_and_len(dict, local->xsel, &xattr_val,
-                                       &xattr->xattr_len);
-            if (xattr->xattr_len == 0)
-                goto out;
-
-            xattr->pos = cky;
-            xattr->xattr_value = gf_memdup(xattr_val, xattr->xattr_len);
-
-            if (xattr->xattr_value != NULL)
-                local->xattr_total_len += xattr->xattr_len + 1;
-        }
-    }
-out:
-    UNLOCK(&frame->lock);
-
-    if (!callcnt) {
-        if (!local->xattr_total_len)
-            goto unwind;
-
-        stripe_xattr = dict_new();
-        if (!stripe_xattr)
-            goto unwind;
-
-        /* select filler based on ->xsel */
-        if (XATTR_IS_PATHINFO(local->xsel))
-            ret = stripe_fill_pathinfo_xattr(this, local, (char **)&xattr_serz);
-        else if (XATTR_IS_LOCKINFO(local->xsel)) {
-            ret = stripe_fill_lockinfo_xattr(this, local, &xattr_serz);
-        } else {
-            gf_log(this->name, GF_LOG_WARNING,
-                   "Unknown xattr in xattr request");
-            goto unwind;
-        }
-
-        if (!ret) {
-            ret = dict_set_dynptr(stripe_xattr, local->xsel, xattr_serz,
-                                  local->xattr_total_len);
-            if (ret)
-                gf_log(this->name, GF_LOG_ERROR, "Can't set %s key in dict",
-                       local->xsel);
-        }
-
-    unwind:
-        /*
-         * Among other things, STRIPE_STACK_UNWIND will free "local"
-         * for us.  That means we can't dereference it afterward.
-         * Fortunately, the actual result is in stripe_xattr now, so we
-         * can simply clean up before unwinding.
-         */
-        ret = stripe_free_xattr_str(local);
-        GF_FREE(local->xattr_list);
-        local->xattr_list = NULL;
-
-        STRIPE_STACK_UNWIND(getxattr, frame, op_ret, op_errno, stripe_xattr,
-                            NULL);
-
-        if (stripe_xattr)
-            dict_unref(stripe_xattr);
-    }
-
-    return ret;
-}
-
-int
-stripe_marker_populate_args(call_frame_t *frame, int type, int *gauge,
-                            xlator_t **subvols)
-{
-    xlator_t *this = frame->this;
-    stripe_private_t *priv = this->private;
-    stripe_local_t *local = frame->local;
-    int count = 0;
-
-    count = priv->child_count;
-    if (MARKER_XTIME_TYPE == type) {
-        if (!IA_FILE_OR_DIR(local->loc.inode->ia_type))
-            count = 1;
-    }
-    memcpy(subvols, priv->xl_array, sizeof(*subvols) * count);
-
-    return count;
-}
-
-int32_t
-stripe_getxattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
-                const char *name, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    xlator_list_t *trav = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t op_errno = EINVAL;
-    int i = 0;
-    int ret = 0;
-
-    VALIDATE_OR_GOTO(frame, err);
-    VALIDATE_OR_GOTO(this, err);
-    VALIDATE_OR_GOTO(loc, err);
-    VALIDATE_OR_GOTO(loc->path, err);
-    VALIDATE_OR_GOTO(loc->inode, err);
-
-    priv = this->private;
-    trav = this->children;
-
-    /* Initialization */
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-    local->op_ret = -1;
-    frame->local = local;
-    loc_copy(&local->loc, loc);
-
-    if (name && strncmp(name, QUOTA_SIZE_KEY, SLEN(QUOTA_SIZE_KEY)) == 0) {
-        local->wind_count = priv->child_count;
-
-        for (i = 0, trav = this->children; i < priv->child_count;
-             i++, trav = trav->next) {
-            STACK_WIND(frame, stripe_getxattr_cbk, trav->xlator,
-                       trav->xlator->fops->getxattr, loc, name, xdata);
-        }
-
-        return 0;
-    }
-
-    if (name && (XATTR_IS_PATHINFO(name))) {
-        if (IA_ISREG(loc->inode->ia_type)) {
-            ret = inode_ctx_get(loc->inode, this, (uint64_t *)&local->fctx);
-            if (ret)
-                gf_log(this->name, GF_LOG_ERROR,
-                       "stripe size unavailable from fctx"
-                       " relying on pathinfo could lead to"
-                       " wrong results");
-        }
-
-        local->nallocs = local->wind_count = priv->child_count;
-        (void)strncpy(local->xsel, name, strlen(name));
-
-        /**
-         * for xattrs that need info from all children, fill ->xsel
-         * as above and call the filler function in cbk based on
-         * it
-         */
-        for (i = 0, trav = this->children; i < priv->child_count;
-             i++, trav = trav->next) {
-            STACK_WIND_COOKIE(frame, stripe_vgetxattr_cbk, (void *)(long)i,
-                              trav->xlator, trav->xlator->fops->getxattr, loc,
-                              name, xdata);
-        }
-
-        return 0;
-    }
-
-    if (cluster_handle_marker_getxattr(frame, loc, name, priv->vol_uuid,
-                                       stripe_getxattr_unwind,
-                                       stripe_marker_populate_args) == 0)
-        return 0;
-
-    STACK_WIND(frame, stripe_internal_getxattr_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->getxattr, loc, name, xdata);
-
-    return 0;
-
-err:
-    STRIPE_STACK_UNWIND(getxattr, frame, -1, op_errno, NULL, NULL);
-    return 0;
-}
-
-static gf_boolean_t
-stripe_is_special_xattr(const char *name)
-{
-    gf_boolean_t is_spl = _gf_false;
-
-    if (!name) {
-        goto out;
-    }
-
-    if (!strncmp(name, GF_XATTR_LOCKINFO_KEY, SLEN(GF_XATTR_LOCKINFO_KEY)) ||
-        XATTR_IS_PATHINFO(name))
-        is_spl = _gf_true;
-out:
-    return is_spl;
-}
-
-int32_t
-stripe_fgetxattr_from_everyone(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                               const char *name, dict_t *xdata)
-{
-    stripe_local_t *local = NULL;
-    stripe_private_t *priv = NULL;
-    int32_t ret = -1, op_errno = 0;
-    int i = 0;
-    xlator_list_t *trav = NULL;
-
-    priv = this->private;
-
-    local = mem_get0(this->local_pool);
-    if (!local) {
-        op_errno = ENOMEM;
-        goto err;
-    }
-
-    local->op_ret = -1;
-    frame->local = local;
-
-    strncpy(local->xsel, name, strlen(name));
-    local->nallocs = local->wind_count = priv->child_count;
-
-    for (i = 0, trav = this->children; i < priv->child_count;
-         i++, trav = trav->next) {
-        STACK_WIND_COOKIE(frame, stripe_vgetxattr_cbk, (void *)(long)i,
-                          trav->xlator, trav->xlator->fops->fgetxattr, fd, name,
-                          xdata);
-    }
-
-    return 0;
-
-err:
-    STACK_UNWIND_STRICT(fgetxattr, frame, -1, op_errno, NULL, NULL);
-    return ret;
-}
-
-int32_t
-stripe_fgetxattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
-                 const char *name, dict_t *xdata)
-{
-    if (stripe_is_special_xattr(name)) {
-        stripe_fgetxattr_from_everyone(frame, this, fd, name, xdata);
-        goto out;
-    }
-
-    STACK_WIND(frame, stripe_internal_getxattr_cbk, FIRST_CHILD(this),
-               FIRST_CHILD(this)->fops->fgetxattr, fd, name, xdata);
-
-out:
-    return 0;
-}
-
-int32_t
-stripe_priv_dump(xlator_t *this)
-{
-    char key[GF_DUMP_MAX_BUF_LEN];
-    int i = 0;
-    stripe_private_t *priv = NULL;
-    int ret = -1;
-    struct stripe_options *options = NULL;
-
-    GF_VALIDATE_OR_GOTO("stripe", this, out);
-
-    priv = this->private;
-    if (!priv)
-        goto out;
-
-    ret = TRY_LOCK(&priv->lock);
-    if (ret != 0)
-        goto out;
-
-    gf_proc_dump_add_section("xlator.cluster.stripe.%s.priv", this->name);
-    gf_proc_dump_write("child_count", "%d", priv->child_count);
-
-    for (i = 0; i < priv->child_count; i++) {
-        sprintf(key, "subvolumes[%d]", i);
-        gf_proc_dump_write(key, "%s.%s", priv->xl_array[i]->type,
-                           priv->xl_array[i]->name);
-    }
-
-    options = priv->pattern;
-    while (options != NULL) {
-        gf_proc_dump_write("path_pattern", "%s", priv->pattern->path_pattern);
-        gf_proc_dump_write("options_block_size", "%ul", options->block_size);
-
-        options = options->next;
-    }
-
-    gf_proc_dump_write("block_size", "%ul", priv->block_size);
-    gf_proc_dump_write("nodes-down", "%d", priv->nodes_down);
-    gf_proc_dump_write("first-child_down", "%d", priv->first_child_down);
-    gf_proc_dump_write("xattr_supported", "%d", priv->xattr_supported);
-
-    UNLOCK(&priv->lock);
-
-out:
-    return ret;
-}
-
-struct xlator_fops fops = {
-    .stat = stripe_stat,
-    .unlink = stripe_unlink,
-    .rename = stripe_rename,
-    .link = stripe_link,
-    .truncate = stripe_truncate,
-    .create = stripe_create,
-    .open = stripe_open,
-    .readv = stripe_readv,
-    .writev = stripe_writev,
-    .statfs = stripe_statfs,
-    .flush = stripe_flush,
-    .fsync = stripe_fsync,
-    .ftruncate = stripe_ftruncate,
-    .fstat = stripe_fstat,
-    .mkdir = stripe_mkdir,
-    .rmdir = stripe_rmdir,
-    .lk = stripe_lk,
-    .opendir = stripe_opendir,
-    .fsyncdir = stripe_fsyncdir,
-    .setattr = stripe_setattr,
-    .fsetattr = stripe_fsetattr,
-    .lookup = stripe_lookup,
-    .mknod = stripe_mknod,
-    .setxattr = stripe_setxattr,
-    .fsetxattr = stripe_fsetxattr,
-    .getxattr = stripe_getxattr,
-    .fgetxattr = stripe_fgetxattr,
-    .removexattr = stripe_removexattr,
-    .fremovexattr = stripe_fremovexattr,
-    .readdirp = stripe_readdirp,
-    .fallocate = stripe_fallocate,
-    .discard = stripe_discard,
-    .zerofill = stripe_zerofill,
-    .seek = stripe_seek,
-};
-
-struct xlator_cbks cbks = {
-    .release = stripe_release,
-    .forget = stripe_forget,
-};
-
-struct xlator_dumpops dumpops = {
-    .priv = stripe_priv_dump,
-};
-
-struct volume_options options[] = {
-    {
-        .key = {"block-size"},
-        .type = GF_OPTION_TYPE_SIZE_LIST,
-        .default_value = "128KB",
-        .min = STRIPE_MIN_BLOCK_SIZE,
-        .description = "Size of the stripe unit that would be read "
-                       "from or written to the striped servers.",
-        .op_version = {1},
-        .tags = {"stripe"},
-        .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE,
-    },
-    {
-        .key = {"use-xattr"},
-        .type = GF_OPTION_TYPE_BOOL,
-        .default_value = "true",
-        .description = "handle the stripe without the xattr",
-        .tags = {"stripe", "dev-only"},
-        .flags = OPT_FLAG_CLIENT_OPT,
-    },
-    {
-        .key = {"coalesce"},
-        .type = GF_OPTION_TYPE_BOOL,
-        .default_value = "true",
-        .description = "Enable/Disable coalesce mode to flatten striped "
