[LU-12982] interop: conf-sanity test 5i fails with “ start mdt should fail “

Fri, 6 Aug 2021 16:54:30 +0000

conf-sanity test_5i started failing with “ start mdt should fail “ on 27 MAY 2019 for 2.12.2 and later servers and Lustre version 2.12.53.62 and later clients.

The issue is that conf-sanity test 5i was added by patch https://review.whamcloud.com/34724 on 25 MAY 2019 and included in Lustre version 2.12.54 which is after 2.12.2 was produced. We should skip this test for Lustre server versions 2.12.53 and earlier.

After test 5i fails, we see tests 6, 7, 8, 17 fail because they try to mount the MDT, but it is already mounted. For example, conf-sanity test 6, we see the following error

Starting mds1:   /dev/mapper/mds1_flakey /mnt/lustre-mds1
CMD: trevis-45vm12 mkdir -p /mnt/lustre-mds1; mount -t lustre   /dev/mapper/mds1_flakey /mnt/lustre-mds1
trevis-45vm12: mount.lustre: according to /etc/mtab /dev/mapper/mds1_flakey is already mounted on /mnt/lustre-mds1
Start of /dev/mapper/mds1_flakey on mds1 failed 17
 conf-sanity test_6: @@@@@@ FAIL: MDT start failed

Similarly, test 9 fails because the OSTs were never unmounted/cleaned up after test 5i failed.

Starting ost1:   /dev/mapper/ost1_flakey /mnt/lustre-ost1
CMD: trevis-45vm11 mkdir -p /mnt/lustre-ost1; mount -t lustre   /dev/mapper/ost1_flakey /mnt/lustre-ost1
trevis-45vm11: mount.lustre: according to /etc/mtab /dev/mapper/ost1_flakey is already mounted on /mnt/lustre-ost1
Start of /dev/mapper/ost1_flakey on ost1 failed 17
 conf-sanity test_9: @@@@@@ FAIL: OST start failed

Logs for these failures are at
https://testing.whamcloud.com/test_sets/cc8aab4a-07a6-11ea-bbc3-52540065bddc
https://testing.whamcloud.com/test_sets/9d33a094-cc4c-11e9-90ad-52540065bddc

Whamcloud Community JIRA

[LU-12982] interop: conf-sanity test 5i fails with “ start mdt should fail “