glusterd: hold off volume deletes while still restarting bricks

We need to do this because modifying the volume/brick tree while
glusterd_restart_bricks is still walking it can lead to segfaults.
Without waiting we could accidentally "slip in" while attach_brick has
released big_lock between retries and make such a modification.

Change-Id: I30ccc4efa8d286aae847250f5d4fb28956a74b03
BUG: 1432542
Signed-off-by: Jeff Darcy <jeff@pl.atyp.us>
Reviewed-on: https://review.gluster.org/16927
Smoke: Gluster Build System <jenkins@build.gluster.org>
NetBSD-regression: NetBSD Build System <jenkins@build.gluster.org>
CentOS-regression: Gluster Build System <jenkins@build.gluster.org>
Reviewed-by: Atin Mukherjee <amukherj@redhat.com>

5 files changed, 45 insertions, 14 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index 55c2bda1dd1..3558ae91059 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -6044,6 +6044,15 @@ glusterd_op_stage_validate (glusterd_op_t op, dict_t *dict, char **op_errstr,
         return ret;
 }
 
+static void
+glusterd_wait_for_blockers (glusterd_conf_t *priv)
+{
+        while (priv->blockers) {
+                synclock_unlock (&priv->big_lock);
+                sleep (1);
+                synclock_lock (&priv->big_lock);
+        }
+}
 
 int32_t
 glusterd_op_commit_perform (glusterd_op_t op, dict_t *dict, char **op_errstr,
@@ -6063,18 +6072,22 @@ glusterd_op_commit_perform (glusterd_op_t op, dict_t *dict, char **op_errstr,
                         break;
 
                 case GD_OP_STOP_VOLUME:
+                        glusterd_wait_for_blockers (this->private);
                         ret = glusterd_op_stop_volume (dict);
                         break;
 
                 case GD_OP_DELETE_VOLUME:
+                        glusterd_wait_for_blockers (this->private);
                         ret = glusterd_op_delete_volume (dict);
                         break;
 
                 case GD_OP_ADD_BRICK:
+                        glusterd_wait_for_blockers (this->private);
                         ret = glusterd_op_add_brick (dict, op_errstr);
                         break;
 
                 case GD_OP_REPLACE_BRICK:
+                        glusterd_wait_for_blockers (this->private);
                         ret = glusterd_op_replace_brick (dict, rsp_dict);
                         break;
 
@@ -6082,11 +6095,13 @@ glusterd_op_commit_perform (glusterd_op_t op, dict_t *dict, char **op_errstr,
                         ret = glusterd_op_set_volume (dict, op_errstr);
                         break;
 
+
                 case GD_OP_RESET_VOLUME:
                         ret = glusterd_op_reset_volume (dict, op_errstr);
                         break;
 
                 case GD_OP_REMOVE_BRICK:
+                        glusterd_wait_for_blockers (this->private);
                         ret = glusterd_op_remove_brick (dict, op_errstr);
                         break;
 
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index f8525611ed4..a0df27c8bae 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -3134,8 +3134,8 @@ out:
 int
 glusterd_spawn_daemons (void *opaque)
 {
-        glusterd_conf_t *conf = THIS->private;
-        int             ret             = -1;
+        glusterd_conf_t *conf   = THIS->private;
+        int             ret     = -1;
 
         synclock_lock (&conf->big_lock);
         glusterd_restart_bricks (conf);
@@ -4904,9 +4904,13 @@ static int32_t
 my_callback (struct rpc_req *req, struct iovec *iov, int count, void *v_frame)
 {
         call_frame_t    *frame  = v_frame;
+        glusterd_conf_t *conf   = frame->this->private;
 
-        STACK_DESTROY (frame->root);
+        synclock_lock (&conf->big_lock);
+        --(conf->blockers);
+        synclock_unlock (&conf->big_lock);
 
+        STACK_DESTROY (frame->root);
         return 0;
 }
 
@@ -4923,6 +4927,7 @@ send_attach_req (xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
         void                            *req = &brick_req;
         void                            *errlbl   = &&err;
         struct rpc_clnt_connection      *conn;
+        glusterd_conf_t                 *conf     = this->private;
         extern struct rpc_clnt_program  gd_brick_prog;
 
         if (!rpc) {
@@ -4982,9 +4987,13 @@ send_attach_req (xlator_t *this, struct rpc_clnt *rpc, char *path, int op)
         iov.iov_len = ret;
 