-                       "files as stored on the server (i.e., eliminate holes "
-                       "caused by the traditional format).",
-        .op_version = {1},
-        .flags = OPT_FLAG_CLIENT_OPT | OPT_FLAG_SETTABLE,
-        .tags = {"stripe"},
-    },
-    {.key = {NULL}},
-};
diff --git a/xlators/cluster/stripe/src/stripe.h b/xlators/cluster/stripe/src/stripe.h
deleted file mode 100644
index 103c96491ff..00000000000
--- a/xlators/cluster/stripe/src/stripe.h
+++ /dev/null
@@ -1,291 +0,0 @@
-/*
-  Copyright (c) 2008-2012 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#ifndef _STRIPE_H_
-#define _STRIPE_H_
-
-#include "xlator.h"
-#include "logging.h"
-#include "defaults.h"
-#include "common-utils.h"
-#include "compat.h"
-#include "compat-errno.h"
-#include "stripe-mem-types.h"
-#include "libxlator.h"
-#include <fnmatch.h>
-#include <signal.h>
-
-#define STRIPE_PATHINFO_HEADER "STRIPE:"
-#define STRIPE_MIN_BLOCK_SIZE (16 * GF_UNIT_KB)
-
-#define STRIPE_STACK_UNWIND(fop, frame, params...)                             \
-    do {                                                                       \
-        stripe_local_t *__local = NULL;                                        \
-        if (frame) {                                                           \
-            __local = frame->local;                                            \
-            frame->local = NULL;                                               \
-        }                                                                      \
-        STACK_UNWIND_STRICT(fop, frame, params);                               \
-        if (__local) {                                                         \
-            stripe_local_wipe(__local);                                        \
-            mem_put(__local);                                                  \
-        }                                                                      \
-    } while (0)
-
-#define STRIPE_STACK_DESTROY(frame)                                            \
-    do {                                                                       \
-        stripe_local_t *__local = NULL;                                        \
-        __local = frame->local;                                                \
-        frame->local = NULL;                                                   \
-        STACK_DESTROY(frame->root);                                            \
-        if (__local) {                                                         \
-            stripe_local_wipe(__local);                                        \
-            mem_put(__local);                                                  \
-        }                                                                      \
-    } while (0)
-
-#define STRIPE_VALIDATE_FCTX(fctx, label)                                      \
-    do {                                                                       \
-        int idx = 0;                                                           \
-        if (!fctx) {                                                           \
-            op_errno = EINVAL;                                                 \
-            goto label;                                                        \
-        }                                                                      \
-        for (idx = 0; idx < fctx->stripe_count; idx++) {                       \
-            if (!fctx->xl_array[idx]) {                                        \
-                gf_log(this->name, GF_LOG_ERROR, "fctx->xl_array[%d] is NULL", \
-                       idx);                                                   \
-                op_errno = ESTALE;                                             \
-                goto label;                                                    \
-            }                                                                  \
-        }                                                                      \
-    } while (0)
-
-typedef struct stripe_xattr_sort {
-    int pos;
-    int xattr_len;
-    char *xattr_value;
-} stripe_xattr_sort_t;
-
-/**
- * struct stripe_options : This keeps the pattern and the block-size
- *     information, which is used for striping on a file.
- */
-struct stripe_options {
-    struct stripe_options *next;
-    char path_pattern[256];
-    uint64_t block_size;
-};
-
-/**
- * Private structure for stripe translator
- */
-struct stripe_private {
-    struct stripe_options *pattern;
-    xlator_t **xl_array;
-    uint64_t block_size;
-    gf_lock_t lock;
-    uint8_t nodes_down;
-    int8_t first_child_down;
-    int *last_event;
-    int8_t child_count;
-    gf_boolean_t xattr_supported; /* default yes */
-    gf_boolean_t coalesce;
-    char vol_uuid[UUID_SIZE + 1];
-};
-
-/**
- * Used to keep info about the replies received from readv/writev calls
- */
-struct stripe_replies {
-    struct iovec *vector;
-    int32_t count;   // count of vector
-    int32_t op_ret;  // op_ret of readv
-    int32_t op_errno;
-    int32_t requested_size;
-    struct iatt stbuf; /* 'stbuf' is also a part of reply */
-};
-
-typedef struct _stripe_fd_ctx {
-    off_t stripe_size;
-    int stripe_count;
-    int stripe_coalesce;
-    int static_array;
-    xlator_t **xl_array;
-} stripe_fd_ctx_t;
-
-/**
- * Local structure to be passed with all the frames in case of STACK_WIND
- */
-struct stripe_local; /* this itself is used inside the structure; */
-
-struct stripe_local {
-    struct stripe_local *next;
-    call_frame_t *orig_frame;
-
-    stripe_fd_ctx_t *fctx;
-
-    /* Used by _cbk functions */
-    struct iatt stbuf;
-    struct iatt pre_buf;
-    struct iatt post_buf;
-    struct iatt preparent;
-    struct iatt postparent;
-
-    off_t stbuf_size;
-    off_t prebuf_size;
-    off_t postbuf_size;
-    off_t preparent_size;
-    off_t postparent_size;
-
-    blkcnt_t stbuf_blocks;
-    blkcnt_t prebuf_blocks;
-    blkcnt_t postbuf_blocks;
-    blkcnt_t preparent_blocks;
-    blkcnt_t postparent_blocks;
-
-    struct stripe_replies *replies;
-    struct statvfs statvfs_buf;
-    dir_entry_t *entry;
-
-    int8_t revalidate;
-    int8_t failed;
-    int8_t unwind;
-
-    size_t readv_size;
-    int32_t entry_count;
-    int32_t node_index;
-    int32_t call_count;
-    int32_t wind_count; /* used instead of child_cound
-                           in case of read and write */
-    int32_t op_ret;
-    int32_t op_errno;
-    int32_t count;
-    int32_t flags;
-    char *name;
-    inode_t *inode;
-
-    loc_t loc;
-    loc_t loc2;
-
-    mode_t mode;
-    dev_t rdev;
-    /* For File I/O fops */
-    dict_t *xdata;
-
-    stripe_xattr_sort_t *xattr_list;
-    int32_t xattr_total_len;
-    int32_t nallocs;
-    char xsel[256];
-
-    /* General usage */
-    off_t offset;
-    off_t stripe_size;
-
-    int xattr_self_heal_needed;
-    int entry_self_heal_needed;
-
-    int8_t *list;
-    struct gf_flock lock;
-    fd_t *fd;
-    void *value;
-    struct iobref *iobref;
-    gf_dirent_t entries;
-    gf_dirent_t *dirent;
-    dict_t *xattr;
-    uuid_t ia_gfid;
-
-    int xflag;
-    mode_t umask;
-};
-
-typedef struct stripe_local stripe_local_t;
-typedef struct stripe_private stripe_private_t;
-
-/*
- * Determine the stripe index of a particular frame based on the translator.
- */
-static inline int32_t
-stripe_get_frame_index(stripe_fd_ctx_t *fctx, call_frame_t *prev)
-{
-    int32_t i, idx = -1;
-
-    for (i = 0; i < fctx->stripe_count; i++) {
-        if (fctx->xl_array[i] == prev->this) {
-            idx = i;
-            break;
-        }
-    }
-
-    return idx;
-}
-
-static inline void
-stripe_copy_xl_array(xlator_t **dst, xlator_t **src, int count)
-{
-    int i;
-
-    for (i = 0; i < count; i++)
-        dst[i] = src[i];
-}
-
-void
-stripe_local_wipe(stripe_local_t *local);
-int32_t
-stripe_ctx_handle(xlator_t *this, call_frame_t *prev, stripe_local_t *local,
-                  dict_t *dict);
-void
-stripe_aggregate_xattr(dict_t *dst, dict_t *src);
-int32_t
-stripe_xattr_request_build(xlator_t *this, dict_t *dict, uint64_t stripe_size,
-                           uint32_t stripe_count, uint32_t stripe_index,
-                           uint32_t stripe_coalesce);
-int32_t
-stripe_get_matching_bs(const char *path, stripe_private_t *priv);
-int
-set_stripe_block_size(xlator_t *this, stripe_private_t *priv, char *data);
-int32_t
-stripe_iatt_merge(struct iatt *from, struct iatt *to);
-int32_t
-stripe_fill_pathinfo_xattr(xlator_t *this, stripe_local_t *local,
-                           char **xattr_serz);
-int32_t
-stripe_free_xattr_str(stripe_local_t *local);
-int32_t
-stripe_xattr_aggregate(char *buffer, stripe_local_t *local, int32_t *total);
-off_t
-coalesced_offset(off_t offset, uint64_t stripe_size, int stripe_count);
-off_t
-uncoalesced_size(off_t size, uint64_t stripe_size, int stripe_count,
-                 int stripe_index);
-int32_t
-stripe_fill_lockinfo_xattr(xlator_t *this, stripe_local_t *local,
-                           void **xattr_serz);
-
-/*
- * Adjust the size attribute for files if coalesce is enabled.
- */
-static inline void
-correct_file_size(struct iatt *buf, stripe_fd_ctx_t *fctx, call_frame_t *prev)
-{
-    int index;
-
-    if (!IA_ISREG(buf->ia_type))
-        return;
-
-    if (!fctx || !fctx->stripe_coalesce)
-        return;
-
-    index = stripe_get_frame_index(fctx, prev);
-    buf->ia_size = uncoalesced_size(buf->ia_size, fctx->stripe_size,
-                                    fctx->stripe_count, index);
-}
-
-#endif /* _STRIPE_H_ */
diff --git a/xlators/debug/delay-gen/src/delay-gen-mem-types.h b/xlators/debug/delay-gen/src/delay-gen-mem-types.h
index 63a15a70da3..c89a9217193 100644
--- a/xlators/debug/delay-gen/src/delay-gen-mem-types.h
+++ b/xlators/debug/delay-gen/src/delay-gen-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __DELAY_GEN_MEM_TYPES_H__
 #define __DELAY_GEN_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_delay_gen_mem_types_ {
     gf_delay_gen_mt_dg_t = gf_common_mt_end + 1,
diff --git a/xlators/debug/delay-gen/src/delay-gen-messages.h b/xlators/debug/delay-gen/src/delay-gen-messages.h
index a9046ca14bf..bc98cec2885 100644
--- a/xlators/debug/delay-gen/src/delay-gen-messages.h
+++ b/xlators/debug/delay-gen/src/delay-gen-messages.h
@@ -11,7 +11,7 @@
 #ifndef __DELAY_GEN_MESSAGES_H__
 #define __DELAY_GEN_MESSAGES_H__
 
