summaryrefslogtreecommitdiffstats
path: root/xlators/mgmt/glusterd/src/glusterd-sm.c
diff options
context:
space:
mode:
authorKrishnan Parthasarathi <kparthas@redhat.com>2013-04-15 15:56:57 +0530
committerVijay Bellur <vbellur@redhat.com>2013-04-17 05:48:50 -0700
commit92729add67e2e7b8c7589c2dfab0bde071a7faf2 (patch)
tree4e8c2d48de34938a71abd320402e98e1588aee30 /xlators/mgmt/glusterd/src/glusterd-sm.c
parent47c118e22d9d6fb6662fe96841ed4fe3089739b5 (diff)
glusterd: big lock - a coarse-grained locking to prevent racesv3.4.0alpha3
There are primarily three lists that are part of glusterd process, that are concurrently accessed. Namely, priv->volumes, priv->peers and volinfo->bricks_list. Big-lock approach ----------------- WHAT IS IT? Big lock is a coarse-grained lock which protects all three lists, mentioned above, from racy access. HOW DOES IT WORK? At any given point in time, glusterd's thread(s) are in execution _iff_ there is a preceding, inbound network event. Of course, the sigwaiter thread and timer thread are exceptions. A network event is an external trigger to glusterd, via the epoll thread, in the form of POLLIN and POLLERR. As long as we take the big-lock at all such entry points and yield it when we are done, we are guaranteed that all the network events, accessing the global lists, are serialised. This amounts to holding the big lock at - all the handlers of all the actors in glusterd. (POLLIN) - all the cbks in glusterd. (POLLIN) - rpc_notify (DISCONNECT event), if we access/modify one of the three lists. (POLLERR) In the case of synctask'ized volume operations, we must remember that, if we held the big lock for the entire duration of the handler, we may block other non-synctask rpc actors from executing. For eg, volume-start would block in PMAP SIGNIN, if done incorrectly. To prevent this, we need to yield the big lock, when we yield the synctask, and reacquire on waking up of the synctask. BUG: 948686 Change-Id: I429832f1fed67bcac0813403d58346558a403ce9 Signed-off-by: Krishnan Parthasarathi <kparthas@redhat.com> Reviewed-on: http://review.gluster.org/4835 Tested-by: Gluster Build System <jenkins@build.gluster.com> Reviewed-by: Vijay Bellur <vbellur@redhat.com>
Diffstat (limited to 'xlators/mgmt/glusterd/src/glusterd-sm.c')
-rw-r--r--xlators/mgmt/glusterd/src/glusterd-sm.c19
1 files changed, 18 insertions, 1 deletions
diff --git a/xlators/mgmt/glusterd/src/glusterd-sm.c b/xlators/mgmt/glusterd/src/glusterd-sm.c
index 5a38fdf..a82ca2e 100644
--- a/xlators/mgmt/glusterd/src/glusterd-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-sm.c
@@ -1075,8 +1075,25 @@ glusterd_friend_sm ()
ret = 0;
out:
- if (quorum_action)
+ if (quorum_action) {
+ /* When glusterd is restarted, it needs to wait until the 'friends' view
+ * of the volumes settle, before it starts any of the internal daemons.
+ *
+ * Every friend that was part of the cluster, would send its
+ * cluster-view, 'our' way. For every friend, who belongs to
+ * a partition which has a different cluster-view from our
+ * partition, we may update our cluster-view. For subsequent
+ * friends from that partition would agree with us, if the first
+ * friend wasn't rejected. For every first friend, whom we agreed with,
+ * we would need to start internal daemons/bricks belonging to the
+ * new volumes.
+ * glusterd_spawn_daemons calls functions that are idempotent. ie,
+ * the functions spawn process(es) only if they are not started yet.
+ *
+ * */
+ glusterd_spawn_daemons ((void*) _gf_false);
glusterd_do_quorum_action ();
+ }
return ret;
}