summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorShyamsundarR <srangana@redhat.com>2018-07-26 07:09:40 -0400
committerAtin Mukherjee <amukherj@redhat.com>2018-08-09 03:16:38 +0000
commit511af2e786f5e881bc4c53287407235dd6fe2704 (patch)
treecbd6d85bc34668725225ae84c5f18aa3456c094f
parentbb9b8f61501cc633e585593de4d9f2fe5494d5ce (diff)
tests: Increase timeout for mpx restart crash test
In lcov based regression testing environments, all tests take more time than what occurs in centos7 regressions. Possibly due to code instrumentation for lcov purposes. Due to this the test, bug-1432542-mpx-restart-crash.t constantly times out. This patch increases the timeout for the same to enable lcov tests to pass on a more regular basis. It was also noted by Nithya that the test at times generated an OOM kill on the regression machines. In order to reduce runtime memory foot print of the tests, FUSE mounts are unmounted as soon as the required test is complete. Fixes: bz#1608568 Change-Id: I37f8d4b45807a69c52c7c7df4923c0fc33fab4e4 Signed-off-by: ShyamsundarR <srangana@redhat.com>
-rw-r--r--tests/bugs/core/bug-1432542-mpx-restart-crash.t9
1 files changed, 6 insertions, 3 deletions
diff --git a/tests/bugs/core/bug-1432542-mpx-restart-crash.t b/tests/bugs/core/bug-1432542-mpx-restart-crash.t
index 7f61c950fb7..5eee2b986cb 100644
--- a/tests/bugs/core/bug-1432542-mpx-restart-crash.t
+++ b/tests/bugs/core/bug-1432542-mpx-restart-crash.t
@@ -1,6 +1,6 @@
#!/bin/bash
-SCRIPT_TIMEOUT=400
+SCRIPT_TIMEOUT=800
. $(dirname $0)/../../include.rc
. $(dirname $0)/../../volume.rc
@@ -50,7 +50,7 @@ create_volume () {
TEST $cmd
TEST $CLI volume start $vol_name
# check for 6 bricks and 1 shd daemon to be up and running
- EXPECT_WITHIN $PROCESS_UP_TIMEOUT 7 count_up_bricks $vol_name
+ EXPECT_WITHIN 120 7 count_up_bricks $vol_name
local mount_point=$(get_mount_point $1)
mkdir -p $mount_point
TEST $GFS -s $H0 --volfile-id=$vol_name $mount_point
@@ -77,10 +77,13 @@ TEST $CLI volume set all cluster.brick-multiplex on
# Our infrastructure can't handle an arithmetic expression here. The formula
# is (NUM_VOLS-1)*5 because it sees each TEST/EXPECT once but needs the other
# NUM_VOLS-1 and there are 5 such statements in each iteration.
-TESTS_EXPECTED_IN_LOOP=95
+TESTS_EXPECTED_IN_LOOP=114
for i in $(seq 1 $NUM_VOLS); do
create_volume $i
TEST dd if=/dev/zero of=$(get_mount_point $i)/a_file bs=4k count=1
+ # Unmounting to reduce memory footprint on regression hosts
+ mnt_point=$(get_mount_point $i)
+ EXPECT_WITHIN $UMOUNT_TIMEOUT "Y" force_umount $mnt_point
done
# Kill glusterd, and wait a bit for all traces to disappear.