-#include "glfs-message-id.h"
+#include <glusterfs/glfs-message-id.h>
 
 /* To add new message IDs, append new identifiers at the end of the list.
  *
diff --git a/xlators/debug/delay-gen/src/delay-gen.c b/xlators/debug/delay-gen/src/delay-gen.c
index a2d02527f23..4698f1fd785 100644
--- a/xlators/debug/delay-gen/src/delay-gen.c
+++ b/xlators/debug/delay-gen/src/delay-gen.c
@@ -27,7 +27,7 @@ delay_gen(xlator_t *this, int fop)
         return 0;
 
     if ((rand() % DELAY_GRANULARITY) < dg->delay_ppm)
-        usleep(dg->delay_duration);
+        gf_nanosleep(dg->delay_duration * GF_US_IN_NS);
 
     return 0;
 }
@@ -679,4 +679,19 @@ struct volume_options options[] = {
         .default_value = "",
     },
 
-    {.key = {NULL}}};
+    {.key = {NULL}},
+};
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .notify = notify,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {GD_OP_VERSION_3_12_0},
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "delay-gen",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/debug/delay-gen/src/delay-gen.h b/xlators/debug/delay-gen/src/delay-gen.h
index 5e4d179f0b4..afa95e5eb2d 100644
--- a/xlators/debug/delay-gen/src/delay-gen.h
+++ b/xlators/debug/delay-gen/src/delay-gen.h
@@ -13,9 +13,9 @@
 
 #include "delay-gen-mem-types.h"
 #include "delay-gen-messages.h"
-#include "glusterfs.h"
-#include "xlator.h"
-#include "defaults.h"
+#include <glusterfs/glusterfs.h>
+#include <glusterfs/xlator.h>
+#include <glusterfs/defaults.h>
 
 typedef struct {
     int enable[GF_FOP_MAXVALUE];
diff --git a/xlators/debug/error-gen/src/error-gen-mem-types.h b/xlators/debug/error-gen/src/error-gen-mem-types.h
index 2facd6b27cb..b9b713af8fc 100644
--- a/xlators/debug/error-gen/src/error-gen-mem-types.h
+++ b/xlators/debug/error-gen/src/error-gen-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __ERROR_GEN_MEM_TYPES_H__
 #define __ERROR_GEN_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_error_gen_mem_types_ {
     gf_error_gen_mt_eg_t = gf_common_mt_end + 1,
diff --git a/xlators/debug/error-gen/src/error-gen.c b/xlators/debug/error-gen/src/error-gen.c
index c6595b4c0e4..d45655ef4c3 100644
--- a/xlators/debug/error-gen/src/error-gen.c
+++ b/xlators/debug/error-gen/src/error-gen.c
@@ -7,10 +7,10 @@
    later), or the GNU General Public License, version 2 (GPLv2), in all
    cases as published by the Free Software Foundation.
 */
-#include "xlator.h"
+#include <glusterfs/xlator.h>
 #include "error-gen.h"
-#include "statedump.h"
-#include "defaults.h"
+#include <glusterfs/statedump.h>
+#include <glusterfs/defaults.h>
 
 /*
  * The user can specify an error probability as a float percentage, but we
@@ -31,9 +31,9 @@
 sys_error_t error_no_list[] = {
     [GF_FOP_LOOKUP] = {.error_no_count = 4,
                        .error_no = {ENOENT, ENOTDIR, ENAMETOOLONG, EAGAIN}},
-    [GF_FOP_STAT] = {.error_no_count = 7,
-                     .error_no = {EACCES, EBADF, EFAULT, ENAMETOOLONG, ENOENT,
-                                  ENOMEM, ENOTDIR}},
+    [GF_FOP_STAT] = {.error_no_count = 6,
+                     .error_no = {EACCES, EFAULT, ENAMETOOLONG, ENOENT, ENOMEM,
+                                  ENOTDIR}},
     [GF_FOP_READLINK] = {.error_no_count = 8,
                          .error_no = {EACCES, EFAULT, EINVAL, EIO, ENAMETOOLONG,
                                       ENOENT, ENOMEM, ENOTDIR}},
@@ -79,21 +79,20 @@ sys_error_t error_no_list[] = {
     [GF_FOP_WRITE] = {.error_no_count = 7,
                       .error_no = {EINVAL, EBADF, EFAULT, EISDIR, ENAMETOOLONG,
                                    ENOSPC, GF_ERROR_SHORT_WRITE}},
-    [GF_FOP_STATFS] = {.error_no_count = 10,
-                       .error_no = {EACCES, EBADF, EFAULT, EINTR, EIO,
-                                    ENAMETOOLONG, ENOENT, ENOMEM, ENOSYS,
-                                    ENOTDIR}},
+    [GF_FOP_STATFS] = {.error_no_count = 9,
+                       .error_no = {EACCES, EFAULT, EINTR, EIO, ENAMETOOLONG,
+                                    ENOENT, ENOMEM, ENOSYS, ENOTDIR}},
     [GF_FOP_FLUSH] = {.error_no_count = 5,
                       .error_no = {EACCES, EFAULT, ENAMETOOLONG, ENOSYS,
                                    ENOENT}},
     [GF_FOP_FSYNC] = {.error_no_count = 4,
                       .error_no = {EBADF, EIO, EROFS, EINVAL}},
-    [GF_FOP_SETXATTR] = {.error_no_count = 4,
-                         .error_no = {EACCES, EBADF, EINTR, ENAMETOOLONG}},
-    [GF_FOP_GETXATTR] = {.error_no_count = 4,
-                         .error_no = {EACCES, EBADF, ENAMETOOLONG, EINTR}},
-    [GF_FOP_REMOVEXATTR] = {.error_no_count = 4,
-                            .error_no = {EACCES, EBADF, ENAMETOOLONG, EINTR}},
+    [GF_FOP_SETXATTR] = {.error_no_count = 3,
+                         .error_no = {EACCES, EINTR, ENAMETOOLONG}},
+    [GF_FOP_GETXATTR] = {.error_no_count = 3,
+                         .error_no = {EACCES, ENAMETOOLONG, EINTR}},
+    [GF_FOP_REMOVEXATTR] = {.error_no_count = 3,
+                            .error_no = {EACCES, ENAMETOOLONG, EINTR}},
     [GF_FOP_FSETXATTR] = {.error_no_count = 4,
                           .error_no = {EACCES, EBADF, EINTR, ENAMETOOLONG}},
     [GF_FOP_FGETXATTR] = {.error_no_count = 4,
@@ -125,26 +124,25 @@ sys_error_t error_no_list[] = {
                                      ENOENT}},
     [GF_FOP_FXATTROP] = {.error_no_count = 4,
                          .error_no = {EBADF, EIO, EROFS, EINVAL}},
-    [GF_FOP_INODELK] = {.error_no_count = 4,
-                        .error_no = {EACCES, EBADF, EINTR, ENAMETOOLONG}},
+    [GF_FOP_INODELK] = {.error_no_count = 3,
+                        .error_no = {EACCES, EINTR, ENAMETOOLONG}},
     [GF_FOP_FINODELK] = {.error_no_count = 4,
                          .error_no = {EACCES, EBADF, EINTR, ENAMETOOLONG}},
-    [GF_FOP_ENTRYLK] = {.error_no_count = 4,
-                        .error_no = {EACCES, EBADF, ENAMETOOLONG, EINTR}},
+    [GF_FOP_ENTRYLK] = {.error_no_count = 3,
+                        .error_no = {EACCES, ENAMETOOLONG, EINTR}},
     [GF_FOP_FENTRYLK] = {.error_no_count = 10,
                          .error_no = {EACCES, EEXIST, EFAULT, EISDIR, EMFILE,
                                       ENAMETOOLONG, ENFILE, ENODEV, ENOENT,
                                       ENOMEM}},
-    [GF_FOP_SETATTR] = {.error_no_count = 11,
+    [GF_FOP_SETATTR] = {.error_no_count = 10,
                         .error_no = {EACCES, EFAULT, EIO, ENAMETOOLONG, ENOENT,
-                                     ENOMEM, ENOTDIR, EPERM, EROFS, EBADF,
-                                     EIO}},
+                                     ENOMEM, ENOTDIR, EPERM, EROFS, EIO}},
     [GF_FOP_FSETATTR] = {.error_no_count = 11,
                          .error_no = {EACCES, EFAULT, EIO, ENAMETOOLONG, ENOENT,
                                       ENOMEM, ENOTDIR, EPERM, EROFS, EBADF,
                                       EIO}},
-    [GF_FOP_GETSPEC] = {.error_no_count = 4,
-                        .error_no = {EACCES, EBADF, ENAMETOOLONG, EINTR}}};
+    [GF_FOP_GETSPEC] = {.error_no_count = 3,
+                        .error_no = {EACCES, ENAMETOOLONG, EINTR}}};
 
 int
 generate_rand_no(int op_no)
@@ -761,6 +759,7 @@ error_gen_writev(call_frame_t *frame, xlator_t *this, fd_t *fd,
          */
         shortvec = iov_dup(vector, 1);
         shortvec->iov_len /= 2;
+        count = 1;
         goto wind;
     } else if (op_errno) {
         GF_ERROR(this, "unwind(-1, %s)", strerror(op_errno));
@@ -1396,7 +1395,7 @@ error_gen_priv_dump(xlator_t *this)
 
     gf_proc_dump_write("op_count", "%d", conf->op_count);
     gf_proc_dump_write("failure_iter_no", "%d", conf->failure_iter_no);
-    gf_proc_dump_write("error_no", "%s", conf->error_no);
+    gf_proc_dump_write("error_no", "%d", conf->error_no_int);
     gf_proc_dump_write("random_failure", "%d", conf->random_failure);
 
     UNLOCK(&conf->lock);
@@ -1430,6 +1429,7 @@ reconfigure(xlator_t *this, dict_t *options)
     eg_t *pvt = NULL;
     int32_t ret = 0;
     char *error_enable_fops = NULL;
+    char *error_no = NULL;
     double failure_percent_dbl = 0.0;
 
     if (!this || !this->private)
@@ -1439,10 +1439,10 @@ reconfigure(xlator_t *this, dict_t *options)
 
     ret = -1;
 
-    GF_OPTION_RECONF("error-no", pvt->error_no, options, str, out);
+    GF_OPTION_RECONF("error-no", error_no, options, str, out);
 
-    if (pvt->error_no)
-        pvt->error_no_int = conv_errno_to_int(&pvt->error_no);
+    if (error_no)
+        pvt->error_no_int = conv_errno_to_int(&error_no);
 
     GF_OPTION_RECONF("failure", failure_percent_dbl, options, percent, out);
 
@@ -1466,6 +1466,7 @@ init(xlator_t *this)
     eg_t *pvt = NULL;
     int32_t ret = 0;
     char *error_enable_fops = NULL;
+    char *error_no = NULL;
     double failure_percent_dbl = 0.0;
 
     if (!this->children || this->children->next) {
@@ -1490,10 +1491,10 @@ init(xlator_t *this)
 
     ret = -1;
 
-    GF_OPTION_INIT("error-no", pvt->error_no, str, out);
+    GF_OPTION_INIT("error-no", error_no, str, out);
 
-    if (pvt->error_no)
-        pvt->error_no_int = conv_errno_to_int(&pvt->error_no);
+    if (error_no)
+        pvt->error_no_int = conv_errno_to_int(&error_no);
 
     GF_OPTION_INIT("failure", failure_percent_dbl, percent, out);
 
@@ -1506,8 +1507,8 @@ init(xlator_t *this)
 
     this->private = pvt;
 
