cluster/*: get logic to calculate min() of the 'stime' xattr

* in both distribute and replicate (ignoring stripe for now),
  add logic to calculate the min() of stime values.

* What is a 'stime' ? Why is this required:
  -  stime means 'slave xtime', mainly used to keep track of slave
  node's sync status when distributed geo-replication is used.
  Logic of calculating 'min()' for this stime is very important as
  in case of crashes/reboots/shutdown, we will have to 'restart'
  with crawling from stime time value from the mount point, which
  gives the 'min()' of all the bricks, which means, we don't miss
  syncing any files in the above cases.

Change-Id: I2be8d434326572be9d4986db665570a6181db1ee
BUG: 847839
Original Author: Amar Tumballi <amarts@redhat.com>
Signed-off-by: Avra Sengupta <asengupt@redhat.com>
Reviewed-on: http://review.gluster.org/4893
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

5 files changed, 121 insertions, 2 deletions

diff --git a/libglusterfs/src/glusterfs.h b/libglusterfs/src/glusterfs.h
index fbf4722cd..65c3c6c10 100644
--- a/libglusterfs/src/glusterfs.h
+++ b/libglusterfs/src/glusterfs.h
@@ -112,6 +112,7 @@
 #define GLUSTERFS_PARENT_ENTRYLK "glusterfs.parent-entrylk"
 #define QUOTA_SIZE_KEY "trusted.glusterfs.quota.size"
 #define GFID_TO_PATH_KEY "glusterfs.gfid2path"
+#define GF_XATTR_STIME_PATTERN "trusted.glusterfs.*.stime"
 
 /* Index xlator related */
 #define GF_XATTROP_INDEX_GFID "glusterfs.xattrop_index_gfid"
diff --git a/xlators/cluster/afr/src/afr-inode-read.c b/xlators/cluster/afr/src/afr-inode-read.c
index 051b7b164..1fc8f613e 100644
--- a/xlators/cluster/afr/src/afr-inode-read.c
+++ b/xlators/cluster/afr/src/afr-inode-read.c
@@ -1323,6 +1323,62 @@ afr_getxattr_pathinfo_cbk (call_frame_t *frame, void *cookie,
         return ret;
 }
 
+static int
+afr_aggregate_stime_xattr (dict_t *this, char *key, data_t *value, void *data)
+{
+        int ret = 0;
+
+        if (fnmatch (GF_XATTR_STIME_PATTERN, key, FNM_NOESCAPE) == 0)
+                ret = gf_get_min_stime (THIS, data, key, value);
+
+        return ret;
+}
+
+int32_t
+afr_common_getxattr_stime_cbk (call_frame_t *frame, void *cookie,
+                               xlator_t *this, int32_t op_ret, int32_t op_errno,
+                               dict_t *dict, dict_t *xdata)
+{
+        afr_local_t *local          = NULL;
+        int32_t      callcnt        = 0;
+
+        if (!frame || !frame->local || !this) {
+                gf_log ("", GF_LOG_ERROR, "possible NULL deref");
+                goto out;
+        }
+
+        local = frame->local;
+
+        LOCK (&frame->lock);
+        {
+                callcnt = --local->call_count;
+
+                if (!dict || (op_ret < 0)) {
+                        local->op_errno = op_errno;
+                        goto cleanup;
+                }
+
+                if (!local->dict)
+                        local->dict = dict_copy_with_ref (dict, NULL);
+                else
+                        dict_foreach (dict, afr_aggregate_stime_xattr,
+                                      local->dict);
+                local->op_ret = 0;
+        }
+
+cleanup:
+        UNLOCK (&frame->lock);
+
+        if (!callcnt) {
+                AFR_STACK_UNWIND (getxattr, frame, local->op_ret,
+                                  local->op_errno, local->dict, xdata);
+        }
+
+out:
+        return 0;
+}
+
+
 static gf_boolean_t
 afr_is_special_xattr (const char *name, fop_getxattr_cbk_t *cbk,
                       gf_boolean_t is_fgetxattr)
@@ -1355,6 +1411,8 @@ afr_is_special_xattr (const char *name, fop_getxattr_cbk_t *cbk,
                 } else {
                         *cbk = afr_getxattr_lockinfo_cbk;
                 }
+        } else if (fnmatch (GF_XATTR_STIME_PATTERN, name, FNM_NOESCAPE) == 0) {
+                *cbk = afr_common_getxattr_stime_cbk;
         } else {
                 is_spl = _gf_false;
         }
diff --git a/xlators/cluster/dht/src/dht-common.c b/xlators/cluster/dht/src/dht-common.c
index c7bbfd160..f1bc99be6 100644
--- a/xlators/cluster/dht/src/dht-common.c
+++ b/xlators/cluster/dht/src/dht-common.c
@@ -62,6 +62,11 @@ dht_aggregate (dict_t *this, char *key, data_t *value, void *data)
                 }
 
