glusterd: Fixed replace-brick commit_force algo.

- commit force subcommand of replace-brick (rb) should be allowed
  even if source brick is (irrecoverably) offline.
- modified rb_timer to be active only for start subcommand. This is
  important since, the rb timer event relies on src_brick and dst_brick
  objects to be 'alive' when it 'happens'. In the case of
  abort/commit/commit force it is very likely that src_brick and/or
  dst_brick objects could have been destroyed.

Change-Id: Ib8b8a4d690fbdd6f99b8aff306490eb59c54a437
BUG: 772845
Signed-off-by: Krishnan Parthasarathi <kp@gluster.com>
Reviewed-on: http://review.gluster.com/2620
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Reviewed-by: Amar Tumballi <amarts@redhat.com>
Reviewed-by: Vijay Bellur <vijay@gluster.com>

1 files changed, 5 insertions, 5 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index ce9581644..9718918f3 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -1863,12 +1863,12 @@ glusterd_op_start_rb_timer (dict_t *dict)
                 goto out;
         }
 
-        if (op == GF_REPLACE_OP_START ||
-            op == GF_REPLACE_OP_ABORT)
-                timeout.tv_sec  = 5;
-        else
-                timeout.tv_sec = 1;
+        if (op != GF_REPLACE_OP_START) {
+                ret = glusterd_op_sm_inject_all_acc ();
+                goto out;
+        }
 
+        timeout.tv_sec  = 5;
         timeout.tv_usec = 0;