-    /* Give some seed value here */
-    srand(time(NULL));
+    /* Give some seed value here. */
+    srand(gf_time());
 
     ret = 0;
 out:
@@ -1644,4 +1645,19 @@ struct volume_options options[] = {
         .flags = OPT_FLAG_SETTABLE,
     },
 
-    {.key = {NULL}}};
+    {.key = {NULL}},
+};
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1},
+    .dumpops = &dumpops,
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "error-gen",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/debug/error-gen/src/error-gen.h b/xlators/debug/error-gen/src/error-gen.h
index ffa09252d0f..2478cd5b21c 100644
--- a/xlators/debug/error-gen/src/error-gen.h
+++ b/xlators/debug/error-gen/src/error-gen.h
@@ -36,7 +36,6 @@ typedef struct {
      * It's just not worth blowing up the diff by changing it.
      */
     int failure_iter_no;
-    char *error_no;
     int error_no_int;
     gf_boolean_t random_failure;
     gf_lock_t lock;
diff --git a/xlators/debug/io-stats/src/io-stats-mem-types.h b/xlators/debug/io-stats/src/io-stats-mem-types.h
index bc25fd2ca4e..51d38d8b97c 100644
--- a/xlators/debug/io-stats/src/io-stats-mem-types.h
+++ b/xlators/debug/io-stats/src/io-stats-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __IO_STATS_MEM_TYPES_H__
 #define __IO_STATS_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 extern const char *__progname;
 
diff --git a/xlators/debug/io-stats/src/io-stats.c b/xlators/debug/io-stats/src/io-stats.c
index 249ef5093b4..aa00c446e5a 100644
--- a/xlators/debug/io-stats/src/io-stats.c
+++ b/xlators/debug/io-stats/src/io-stats.c
@@ -7,8 +7,8 @@
    later), or the GNU General Public License, version 2 (GPLv2), in all
    cases as published by the Free Software Foundation.
 */
-#include "xlator.h"
-#include "syscall.h"
+#include <glusterfs/xlator.h>
+#include <glusterfs/syscall.h>
 
 /**
  * xlators/debug/io_stats :
@@ -28,18 +28,18 @@
 
 #include <fnmatch.h>
 #include <errno.h>
-#include "glusterfs.h"
-#include "xlator.h"
+#include <glusterfs/glusterfs.h>
+#include <glusterfs/xlator.h>
 #include "io-stats-mem-types.h"
 #include <stdarg.h>
-#include "defaults.h"
-#include "logging.h"
-#include "cli1-xdr.h"
-#include "statedump.h"
-#include "syncop.h"
+#include <glusterfs/defaults.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/statedump.h>
+#include <glusterfs/syncop.h>
 #include <pwd.h>
 #include <grp.h>
-#include "upcall-utils.h"
+#include <glusterfs/upcall-utils.h>
+#include <glusterfs/async.h>
 
 #define MAX_LIST_MEMBERS 100
 #define DEFAULT_PWD_BUF_SZ 16384
@@ -66,6 +66,17 @@ typedef enum {
     IOS_STATS_THRU_MAX,
 } ios_stats_thru_t;
 
+/* This is same as gf1_cli_info_op */
+/* had to be defined here again, so we have modularity between
+ xdr, xlator, and library functions */
+typedef enum ios_info_op {
+    GF_IOS_INFO_NONE = 0,
+    GF_IOS_INFO_ALL = 1,
+    GF_IOS_INFO_INCREMENTAL = 2,
+    GF_IOS_INFO_CUMULATIVE = 3,
+    GF_IOS_INFO_CLEAR = 4,
+} ios_info_op_t;
+
 struct ios_stat_lat {
     struct timeval time;
     double throughput;
@@ -124,7 +135,7 @@ struct ios_global_stats {
     gf_atomic_t block_count_read[IOS_BLOCK_COUNT_SIZE];
     gf_atomic_t fop_hits[GF_FOP_MAXVALUE];
     gf_atomic_t upcall_hits[GF_UPCALL_FLAGS_MAXVALUE];
-    struct timeval started_at;
+    time_t started_at;
     struct ios_lat latency[GF_FOP_MAXVALUE];
     uint64_t nr_opens;
     uint64_t max_nr_opens;
@@ -173,7 +184,6 @@ struct ios_conf {
      */
     char *unique_id;
     ios_dump_type_t dump_format;
-    char *dump_format_str;
 };
 
 struct ios_fd {
@@ -282,9 +292,7 @@ is_fop_latency_started(call_frame_t *frame)
         begin = &frame->begin;                                                 \
         end = &frame->end;                                                     \
                                                                                \
-        elapsed = ((end->tv_sec - begin->tv_sec) * 1e9 +                       \
-                   (end->tv_nsec - begin->tv_nsec)) /                          \
-                  1000;                                                        \
+        elapsed = gf_tsdiff(begin, end) / 1000.0;                              \
         throughput = op_ret / elapsed;                                         \
                                                                                \
         conf = this->private;                                                  \
@@ -668,10 +676,7 @@ ios_dump_throughput_stats(struct ios_stat_head *list_head, xlator_t *this,
                           FILE *logfp, ios_stats_thru_t type)
 {
     struct ios_stat_list *entry = NULL;
-    struct timeval time = {
-        0,
-    };
-    char timestr[256] = {
+    char timestr[GF_TIMESTR_SIZE] = {
         0,
     };
 
@@ -679,12 +684,9 @@ ios_dump_throughput_stats(struct ios_stat_head *list_head, xlator_t *this,
     {
         list_for_each_entry(entry, &list_head->iosstats->list, list)
         {
-            gf_time_fmt(timestr, sizeof timestr,
-                        entry->iosstat->thru_counters[type].time.tv_sec,
-                        gf_timefmt_FT);
-            snprintf(timestr + strlen(timestr),
-                     sizeof timestr - strlen(timestr), ".%" GF_PRI_SUSECONDS,
-                     time.tv_usec);
+            gf_time_fmt_tv(timestr, sizeof timestr,
+                           &entry->iosstat->thru_counters[type].time,
+                           gf_timefmt_FT);
 
             ios_log(this, logfp, "%s \t %-10.2f  \t  %s", timestr, entry->value,
                     entry->iosstat->filename);
@@ -763,9 +765,8 @@ err:
 
 int
 io_stats_dump_global_to_json_logfp(xlator_t *this,
-                                   struct ios_global_stats *stats,
-                                   struct timeval *now, int interval,
-                                   FILE *logfp)
+                                   struct ios_global_stats *stats, time_t now,
+                                   int interval, FILE *logfp)
 {
     int i = 0;
     int j = 0;
@@ -791,10 +792,7 @@ io_stats_dump_global_to_json_logfp(xlator_t *this,
     };
     dict_t *xattr = NULL;
 
-    interval_sec = ((now->tv_sec * 1000000.0 + now->tv_usec) -
-                    (stats->started_at.tv_sec * 1000000.0 +
-                     stats->started_at.tv_usec)) /
-                   1000000.0;
+    interval_sec = (double)(now - stats->started_at);
 
     conf = this->private;
 
@@ -946,8 +944,8 @@ io_stats_dump_global_to_json_logfp(xlator_t *this,
     }
 
     if (interval == -1) {
-        ios_log(this, logfp, "\"%s.%s.uptime\": %" PRId64 ",", key_prefix,
-                str_prefix, (uint64_t)(now->tv_sec - stats->started_at.tv_sec));
+        ios_log(this, logfp, "\"%s.%s.uptime\": %" PRIu64 ",", key_prefix,
+                str_prefix, (uint64_t)(now - stats->started_at));
         ios_log(this, logfp,
                 "\"%s.%s.bytes_read\": "
                 "%" GF_PRI_ATOMIC ",",
@@ -1199,14 +1197,14 @@ out:
 
 int
 io_stats_dump_global_to_logfp(xlator_t *this, struct ios_global_stats *stats,
-                              struct timeval *now, int interval, FILE *logfp)
+                              time_t now, int interval, FILE *logfp)
 {
     int i = 0;
     int per_line = 0;
     int index = 0;
     struct ios_stat_head *list_head = NULL;
     struct ios_conf *conf = NULL;
-    char timestr[256] = {
+    char timestr[GF_TIMESTR_SIZE] = {
         0,
     };
     char str_header[128] = {0};
@@ -1222,8 +1220,8 @@ io_stats_dump_global_to_logfp(xlator_t *this, struct ios_global_stats *stats,
         ios_log(this, logfp, "\n=== Cumulative stats ===");
     else
         ios_log(this, logfp, "\n=== Interval %d stats ===", interval);
-    ios_log(this, logfp, "      Duration : %" PRId64 " secs",
-            (uint64_t)(now->tv_sec - stats->started_at.tv_sec));
+    ios_log(this, logfp, "      Duration : %" PRIu64 " secs",
+            (uint64_t)(now - stats->started_at));
     ios_log(this, logfp, "     BytesRead : %" GF_PRI_ATOMIC,
             GF_ATOMIC_GET(stats->data_read));
     ios_log(this, logfp, "  BytesWritten : %" GF_PRI_ATOMIC "\n",
@@ -1315,11 +1313,8 @@ io_stats_dump_global_to_logfp(xlator_t *this, struct ios_global_stats *stats,
     if (interval == -1) {
         LOCK(&conf->lock);
         {
-            gf_time_fmt(timestr, sizeof timestr,
-                        conf->cumulative.max_openfd_time.tv_sec, gf_timefmt_FT);
-            snprintf(timestr + strlen(timestr),
-                     sizeof timestr - strlen(timestr), ".%" GF_PRI_SUSECONDS,
-                     conf->cumulative.max_openfd_time.tv_usec);
+            gf_time_fmt_tv(timestr, sizeof timestr,
+                           &conf->cumulative.max_openfd_time, gf_timefmt_FT);
             ios_log(this, logfp,
                     "Current open fd's: %" PRId64 " Max open fd's: %" PRId64
                     " time %s",
@@ -1371,10 +1366,10 @@ io_stats_dump_global_to_logfp(xlator_t *this, struct ios_global_stats *stats,
 
 int
 io_stats_dump_global_to_dict(xlator_t *this, struct ios_global_stats *stats,
-                             struct timeval *now, int interval, dict_t *dict)
+                             time_t now, int interval, dict_t *dict)
 {
     int ret = 0;
-    char key[256] = {0};
+    char key[64] = {0};
     uint64_t sec = 0;
     int i = 0;
     uint64_t count = 0;
@@ -1397,7 +1392,7 @@ io_stats_dump_global_to_dict(xlator_t *this, struct ios_global_stats *stats,
                interval);
 
     snprintf(key, sizeof(key), "%d-duration", interval);
-    sec = (uint64_t)(now->tv_sec - stats->started_at.tv_sec);
+    sec = now - stats->started_at;
     ret = dict_set_uint64(dict, key, sec);
     if (ret) {
         gf_log(this->name, GF_LOG_ERROR,
@@ -1520,9 +1515,8 @@ out:
 }
 
 int
-io_stats_dump_global(xlator_t *this, struct ios_global_stats *stats,
-                     struct timeval *now, int interval,
-                     struct ios_dump_args *args)
+io_stats_dump_global(xlator_t *this, struct ios_global_stats *stats, time_t now,
+                     int interval, struct ios_dump_args *args)
 {
     int ret = -1;
 