                 *size = hton64 (ntoh64 (*size) + ntoh64 (*ptr));
+
+        } else if (fnmatch (GF_XATTR_STIME_PATTERN, key, FNM_NOESCAPE) == 0) {
+                ret = gf_get_min_stime (THIS, dst, key, value);
+                if (ret < 0)
+                        return ret;
         } else {
                 /* compare user xattrs only */
                 if (!strncmp (key, "user.", strlen ("user."))) {
diff --git a/xlators/lib/src/libxlator.c b/xlators/lib/src/libxlator.c
index 624a929d0..8ae3ff314 100644
--- a/xlators/lib/src/libxlator.c
+++ b/xlators/lib/src/libxlator.c
@@ -357,3 +357,58 @@ err:
         return -1;
 
 }
+
+int
+gf_get_min_stime (xlator_t *this, dict_t *dst, char *key, data_t *value)
+{
+        int ret = -1;
+        uint32_t *net_timebuf = NULL;
+        uint32_t *value_timebuf = NULL;
+        uint32_t host_timebuf[2] = {0,};
+        uint32_t host_value_timebuf[2] = {0,};
+
+        /* stime should be minimum of all the other nodes */
+        ret = dict_get_bin (dst, key, (void **)&net_timebuf);
+        if (ret < 0) {
+                net_timebuf = GF_CALLOC (1, sizeof (int64_t),
+                                           gf_common_mt_char);
+                if (!net_timebuf)
+                        goto out;
+
+                ret = dict_set_bin (dst, key, net_timebuf, sizeof (int64_t));
+                if (ret < 0) {
+                        gf_log (this->name, GF_LOG_WARNING,
+                                "key=%s: dict set failed", key);
+                        goto error;
+                }
+        }
+
+        value_timebuf = data_to_bin (value);
+        if (!value_timebuf) {
+                gf_log (this->name, GF_LOG_WARNING,
+                        "key=%s: getting value of stime failed", key);
+                ret = -1;
+                goto out;
+        }
+
+        get_hosttime (value_timebuf, host_value_timebuf);
+        get_hosttime (net_timebuf, host_timebuf);
+
+        /* can't use 'min()' macro here as we need to compare two fields
+           in the array, selectively */
+        if ((host_value_timebuf[0] > host_timebuf[0]) ||
+            ((host_value_timebuf[0] == host_timebuf[0]) &&
+             (host_value_timebuf[1] > host_timebuf[1]))) {
+                update_timebuf (value_timebuf, net_timebuf);
+        }
+
+        ret = 0;
+out:
+        return ret;
+error:
+        /* To be used only when net_timebuf is not set in the dict */
+        if (net_timebuf)
+                GF_FREE (net_timebuf);
+
+        return ret;
+}
diff --git a/xlators/lib/src/libxlator.h b/xlators/lib/src/libxlator.h
index 7e75c9482..c89629e9d 100644
--- a/xlators/lib/src/libxlator.h
+++ b/xlators/lib/src/libxlator.h
@@ -98,7 +98,7 @@ cluster_getmarkerattr (call_frame_t *frame,xlator_t *this, loc_t *loc,
 int
 match_uuid_local (const char *name, char *uuid);
 
-
-
+int
+gf_get_min_stime (xlator_t *this, dict_t *dst, char *key, data_t *value);
 
 #endif /* !_LIBXLATOR_H */