featuress/changelog: On snapshot, notify irrespective of failures

During snapshot, changelog barrier is enabled and a
explicit rollover of changelog is initiated. During
rollover of changelog, if any error or changelog is
empty, the notification was not sent to reconfigure
and hence snapshot was failing because of timeout.
This patch addresses it by sending notification
irrespective of failures and sends error if any
back to barrier.

Change-Id: I898af624b44555281a9e43c69066077e0e121c17
BUG: 1225542
Signed-off-by: Kotresh HR <khiremat@redhat.com>
Reviewed-on: http://review.gluster.org/10951
Tested-by: NetBSD Build System <jenkins@build.gluster.org>
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Aravinda VK <avishwan@redhat.com>
Reviewed-by: Venky Shankar <vshankar@redhat.com>

4 files changed, 63 insertions, 23 deletions

diff --git a/tests/bugs/changelog/bug-1225542.t b/tests/bugs/changelog/bug-1225542.t
new file mode 100644
index 00000000000..a646aa88014
--- /dev/null
+++ b/tests/bugs/changelog/bug-1225542.t
@@ -0,0 +1,30 @@
+#!/bin/bash
+
+#Testcase:
+#On snapshot, notify changelog reconfigure upon explicit rollover
+#irrespective of any failures and send error back to barrier if any.
+
+. $(dirname $0)/../../include.rc
+. $(dirname $0)/../../snapshot.rc
+
+cleanup;
+TEST verify_lvm_version
+TEST glusterd;
+TEST pidof glusterd;
+
+TEST setup_lvm 1
+
+## Create a volume
+TEST $CLI volume create $V0 $H0:$L1
+
+## Start volume and verify
+TEST $CLI volume start $V0
+
+TEST $CLI volume set $V0 changelog.changelog on
+##Wait for changelog init to complete.
+sleep 1
+
+## Take snapshot
+TEST $CLI snapshot create snap1 $V0
+
+cleanup;
diff --git a/xlators/features/changelog/src/changelog-helpers.c b/xlators/features/changelog/src/changelog-helpers.c
index 11e60dcbf4a..a604891f0b3 100644
--- a/xlators/features/changelog/src/changelog-helpers.c
+++ b/xlators/features/changelog/src/changelog-helpers.c
@@ -493,30 +493,33 @@ changelog_rollover_changelog (xlator_t *this,
                 ev.ev_type = CHANGELOG_OP_TYPE_JOURNAL;
                 memcpy (ev.u.journal.path, nfile, strlen (nfile) + 1);
                 changelog_dispatch_event (this, priv, &ev);
+        }
+ out:
+        /* If this is explicit rollover initiated by snapshot,
+         * wakeup reconfigure thread waiting for changelog to
+         * rollover. This should happen even in failure cases as
+         * well otherwise snapshot will timeout and fail. Hence
+         * moved under out.
+         */
+        if (priv->explicit_rollover) {
+                priv->explicit_rollover = _gf_false;
 
-                /* If this is explicit rollover initiated by snapshot,
-                 * wakeup reconfigure thread waiting for changelog to
-                 * rollover
-                 */
-                if (priv->explicit_rollover) {
-                        priv->explicit_rollover = _gf_false;
-
-                        ret = pthread_mutex_lock (&priv->bn.bnotify_mutex);
-                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
-                        {
-                                priv->bn.bnotify = _gf_false;
-                                ret = pthread_cond_signal
-                                        (&priv->bn.bnotify_cond);
-                                CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
-                                gf_log (this->name, GF_LOG_INFO,
-                                        "Changelog published: %s signalled"
-                                        " bnotify", nfile);
+                pthread_mutex_lock (&priv->bn.bnotify_mutex);
+                {
+                        if (ret) {
+                                priv->bn.bnotify_error = _gf_true;
+                                gf_log (this->name, GF_LOG_ERROR, "Fail "
+                                        "snapshot because of previous errors");
+                        } else {
+                                gf_log (this->name, GF_LOG_INFO, "Explicit "
+                                        "rollover changelog: %s signaling "
+                                        "bnotify", nfile);
                         }
-                        ret = pthread_mutex_unlock (&priv->bn.bnotify_mutex);
-                        CHANGELOG_PTHREAD_ERROR_HANDLE_0 (ret, out);
+                        priv->bn.bnotify = _gf_false;
+                        pthread_cond_signal (&priv->bn.bnotify_cond);
                 }
+                pthread_mutex_unlock (&priv->bn.bnotify_mutex);
         }
- out:
         return ret;
 }
 
diff --git a/xlators/features/changelog/src/changelog-helpers.h b/xlators/features/changelog/src/changelog-helpers.h
index ccfbf9113d8..4416b9cdfdb 100644
--- a/xlators/features/changelog/src/changelog-helpers.h
+++ b/xlators/features/changelog/src/changelog-helpers.h
@@ -152,6 +152,7 @@ typedef struct barrier_notify {
          pthread_mutex_t        bnotify_mutex;
          pthread_cond_t         bnotify_cond;
          gf_boolean_t           bnotify;
+         gf_boolean_t           bnotify_error;
 } barrier_notify_t;
 
 /* Two separate mutex and conditional variable set is used
diff --git a/xlators/features/changelog/src/changelog.c b/xlators/features/changelog/src/changelog.c
index 91b0ae7a6e5..ed29bb83556 100644
--- a/xlators/features/changelog/src/changelog.c
+++ b/xlators/features/changelog/src/changelog.c
@@ -1958,6 +1958,7 @@ notify (xlator_t *this, int event, void *data, ...)
         int                     barrier         = DICT_DEFAULT;
         gf_boolean_t            bclean_req      = _gf_false;
         int                     ret             = 0;
+        int                     ret1            = 0;
         struct list_head        queue           = {0, };
 
         INIT_LIST_HEAD (&queue);
@@ -2108,13 +2109,17 @@ notify (xlator_t *this, int event, void *data, ...)
                                                                           out,
                                                                     bclean_req);
                                 }
+                                if (priv->bn.bnotify_error == _gf_true) {
+                                        ret = -1;
+                                        priv->bn.bnotify_error = _gf_false;
+                                }
                         }
-                        ret = pthread_mutex_unlock (&priv->bn.bnotify_mutex);
-                        CHANGELOG_PTHREAD_ERROR_HANDLE_1 (ret, out, bclean_req);
+                        ret1 = pthread_mutex_unlock (&priv->bn.bnotify_mutex);
+                        CHANGELOG_PTHREAD_ERROR_HANDLE_1 (ret1, out,
+                                                          bclean_req);
                         gf_log (this->name, GF_LOG_INFO,
                                 "Woke up: bnotify conditional wait");
 
-                        ret = 0;
                         goto out;
 
                 case DICT_DEFAULT:
@@ -2632,6 +2637,7 @@ init (xlator_t *this)
 
         /* Mutex is not needed as threads are not spawned yet */
         priv->bn.bnotify = _gf_false;
+        priv->bn.bnotify_error = _gf_false;
         ret = changelog_barrier_pthread_init (this, priv);
         if (ret)
                 goto cleanup_options;