@@ -1580,24 +1574,24 @@ ios_dump_args_init(struct ios_dump_args *args, ios_dump_type_t type,
 }
 
 static void
-ios_global_stats_clear(struct ios_global_stats *stats, struct timeval *now)
+ios_global_stats_clear(struct ios_global_stats *stats, time_t now)
 {
     GF_ASSERT(stats);
     GF_ASSERT(now);
 
     memset(stats, 0, sizeof(*stats));
-    stats->started_at = *now;
+    stats->started_at = now;
 }
 
 int
-io_stats_dump(xlator_t *this, struct ios_dump_args *args, gf1_cli_info_op op,
+io_stats_dump(xlator_t *this, struct ios_dump_args *args, ios_info_op_t op,
               gf_boolean_t is_peek)
 {
     struct ios_conf *conf = NULL;
     struct ios_global_stats cumulative = {};
     struct ios_global_stats incremental = {};
     int increment = 0;
-    struct timeval now;
+    time_t now = 0;
 
     GF_ASSERT(this);
     GF_ASSERT(args);
@@ -1605,31 +1599,31 @@ io_stats_dump(xlator_t *this, struct ios_dump_args *args, gf1_cli_info_op op,
     GF_ASSERT(args->type < IOS_DUMP_TYPE_MAX);
 
     conf = this->private;
+    now = gf_time();
 
-    gettimeofday(&now, NULL);
     LOCK(&conf->lock);
     {
-        if (op == GF_CLI_INFO_ALL || op == GF_CLI_INFO_CUMULATIVE)
+        if (op == GF_IOS_INFO_ALL || op == GF_IOS_INFO_CUMULATIVE)
             cumulative = conf->cumulative;
 
-        if (op == GF_CLI_INFO_ALL || op == GF_CLI_INFO_INCREMENTAL) {
+        if (op == GF_IOS_INFO_ALL || op == GF_IOS_INFO_INCREMENTAL) {
             incremental = conf->incremental;
             increment = conf->increment;
 
             if (!is_peek) {
                 increment = conf->increment++;
 
-                ios_global_stats_clear(&conf->incremental, &now);
+                ios_global_stats_clear(&conf->incremental, now);
             }
         }
     }
     UNLOCK(&conf->lock);
 
-    if (op == GF_CLI_INFO_ALL || op == GF_CLI_INFO_CUMULATIVE)
-        io_stats_dump_global(this, &cumulative, &now, -1, args);
+    if (op == GF_IOS_INFO_ALL || op == GF_IOS_INFO_CUMULATIVE)
+        io_stats_dump_global(this, &cumulative, now, -1, args);
 
-    if (op == GF_CLI_INFO_ALL || op == GF_CLI_INFO_INCREMENTAL)
-        io_stats_dump_global(this, &incremental, &now, increment, args);
+    if (op == GF_IOS_INFO_ALL || op == GF_IOS_INFO_INCREMENTAL)
+        io_stats_dump_global(this, &incremental, now, increment, args);
 
     return 0;
 }
@@ -1639,9 +1633,8 @@ io_stats_dump_fd(xlator_t *this, struct ios_fd *iosfd)
 {
     struct ios_conf *conf = NULL;
     struct timeval now;
-    uint64_t sec = 0;
-    uint64_t usec = 0;
     int i = 0;
+    double usecs = 0;
     uint64_t data_read = 0;
     uint64_t data_written = 0;
     uint64_t block_count_read = 0;
@@ -1656,23 +1649,15 @@ io_stats_dump_fd(xlator_t *this, struct ios_fd *iosfd)
         return 0;
 
     gettimeofday(&now, NULL);
-
-    if (iosfd->opened_at.tv_usec > now.tv_usec) {
-        now.tv_usec += 1000000;
-        now.tv_usec--;
-    }
-
-    sec = now.tv_sec - iosfd->opened_at.tv_sec;
-    usec = now.tv_usec - iosfd->opened_at.tv_usec;
+    usecs = gf_tvdiff(&iosfd->opened_at, &now);
 
     gf_log(this->name, GF_LOG_INFO, "--- fd stats ---");
 
     if (iosfd->filename)
         gf_log(this->name, GF_LOG_INFO, "      Filename : %s", iosfd->filename);
 
-    if (sec)
-        gf_log(this->name, GF_LOG_INFO,
-               "      Lifetime : %" PRId64 "secs, %" PRId64 "usecs", sec, usec);
+    if (usecs)
+        gf_log(this->name, GF_LOG_INFO, "      Lifetime : %lf secs", usecs);
 
     data_read = GF_ATOMIC_GET(iosfd->data_read);
     if (data_read)
@@ -1775,9 +1760,7 @@ update_ios_latency(struct ios_conf *conf, call_frame_t *frame,
     begin = &frame->begin;
     end = &frame->end;
 
-    elapsed = ((end->tv_sec - begin->tv_sec) * 1e9 +
-               (end->tv_nsec - begin->tv_nsec)) /
-              1000;
+    elapsed = gf_tsdiff(begin, end) / 1000.0;
 
     update_ios_latency_stats(&conf->cumulative, elapsed, op);
     update_ios_latency_stats(&conf->incremental, elapsed, op);
@@ -1792,12 +1775,13 @@ io_stats_dump_stats_to_dict(xlator_t *this, dict_t *resp,
 {
     struct ios_conf *conf = NULL;
     int cnt = 0;
-    char key[256];
+    char key[32];
+    int keylen;
     struct ios_stat_head *list_head = NULL;
     struct ios_stat_list *entry = NULL;
     int ret = -1;
     ios_stats_thru_t index = IOS_STATS_THRU_MAX;
-    char timestr[256] = {
+    char timestr[GF_TIMESTR_SIZE] = {
         0,
     };
     char *dict_timestr = NULL;
@@ -1816,14 +1800,9 @@ io_stats_dump_stats_to_dict(xlator_t *this, dict_t *resp,
                 ret = dict_set_uint64(resp, "max-open",
                                       conf->cumulative.max_nr_opens);
 
-                gf_time_fmt(timestr, sizeof timestr,
-                            conf->cumulative.max_openfd_time.tv_sec,
-                            gf_timefmt_FT);
-                if (conf->cumulative.max_openfd_time.tv_sec)
-                    snprintf(timestr + strlen(timestr),
-                             sizeof timestr - strlen(timestr),
-                             ".%" GF_PRI_SUSECONDS,
-                             conf->cumulative.max_openfd_time.tv_usec);
+                gf_time_fmt_tv(timestr, sizeof timestr,
+                               &conf->cumulative.max_openfd_time,
+                               gf_timefmt_FT);
 
                 dict_timestr = gf_strdup(timestr);
                 if (!dict_timestr)
@@ -1863,7 +1842,7 @@ io_stats_dump_stats_to_dict(xlator_t *this, dict_t *resp,
         default:
             goto out;
     }
-    ret = dict_set_int32(resp, "top-op", flags);
+    ret = dict_set_int32_sizen(resp, "top-op", flags);
     if (!list_cnt)
         goto out;
     LOCK(&list_head->lock);
@@ -1871,24 +1850,24 @@ io_stats_dump_stats_to_dict(xlator_t *this, dict_t *resp,
         list_for_each_entry(entry, &list_head->iosstats->list, list)
         {
             cnt++;
-            snprintf(key, 256, "%s-%d", "filename", cnt);
-            ret = dict_set_str(resp, key, entry->iosstat->filename);
+            keylen = snprintf(key, sizeof(key), "filename-%d", cnt);
+            ret = dict_set_strn(resp, key, keylen, entry->iosstat->filename);
             if (ret)
                 goto unlock_list_head;
-            snprintf(key, 256, "%s-%d", "value", cnt);
+            snprintf(key, sizeof(key), "value-%d", cnt);
             ret = dict_set_uint64(resp, key, entry->value);
             if (ret)
                 goto unlock_list_head;
             if (index != IOS_STATS_THRU_MAX) {
-                snprintf(key, 256, "%s-%d", "time-sec", cnt);
-                ret = dict_set_int32(
-                    resp, key,
+                keylen = snprintf(key, sizeof(key), "time-sec-%d", cnt);
+                ret = dict_set_int32n(
+                    resp, key, keylen,
                     entry->iosstat->thru_counters[index].time.tv_sec);
                 if (ret)
                     goto unlock_list_head;
-                snprintf(key, 256, "%s-%d", "time-usec", cnt);
-                ret = dict_set_int32(
-                    resp, key,
+                keylen = snprintf(key, sizeof(key), "time-usec-%d", cnt);
+                ret = dict_set_int32n(
+                    resp, key, keylen,
                     entry->iosstat->thru_counters[index].time.tv_usec);
                 if (ret)
                     goto unlock_list_head;
@@ -1903,7 +1882,7 @@ unlock_list_head:
      * failed. */
     if (ret)
         goto out;
-    ret = dict_set_int32(resp, "members", cnt);
+    ret = dict_set_int32_sizen(resp, "members", cnt);
 out:
     return ret;
 }
@@ -2119,6 +2098,19 @@ io_stats_writev_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
 }
 
 int
+io_stats_copy_file_range_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
+                             int32_t op_ret, int32_t op_errno,
+                             struct iatt *stbuf, struct iatt *prebuf_dst,
+                             struct iatt *postbuf_dst, dict_t *xdata)
+{
+    UPDATE_PROFILE_STATS(frame, COPY_FILE_RANGE);
+
+    STACK_UNWIND_STRICT(copy_file_range, frame, op_ret, op_errno, stbuf,
+                        prebuf_dst, postbuf_dst, xdata);
+    return 0;
+}
+
+int
 io_stats_readdirp_cbk(call_frame_t *frame, void *cookie, xlator_t *this,
                       int32_t op_ret, int32_t op_errno, gf_dirent_t *buf,
                       dict_t *xdata)
@@ -2873,6 +2865,19 @@ io_stats_writev(call_frame_t *frame, xlator_t *this, fd_t *fd,
 }
 
 int
+io_stats_copy_file_range(call_frame_t *frame, xlator_t *this, fd_t *fd_in,
+                         off_t off_in, fd_t *fd_out, off_t off_out, size_t len,
+                         uint32_t flags, dict_t *xdata)
+{
+    START_FOP_LATENCY(frame);
+
+    STACK_WIND(frame, io_stats_copy_file_range_cbk, FIRST_CHILD(this),
+               FIRST_CHILD(this)->fops->copy_file_range, fd_in, off_in, fd_out,
+               off_out, len, flags, xdata);
+    return 0;
+}
+
+int
 io_stats_statfs(call_frame_t *frame, xlator_t *this, loc_t *loc, dict_t *xdata)
 {
     START_FOP_LATENCY(frame);
@@ -3005,7 +3010,7 @@ conditional_dump(dict_t *dict, char *key, data_t *value, void *data)
     } else {
         (void)ios_dump_args_init(&args, IOS_DUMP_TYPE_FILE, logfp);
     }
-    io_stats_dump(this, &args, GF_CLI_INFO_ALL, _gf_false);
+    io_stats_dump(this, &args, GF_IOS_INFO_ALL, _gf_false);
     fclose(logfp);
     return 0;
 }
@@ -3108,7 +3113,7 @@ _ios_dump_thread(xlator_t *this)
         stats_logfp = fopen(stats_filename, "w+");
         if (stats_logfp) {
             (void)ios_dump_args_init(&args, conf->dump_format, stats_logfp);
-            io_stats_dump(this, &args, GF_CLI_INFO_ALL, _gf_false);
+            io_stats_dump(this, &args, GF_IOS_INFO_ALL, _gf_false);
             fclose(stats_logfp);
             log_stats_fopen_failure = _gf_true;
         } else if (log_stats_fopen_failure) {
@@ -3569,26 +3574,21 @@ ios_destroy_top_stats(struct ios_conf *conf)
     return;
 }
 