         /* Send the msg */
+        ++(conf->blockers);
         ret = rpc_clnt_submit (rpc, &gd_brick_prog, op,
-                               my_callback, &iov, 1, NULL, 0, iobref, frame,
-                               NULL, 0, NULL, 0, NULL);
+                               my_callback, &iov, 1, NULL, 0, iobref,
+                               frame, NULL, 0, NULL, 0, NULL);
+        if (ret) {
+                --(conf->blockers);
+        }
         return ret;
 
 free_iobref:
@@ -5016,6 +5025,8 @@ attach_brick (xlator_t *this,
         char            full_id[PATH_MAX]       = {'\0',};
         char            path[PATH_MAX]          = {'\0',};
         int             ret;
+        int             tries;
+        rpc_clnt_t      *rpc;
 
         gf_log (this->name, GF_LOG_INFO,
                 "add brick %s to existing process for %s",
@@ -5052,12 +5063,15 @@ attach_brick (xlator_t *this,
         }
         (void) build_volfile_path (full_id, path, sizeof(path), NULL);
 
-        int tries = 0;
-        while (tries++ <= 15) {
-                ret = send_attach_req (this, other_brick->rpc, path,
-                                       GLUSTERD_BRICK_ATTACH);
-                if (!ret) {
-                        return 0;
+        for (tries = 15; tries > 0; --tries) {
+                rpc = rpc_clnt_ref (other_brick->rpc);
+                if (rpc) {
+                        ret = send_attach_req (this, rpc, path,
+                                               GLUSTERD_BRICK_ATTACH);
+                        rpc_clnt_unref (rpc);
+                        if (!ret) {
+                                return 0;
+                        }
                 }
                 /*
                  * It might not actually be safe to manipulate the lock like
@@ -5423,6 +5437,8 @@ glusterd_restart_bricks (glusterd_conf_t *conf)
         conf = this->private;
         GF_VALIDATE_OR_GOTO (this->name, conf, out);
 
+        ++(conf->blockers);
+
         ret = glusterd_get_quorum_cluster_counts (this, &active_count,
                                                   &quorum_count);
         if (ret)
@@ -5502,6 +5518,7 @@ glusterd_restart_bricks (glusterd_conf_t *conf)
         ret = 0;
 
 out:
+        --(conf->blockers);
         conf->restart_done = _gf_true;
         return ret;
 }
diff --git a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
index d7983f6fd71..5d164fd8df3 100644
--- a/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-volume-ops.c
@@ -2761,14 +2761,11 @@ glusterd_op_delete_volume (dict_t *dict)
 {
         int                                     ret = 0;
         char                                    *volname = NULL;
-        glusterd_conf_t                         *priv = NULL;
         glusterd_volinfo_t                      *volinfo = NULL;
         xlator_t                                *this = NULL;
 
         this = THIS;
         GF_ASSERT (this);
-        priv = this->private;
-        GF_ASSERT (priv);
 
         ret = dict_get_str (dict, "volname", &volname);
         if (ret) {
diff --git a/xlators/mgmt/glusterd/src/glusterd.c b/xlators/mgmt/glusterd/src/glusterd.c
index 11d4b8bc79f..d26e959d3a3 100644
--- a/xlators/mgmt/glusterd/src/glusterd.c
+++ b/xlators/mgmt/glusterd/src/glusterd.c
@@ -1867,6 +1867,7 @@ init (xlator_t *this)
         if (ret < 0)
                 goto out;
 
+        conf->blockers = 0;
         /* If the peer count is less than 2 then this would be the best time to
          * spawn process/bricks that may need (re)starting since last time
          * (this) glusterd was up. */
diff --git a/xlators/mgmt/glusterd/src/glusterd.h b/xlators/mgmt/glusterd/src/glusterd.h
index e5b7b9f8f38..a773a2b6221 100644
--- a/xlators/mgmt/glusterd/src/glusterd.h
+++ b/xlators/mgmt/glusterd/src/glusterd.h
@@ -190,6 +190,7 @@ typedef struct {
         int                        ping_timeout;
         uint32_t                   generation;
         int32_t                    workers;
+        uint32_t                   blockers;
 } glusterd_conf_t;