Tests: workaround NetBSD failures in cdc.t

The volume reset network.compression operation cause brick processes
to be restarted. If the volume is already started, a brick process
is already there and the restart will fail, as the brick TCP port
is already in use.

Because the new brick process is not started, the volume is left
with no brick online, and the volume stop operation will timeout
waiting for bricks to stop.

Obviosuly we have two bugs here
- If volume reset network.compression needs to restart the bricks,
  it should first make sure the previous brick process is terminated
- volume stop should not wait forever for bricks to come back online

This change does not fix the bugs but just makes sure the volume
is stoped before volume reset network.compression, so that the
failure oes not happen.

backport of: I9cd5cdc767ef6ee9dd31f2121d672dc3bfdce45f

BUG: 1212676
Change-Id: If85c46f7ed675c4c84cc4170292b1f0f41e5b456
Signed-off-by: Emmanuel Dreyfus <manu@netbsd.org>
Reviewed-on: http://review.gluster.org/10587
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Tested-by: NetBSD Build System
Reviewed-by: Vijay Bellur <vbellur@redhat.com>

1 files changed, 4 insertions, 4 deletions

diff --git a/tests/basic/cdc.t b/tests/basic/cdc.t
index 6a80b9287de..8653a77207a 100755
--- a/tests/basic/cdc.t
+++ b/tests/basic/cdc.t
@@ -132,15 +132,15 @@ TEST ! test -e /tmp/cdcdump.gz
 TEST rm -f /tmp/cdc* $M0/cdc*
 EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $M0
 
+## Stop the volume
+TEST $CLI volume stop $V0;
+EXPECT 'Stopped' volinfo_field $V0 'Status';
+
 ## Reset the network.compression options
 TEST $CLI volume reset $V0 network.compression.debug
 TEST $CLI volume reset $V0 network.compression.min-size
 TEST $CLI volume reset $V0 network.compression
 
-## Stop the volume
-TEST $CLI volume stop $V0;
-EXPECT 'Stopped' volinfo_field $V0 'Status';
-
 ## Delete the volume
 TEST $CLI volume delete $V0;
 TEST ! $CLI volume info $V0;