-static int
+static void
 io_stats_clear(struct ios_conf *conf)
 {
-    struct timeval now;
-    int ret = -1;
+    time_t now = 0;
 
     GF_ASSERT(conf);
+    now = gf_time();
 
-    if (!gettimeofday(&now, NULL)) {
-        LOCK(&conf->lock);
-        {
-            ios_global_stats_clear(&conf->cumulative, &now);
-            ios_global_stats_clear(&conf->incremental, &now);
-            conf->increment = 0;
-        }
-        UNLOCK(&conf->lock);
-        ret = 0;
+    LOCK(&conf->lock);
+    {
+        ios_global_stats_clear(&conf->cumulative, now);
+        ios_global_stats_clear(&conf->incremental, now);
+        conf->increment = 0;
     }
-
-    return ret;
+    UNLOCK(&conf->lock);
 }
 
 int32_t
@@ -3654,15 +3654,15 @@ io_priv(xlator_t *this)
 }
 
 static void
-ios_set_log_format_code(struct ios_conf *conf)
+ios_set_log_format_code(struct ios_conf *conf, char *dump_format_str)
 {
-    if (strcmp(conf->dump_format_str, "json") == 0)
+    if (strcmp(dump_format_str, "json") == 0)
         conf->dump_format = IOS_DUMP_TYPE_JSON_FILE;
-    else if (strcmp(conf->dump_format_str, "text") == 0)
+    else if (strcmp(dump_format_str, "text") == 0)
         conf->dump_format = IOS_DUMP_TYPE_FILE;
-    else if (strcmp(conf->dump_format_str, "dict") == 0)
+    else if (strcmp(dump_format_str, "dict") == 0)
         conf->dump_format = IOS_DUMP_TYPE_DICT;
-    else if (strcmp(conf->dump_format_str, "samples") == 0)
+    else if (strcmp(dump_format_str, "samples") == 0)
         conf->dump_format = IOS_DUMP_TYPE_SAMPLES;
 }
 
@@ -3679,19 +3679,23 @@ xlator_set_loglevel(xlator_t *this, int log_level)
     active = ctx->active;
     top = active->first;
 
-    if (strcmp(top->type, "protocol/server") || (log_level == -1))
+    if (log_level == -1)
         return;
 
-    /* Set log-level for server xlator */
-    top->loglevel = log_level;
+    if (ctx->cmd_args.brick_mux) {
+        /* Set log-level for all brick xlators */
+        top->loglevel = log_level;
 
-    /* Set log-level for parent xlator */
-    if (this->parents)
-        this->parents->xlator->loglevel = log_level;
+        /* Set log-level for parent xlator */
+        if (this->parents)
+            this->parents->xlator->loglevel = log_level;
 
-    while (trav) {
-        trav->loglevel = log_level;
-        trav = trav->next;
+        while (trav) {
+            trav->loglevel = log_level;
+            trav = trav->next;
+        }
+    } else {
+        gf_log_set_loglevel(this->ctx, log_level);
     }
 }
 
@@ -3703,6 +3707,7 @@ reconfigure(xlator_t *this, dict_t *options)
     char *sys_log_str = NULL;
     char *log_format_str = NULL;
     char *logger_str = NULL;
+    char *dump_format_str = NULL;
     int sys_log_level = -1;
     char *log_str = NULL;
     int log_level = -1;
@@ -3711,6 +3716,7 @@ reconfigure(xlator_t *this, dict_t *options)
     uint32_t log_buf_size = 0;
     uint32_t log_flush_timeout = 0;
     int32_t old_dump_interval;
+    int32_t threads;
 
     if (!this || !this->private)
         goto out;
@@ -3747,9 +3753,8 @@ reconfigure(xlator_t *this, dict_t *options)
 
     GF_OPTION_RECONF("ios-sample-interval", conf->ios_sample_interval, options,
                      int32, out);
-    GF_OPTION_RECONF("ios-dump-format", conf->dump_format_str, options, str,
-                     out);
-    ios_set_log_format_code(conf);
+    GF_OPTION_RECONF("ios-dump-format", dump_format_str, options, str, out);
+    ios_set_log_format_code(conf, dump_format_str);
     GF_OPTION_RECONF("ios-sample-buf-size", conf->ios_sample_buf_size, options,
                      int32, out);
     GF_OPTION_RECONF("sys-log-level", sys_log_str, options, str, out);
@@ -3784,6 +3789,9 @@ reconfigure(xlator_t *this, dict_t *options)
                      out);
     gf_log_set_log_flush_timeout(log_flush_timeout);
 
+    GF_OPTION_RECONF("threads", threads, options, int32, out);
+    gf_async_adjust_threads(threads);
+
     ret = 0;
 out:
     gf_log(this ? this->name : "io-stats", GF_LOG_DEBUG,
@@ -3821,7 +3829,7 @@ ios_conf_destroy(struct ios_conf *conf)
     _ios_destroy_dump_thread(conf);
     ios_destroy_sample_buf(conf->ios_sample_buf);
     LOCK_DESTROY(&conf->lock);
-    GF_FREE(conf->dnscache);
+    gf_dnscache_deinit(conf->dnscache);
     GF_FREE(conf);
 }
 
@@ -3844,16 +3852,18 @@ ios_init_stats(struct ios_global_stats *stats)
     for (i = 0; i < GF_UPCALL_FLAGS_MAXVALUE; i++)
         GF_ATOMIC_INIT(stats->upcall_hits[i], 0);
 
-    gettimeofday(&stats->started_at, NULL);
+    stats->started_at = gf_time();
 }
 
 int
 init(xlator_t *this)
 {
     struct ios_conf *conf = NULL;
+    char *volume_id = NULL;
     char *sys_log_str = NULL;
     char *logger_str = NULL;
     char *log_format_str = NULL;
+    char *dump_format_str = NULL;
     int logger = -1;
     int log_format = -1;
     int sys_log_level = -1;
@@ -3862,6 +3872,7 @@ init(xlator_t *this)
     int ret = -1;
     uint32_t log_buf_size = 0;
     uint32_t log_flush_timeout = 0;
+    int32_t threads;
 
     if (!this)
         return -1;
@@ -3889,6 +3900,11 @@ init(xlator_t *this)
         conf->unique_id = this->name;
     }
 
+    ret = dict_get_strn(this->options, "volume-id", SLEN("volume-id"),
+                        &volume_id);
+    if (!ret) {
+        strncpy(this->graph->volume_id, volume_id, GF_UUID_BUF_SIZE);
+    }
     /*
      * Init it just after calloc, so that we are sure the lock is inited
      * in case of error paths.
@@ -3914,8 +3930,8 @@ init(xlator_t *this)
     GF_OPTION_INIT("ios-sample-interval", conf->ios_sample_interval, int32,
                    out);
 
-    GF_OPTION_INIT("ios-dump-format", conf->dump_format_str, str, out);
-    ios_set_log_format_code(conf);
+    GF_OPTION_INIT("ios-dump-format", dump_format_str, str, out);
+    ios_set_log_format_code(conf, dump_format_str);
 
     GF_OPTION_INIT("ios-sample-buf-size", conf->ios_sample_buf_size, int32,
                    out);
@@ -3929,6 +3945,10 @@ init(xlator_t *this)
     GF_OPTION_INIT("ios-dnscache-ttl-sec", conf->ios_dnscache_ttl_sec, int32,
                    out);
     conf->dnscache = gf_dnscache_init(conf->ios_dnscache_ttl_sec);
+    if (!conf->dnscache) {
+        ret = -1;
+        goto out;
+    }
 
     GF_OPTION_INIT("sys-log-level", sys_log_str, str, out);
     if (sys_log_str) {
@@ -3961,6 +3981,9 @@ init(xlator_t *this)
     GF_OPTION_INIT("log-flush-timeout", log_flush_timeout, time, out);
     gf_log_set_log_flush_timeout(log_flush_timeout);
 
+    GF_OPTION_INIT("threads", threads, int32, out);
+    gf_async_adjust_threads(threads);
+
     this->private = conf;
     if (conf->ios_dump_interval > 0) {
         conf->dump_thread_running = _gf_true;
@@ -4065,8 +4088,8 @@ notify(xlator_t *this, int32_t event, void *data, ...)
                 }
             } else {
                 ret = dict_get_int32(dict, "info-op", &op);
-                if (ret || op < GF_CLI_INFO_ALL || GF_CLI_INFO_CLEAR < op)
-                    op = GF_CLI_INFO_ALL;
+                if (ret || op < GF_IOS_INFO_ALL || GF_IOS_INFO_CLEAR < op)
+                    op = GF_IOS_INFO_ALL;
 
                 ret = dict_set_int32(output, "info-op", op);
                 if (ret) {
@@ -4075,13 +4098,10 @@ notify(xlator_t *this, int32_t event, void *data, ...)
                     goto out;
                 }
 
-                if (GF_CLI_INFO_CLEAR == op) {
-                    ret = io_stats_clear(this->private);
-                    if (ret)
-                        gf_log(this->name, GF_LOG_ERROR,
-                               "Failed to clear info stats");
+                if (GF_IOS_INFO_CLEAR == op) {
+                    io_stats_clear(this->private);
 
-                    ret = dict_set_int32(output, "stats-cleared", ret ? 0 : 1);
+                    ret = dict_set_int32(output, "stats-cleared", 1);
                     if (ret)
                         gf_log(this->name, GF_LOG_ERROR,
                                "Failed to set stats-cleared"
@@ -4189,6 +4209,7 @@ struct xlator_fops fops = {
     .getactivelk = io_stats_getactivelk,
     .setactivelk = io_stats_setactivelk,
     .compound = io_stats_compound,
+    .copy_file_range = io_stats_copy_file_range,
 };
 
 struct xlator_cbks cbks = {
@@ -4403,6 +4424,57 @@ struct volume_options options[] = {
      .type = GF_OPTION_TYPE_STR,
      .default_value = "/no/such/path",
      .description = "Unique ID for our files."},
+    {.key = {"global-threading"},
+     .type = GF_OPTION_TYPE_BOOL,
+     .default_value = "off",
+     .op_version = {GD_OP_VERSION_6_0},
+     .flags = OPT_FLAG_SETTABLE,
+     .tags = {"io-stats", "threading"},
+     .description = "This option enables the global threading support for "
+                    "bricks. If enabled, it's recommended to also enable "
+                    "'performance.iot-pass-through'"},
+    {.key = {"threads"}, .type = GF_OPTION_TYPE_INT},
+    {.key = {"brick-threads"},
+     .type = GF_OPTION_TYPE_INT,
+     .default_value = "16",
+     .min = 0,
+     .max = GF_ASYNC_MAX_THREADS,
+     .op_version = {GD_OP_VERSION_6_0},
+     .flags = OPT_FLAG_SETTABLE | OPT_FLAG_DOC,
+     .tags = {"io-stats", "threading"},
+     .description = "When global threading is used, this value determines the "
+                    "maximum amount of threads that can be created on bricks"},
+    {.key = {"client-threads"},
+     .type = GF_OPTION_TYPE_INT,
+     .default_value = "16",
+     .min = 0,
+     .max = GF_ASYNC_MAX_THREADS,
+     .op_version = {GD_OP_VERSION_6_0},
+     .flags = OPT_FLAG_SETTABLE | OPT_FLAG_DOC | OPT_FLAG_CLIENT_OPT,
+     .tags = {"io-stats", "threading"},
+     .description = "When global threading is used, this value determines the "
+                    "maximum amount of threads that can be created on clients"},
+    {.key = {"volume-id"},
+     .type = GF_OPTION_TYPE_STR,
+     .op_version = {GD_OP_VERSION_7_1},
+     .tags = {"global", "volume-id"},
+     .description =
+         "This option points to the 'unique' UUID particular to this "
+         "volume, which would be set in 'graph->volume_id'"},
     {.key = {NULL}},
+};
 
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .notify = notify,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1}, /* Present from the initial version */
+    .dumpops = &dumpops,
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "io-stats",
+    .category = GF_MAINTAINED,
 };
diff --git a/xlators/debug/sink/src/sink.c b/xlators/debug/sink/src/sink.c
index fbbdd3a4847..9822bbb732e 100644
--- a/xlators/debug/sink/src/sink.c
+++ b/xlators/debug/sink/src/sink.c
@@ -8,8 +8,8 @@
    cases as published by the Free Software Foundation.
 */
 
-#include "xlator.h"
-#include "defaults.h"
+#include <glusterfs/xlator.h>
+#include <glusterfs/defaults.h>
 
 int32_t
 init(xlator_t *this)
@@ -80,3 +80,15 @@ struct xlator_cbks cbks = {};
 struct volume_options options[] = {
     {.key = {NULL}},
 };
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .notify = notify,
+    .op_version = {GD_OP_VERSION_3_12_0},
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "sink",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/debug/trace/src/trace-mem-types.h b/xlators/debug/trace/src/trace-mem-types.h
index cf05a77b9f1..18a7e0414a6 100644
--- a/xlators/debug/trace/src/trace-mem-types.h
+++ b/xlators/debug/trace/src/trace-mem-types.h
@@ -11,7 +11,7 @@
 #ifndef __TRACE_MEM_TYPES_H__
 #define __TRACE_MEM_TYPES_H__
 
-#include "mem-types.h"
+#include <glusterfs/mem-types.h>
 
 enum gf_trace_mem_types_ {
     gf_trace_mt_trace_conf_t = gf_common_mt_end + 1,
diff --git a/xlators/debug/trace/src/trace.c b/xlators/debug/trace/src/trace.c
index 0aca3a9a5bb..6ed0ca00342 100644
--- a/xlators/debug/trace/src/trace.c
+++ b/xlators/debug/trace/src/trace.c
@@ -22,13 +22,13 @@
 static void
 trace_stat_to_str(struct iatt *buf, char *str, size_t len)
 {
-    char atime_buf[200] = {
+    char atime_buf[GF_TIMESTR_SIZE] = {
         0,
     };
-    char mtime_buf[200] = {
+    char mtime_buf[GF_TIMESTR_SIZE] = {
         0,
     };
-    char ctime_buf[200] = {
+    char ctime_buf[GF_TIMESTR_SIZE] = {
         0,
     };
 
@@ -64,7 +64,7 @@ trace_stat_to_str(struct iatt *buf, char *str, size_t len)
 int
 dump_history_trace(circular_buffer_t *cb, void *data)
 {
-    char timestr[256] = {
+    char timestr[GF_TIMESTR_SIZE] = {
         0,
     };
 
@@ -72,9 +72,7 @@ dump_history_trace(circular_buffer_t *cb, void *data)
        gettimeofday () fails, it's safe to check tm and then dump the time
        at which the entry was added to the buffer */
 
-    gf_time_fmt(timestr, sizeof timestr, cb->tv.tv_sec, gf_timefmt_Ymd_T);
-    snprintf(timestr + strlen(timestr), 256 - strlen(timestr),
-             ".%" GF_PRI_SUSECONDS, cb->tv.tv_usec);
+    gf_time_fmt_tv(timestr, sizeof timestr, &cb->tv, gf_timefmt_Ymd_T);
     gf_proc_dump_write("TIME", "%s", timestr);
 
     gf_proc_dump_write("FOP", "%s\n", (char *)cb->data);
@@ -2209,10 +2207,10 @@ int
 trace_setattr(call_frame_t *frame, xlator_t *this, loc_t *loc,
               struct iatt *stbuf, int32_t valid, dict_t *xdata)
 {
-    char actime_str[256] = {
+    char actime_str[GF_TIMESTR_SIZE] = {
         0,
     };
-    char modtime_str[256] = {
+    char modtime_str[GF_TIMESTR_SIZE] = {
         0,
     };
     trace_conf_t *conf = NULL;
@@ -2278,10 +2276,10 @@ int
 trace_fsetattr(call_frame_t *frame, xlator_t *this, fd_t *fd,
                struct iatt *stbuf, int32_t valid, dict_t *xdata)
 {
-    char actime_str[256] = {
+    char actime_str[GF_TIMESTR_SIZE] = {
         0,
     };
-    char modtime_str[256] = {
+    char modtime_str[GF_TIMESTR_SIZE] = {
         0,
     };
     trace_conf_t *conf = NULL;
@@ -3520,3 +3518,17 @@ struct volume_options options[] = {
 };
 
 struct xlator_dumpops dumpops = {.history = trace_dump_history};
+
+xlator_api_t xlator_api = {
+    .init = init,
+    .fini = fini,
+    .reconfigure = reconfigure,
+    .mem_acct_init = mem_acct_init,
+    .op_version = {1},
+    .dumpops = &dumpops,
+    .fops = &fops,
+    .cbks = &cbks,
+    .options = options,
+    .identifier = "trace",
+    .category = GF_TECH_PREVIEW,
+};
diff --git a/xlators/debug/trace/src/trace.h b/xlators/debug/trace/src/trace.h
index cd73e0f34ed..b16304799da 100644
--- a/xlators/debug/trace/src/trace.h
+++ b/xlators/debug/trace/src/trace.h
@@ -10,14 +10,14 @@
 
 #include <time.h>
 #include <errno.h>
-#include "glusterfs.h"
-#include "xlator.h"
-#include "common-utils.h"
-#include "event-history.h"
-#include "logging.h"
-#include "circ-buff.h"
-#include "statedump.h"
-#include "options.h"
+#include <glusterfs/glusterfs.h>
+#include <glusterfs/xlator.h>
+#include <glusterfs/common-utils.h>
+#include <glusterfs/event-history.h>
+#include <glusterfs/logging.h>
+#include <glusterfs/circ-buff.h>
+#include <glusterfs/statedump.h>
+#include <glusterfs/options.h>
 
 #define TRACE_DEFAULT_HISTORY_SIZE 1024
 
diff --git a/xlators/encryption/Makefile.am b/xlators/encryption/Makefile.am
deleted file mode 100644
index 36efc6698bd..00000000000
--- a/xlators/encryption/Makefile.am
+++ /dev/null
@@ -1,3 +0,0 @@
-SUBDIRS = rot-13 crypt
-
-CLEANFILES = 
diff --git a/xlators/encryption/crypt/Makefile.am b/xlators/encryption/crypt/Makefile.am
deleted file mode 100644
index d471a3f9243..00000000000
--- a/xlators/encryption/crypt/Makefile.am
+++ /dev/null
@@ -1,3 +0,0 @@
-SUBDIRS = src
-
-CLEANFILES = 
diff --git a/xlators/encryption/crypt/src/Makefile.am b/xlators/encryption/crypt/src/Makefile.am
deleted file mode 100644
index 05fd3d5096b..00000000000
--- a/xlators/encryption/crypt/src/Makefile.am
+++ /dev/null
@@ -1,26 +0,0 @@
-if ENABLE_CRYPT_XLATOR
-
-xlator_LTLIBRARIES = crypt.la
-xlatordir = $(libdir)/glusterfs/$(PACKAGE_VERSION)/xlator/encryption
-
-crypt_la_LDFLAGS = -module $(GF_XLATOR_DEFAULT_LDFLAGS)
-
-crypt_la_SOURCES = keys.c data.c metadata.c atom.c crypt.c
-crypt_la_LIBADD = $(top_builddir)/libglusterfs/src/libglusterfs.la \
-	-lssl -lcrypto
-
-noinst_HEADERS = crypt-common.h crypt-mem-types.h crypt.h metadata.h
-
-AM_CPPFLAGS = $(GF_CPPFLAGS) -I$(top_srcdir)/libglusterfs/src \
-	-I$(top_srcdir)/rpc/xdr/src -I$(top_builddir)/rpc/xdr/src
-
-AM_CFLAGS = -Wall $(GF_CFLAGS)
-
-CLEANFILES =
-
-else
-
-noinst_DIST = keys.c data.c metadata.c atom.c crypt.c
-noinst_HEADERS = crypt-common.h crypt-mem-types.h crypt.h metadata.h
-
-endif
diff --git a/xlators/encryption/crypt/src/atom.c b/xlators/encryption/crypt/src/atom.c
deleted file mode 100644
index 8e9c4940abd..00000000000
--- a/xlators/encryption/crypt/src/atom.c
+++ /dev/null
@@ -1,861 +0,0 @@
-/*
-  Copyright (c) 2008-2013 Red Hat, Inc. <http://www.redhat.com>
-  This file is part of GlusterFS.
-
-  This file is licensed to you under your choice of the GNU Lesser
-  General Public License, version 3 or any later version (LGPLv3 or
-  later), or the GNU General Public License, version 2 (GPLv2), in all
-  cases as published by the Free Software Foundation.
-*/
-
-#include "defaults.h"
-#include "crypt-common.h"
-#include "crypt.h"
-
-/*
- *                              Glossary
- *
- *
- * cblock            (or cipher block). A logical unit in a file.
- *                   cblock size is defined as the number of bits
- *                   in an input (or output) block of the block
- *                   cipher (*). Cipher block size is a property of
- *                   cipher algorithm. E.g. cblock size is 64 bits
- *                   for DES, 128 bits for AES, etc.
- *
- * atomic cipher     A cipher algorithm, which requires some chunks of
- * algorithm         text to be padded at left and(or) right sides before
- *                   cipher transaform.
- *
- *
- * block (atom)      Minimal chunk of file's data, which doesn't require
- *                   padding. We'll consider logical units in a file of
- *                   block size (atom size).
- *
- * cipher algorithm  Atomic cipher algorithm, which requires the last
- * with EOF issue    incomplete cblock in a file to be padded with some
- *                   data (usually zeros).
- *
- *
- * operation, which  reading/writing from offset, which is not aligned to
- * forms a gap at    to atom size
- * the beginning
- *
- *
- * operation, which  reading/writing count bytes starting from offset off,
- * forms a gap at    so that off+count is not aligned to atom_size
- * the end
- *
- * head block        the first atom affected by an operation, which forms
- *                   a gap at the beginning, or(and) at the end.
- *                   Сomment. Head block has at least one gap (either at
- *                   the beginning, or at the end)
- *
- *
- * tail block        the last atom different from head, affected by an
- *                   operation, which forms a gap at the end.
- *                   Сomment: Tail block has exactly one gap (at the end).
- *
- *
- * partial block     head or tail block
- *
- *
- * full block        block without gaps.
- *
- *
- * (*) Recommendation for Block Cipher Modes of Operation
- *     Methods and Techniques
- *     NIST Special Publication 800-38A Edition 2001
- */
-
-/*
- * atom->offset_at()
- */
-static off_t
-offset_at_head(struct avec_config *conf)
-{
-    return conf->aligned_offset;
-}
-
-static off_t
-offset_at_hole_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_head(get_hole_conf(frame));
-}
-
-static off_t
-offset_at_data_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_head(get_data_conf(frame));
-}
-
-static off_t
-offset_at_tail(struct avec_config *conf, struct object_cipher_info *object)
-{
-    return conf->aligned_offset +
-           (conf->off_in_head ? get_atom_size(object) : 0) +
-           (conf->nr_full_blocks << get_atom_bits(object));
-}
-
-static off_t
-offset_at_hole_tail(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_tail(get_hole_conf(frame), object);
-}
-
-static off_t
-offset_at_data_tail(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_tail(get_data_conf(frame), object);
-}
-
-static off_t
-offset_at_full(struct avec_config *conf, struct object_cipher_info *object)
-{
-    return conf->aligned_offset +
-           (conf->off_in_head ? get_atom_size(object) : 0);
-}
-
-static off_t
-offset_at_data_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_full(get_data_conf(frame), object);
-}
-
-static off_t
-offset_at_hole_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_at_full(get_hole_conf(frame), object);
-}
-
-/*
- * atom->io_size_nopad()
- */
-
-static uint32_t
-io_size_nopad_head(struct avec_config *conf, struct object_cipher_info *object)
-{
-    uint32_t gap_at_beg;
-    uint32_t gap_at_end;
-
-    check_head_block(conf);
-
-    gap_at_beg = conf->off_in_head;
-
-    if (has_tail_block(conf) || has_full_blocks(conf) || conf->off_in_tail == 0)
-        gap_at_end = 0;
-    else
-        gap_at_end = get_atom_size(object) - conf->off_in_tail;
-
-    return get_atom_size(object) - (gap_at_beg + gap_at_end);
-}
-
-static uint32_t
-io_size_nopad_tail(struct avec_config *conf, struct object_cipher_info *object)
-{
-    check_tail_block(conf);
-    return conf->off_in_tail;
-}
-
-static uint32_t
-io_size_nopad_full(struct avec_config *conf, struct object_cipher_info *object)
-{
-    check_full_block(conf);
-    return get_atom_size(object);
-}
-
-static uint32_t
-io_size_nopad_data_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_head(get_data_conf(frame), object);
-}
-
-static uint32_t
-io_size_nopad_hole_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_head(get_hole_conf(frame), object);
-}
-
-static uint32_t
-io_size_nopad_data_tail(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_tail(get_data_conf(frame), object);
-}
-
-static uint32_t
-io_size_nopad_hole_tail(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_tail(get_hole_conf(frame), object);
-}
-
-static uint32_t
-io_size_nopad_data_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_full(get_data_conf(frame), object);
-}
-
-static uint32_t
-io_size_nopad_hole_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return io_size_nopad_full(get_hole_conf(frame), object);
-}
-
-static uint32_t
-offset_in_head(struct avec_config *conf)
-{
-    check_cursor_head(conf);
-
-    return conf->off_in_head;
-}
-
-static uint32_t
-offset_in_tail(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return 0;
-}
-
-static uint32_t
-offset_in_full(struct avec_config *conf, struct object_cipher_info *object)
-{
-    check_cursor_full(conf);
-
-    if (has_head_block(conf))
-        return (conf->cursor - 1) << get_atom_bits(object);
-    else
-        return conf->cursor << get_atom_bits(object);
-}
-
-static uint32_t
-offset_in_data_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_in_head(get_data_conf(frame));
-}
-
-static uint32_t
-offset_in_hole_head(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_in_head(get_hole_conf(frame));
-}
-
-static uint32_t
-offset_in_data_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_in_full(get_data_conf(frame), object);
-}
-
-static uint32_t
-offset_in_hole_full(call_frame_t *frame, struct object_cipher_info *object)
-{
-    return offset_in_full(get_hole_conf(frame), object);
-}
-
-/*
- * atom->rmw()
- */
-/*
- * Pre-conditions:
- * @vec contains plain text of the latest
- * version.
- *
- * Uptodate gaps of the @partial block with
- * this plain text, encrypt the whole block
- * and write the result to disk.
- */
-static int32_t
-rmw_partial_block(call_frame_t *frame, void *cookie, xlator_t *this,
-                  int32_t op_ret, int32_t op_errno, struct iovec *vec,
-                  int32_t count, struct iatt *stbuf, struct iobref *iobref,
-                  struct rmw_atom *atom)
-{
-    size_t was_read = 0;
-    uint64_t file_size;
-    crypt_local_t *local = frame->local;
-    struct object_cipher_info *object = &local->info->cinfo;
-
-    struct iovec *partial = atom->get_iovec(frame, 0);
-    struct avec_config *conf = atom->get_config(frame);
-    end_writeback_handler_t end_writeback_partial_block;
-#if DEBUG_CRYPT
-    gf_boolean_t check_last_cblock = _gf_false;
-#endif
-    local->op_ret = op_ret;
-    local->op_errno = op_errno;
-
-    if (op_ret < 0)
-        goto exit;
-
-    file_size = local->cur_file_size;
-    was_read = op_ret;
-
-    if (atom->locality == HEAD_ATOM && conf->off_in_head) {
-        /*
-         * head atom with a non-uptodate gap
-         * at the beginning
-         *
-         * fill the gap with plain text of the
-         * latest version. Convert a part of hole
-         * (if any) to zeros.
-         */
-        int32_t i;
-        int32_t copied = 0;
-        int32_t to_gap; /* amount of data needed to uptodate
-                           the gap at the beginning */
-#if 0
-		int32_t hole = 0; /* The part of the hole which
-				   * got in the head block */
-#endif /* 0 */
-        to_gap = conf->off_in_head;
-
-        if (was_read < to_gap) {
-            if (file_size > offset_at_head(conf) + was_read) {
-                /*
-                 * It is impossible to uptodate
-                 * head block: too few bytes have
-                 * been read from disk, so that
-                 * partial write is impossible.
-                 *
-                 * It could happen because of many
-                 * reasons: IO errors, (meta)data
-                 * corruption in the local file system,
-                 * etc.
-                 */
-                gf_log(this->name, GF_LOG_WARNING,
-                       "Can not uptodate a gap at the beginning");
-                local->op_ret = -1;
-                local->op_errno = EIO;
-                goto exit;
-            }
-#if 0
-			hole = to_gap - was_read;
-#endif /* 0 */
-            to_gap = was_read;
-        }
-        /*
-         * uptodate the gap at the beginning
-         */
-        for (i = 0; i < count && copied < to_gap; i++) {
-            int32_t to_copy;
-
-            to_copy = vec[i].iov_len;
-            if (to_copy > to_gap - copied)
-                to_copy = to_gap - copied;
-
-            memcpy(partial->iov_base, vec[i].iov_base, to_copy);
-            copied += to_copy;
-        }
-#if 0
-		/*
-		 * If possible, convert part of the
-		 * hole, which got in the head block
-		 */
-		ret = TRY_LOCK(&local->hole_lock);
-		if (!ret) {
-			if (local->hole_handled)
-				/*
-				 * already converted by
-				 * crypt_writev_cbk()
-				 */
-				UNLOCK(&local->hole_lock);
-			else {
-				/*
-				 * convert the part of the hole
-				 * which got in the head block
-				 * to zeros.
-				 *
-				 * Update the orig_offset to make
-				 * sure writev_cbk() won't care
-				 * about this part of the hole.
-				 *
-				 */
-				memset(partial->iov_base + to_gap, 0, hole);
-
-				conf->orig_offset -= hole;
-				conf->orig_size += hole;
-				UNLOCK(&local->hole_lock);
-			}
-		}
-		else /*
-		      * conversion is being performed
-		      * by crypt_writev_cbk()
-		      */
-			;
-#endif /* 0 */
-    }
-    if (atom->locality == TAIL_ATOM ||
-        (!has_tail_block(conf) && conf->off_in_tail)) {
-        /*
-         * tail atom, or head atom with a non-uptodate
-         * gap at the end.
-         *
-         * fill the gap at the end of the block
-         * with plain text of the latest version.
-         * Pad the result, (if needed)
-         */
-        int32_t i;
-        int32_t to_gap;
-        int copied;
-        off_t off_in_tail;
-        int32_t to_copy;
-
-        off_in_tail = conf->off_in_tail;
-        to_gap = conf->gap_in_tail;
-
-        if (to_gap && was_read < off_in_tail + to_gap) {
-            /*
-             * It is impossible to uptodate
-             * the gap at the end: too few bytes
-             * have been read from disk, so that
-             * partial write is impossible.
-             *
-             * It could happen because of many
-             * reasons: IO errors, (meta)data
-             * corruption in the local file system,
-             * etc.
-             */
-            gf_log(this->name, GF_LOG_WARNING,
-                   "Can not uptodate a gap at the end");
-            local->op_ret = -1;
-            local->op_errno = EIO;
-            goto exit;
-        }
-        /*
-         * uptodate the gap at the end
-         */
-        copied = 0;
-        to_copy = to_gap;
-        for (i = count - 1; i >= 0 && to_copy > 0; i--) {
-            uint32_t from_vec, off_in_vec;
-
-            off_in_vec = 0;
-            from_vec = vec[i].iov_len;
-            if (from_vec > to_copy) {
-                off_in_vec = from_vec - to_copy;
-                from_vec = to_copy;
-            }
-            memcpy(partial->iov_base + off_in_tail + to_gap - copied - from_vec,
-                   vec[i].iov_base + off_in_vec, from_vec);
-
-            gf_log(
-                this->name, GF_LOG_DEBUG,
-                "uptodate %d bytes at tail. Offset at target(source): %d(%d)",
-                (int)from_vec, (int)off_in_tail + to_gap - copied - from_vec,
-                (int)off_in_vec);
-
-            copied += from_vec;
-            to_copy -= from_vec;
-        }
-        partial->iov_len = off_in_tail + to_gap;
-
-        if (object_alg_should_pad(object)) {
-            int32_t resid = 0;
-            resid = partial->iov_len & (object_alg_blksize(object) - 1);
-            if (resid) {
-                /*
-                 * append a new EOF padding
-                 */
-                local->eof_padding_size = object_alg_blksize(object) - resid;
-
-                gf_log(this->name, GF_LOG_DEBUG, "set padding size %d",
author	Vijay Bellur <vijay@gluster.com>	2010-08-17 03:37:46 +0000
committer	Anand V. Avati <avati@dev.gluster.com>	2010-08-17 02:23:59 -0700
commit	e5737546071bd1d38430503bb64b9998bd7870dd (patch)
tree	c5a9fe82f2474ff6c8792381c1b77384c9f82983 /cli/src/cli-cmd.h
parent	ead92558ee8ec71442b2fa2da1a0df197ee78467 (diff)