[LU-11161] sanity test_160g fails for DNE with ''mds2: User cl9 still registered''

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.13.0, Lustre 2.12.1
• Affects Version/s: Lustre 2.12.0
• Labels: DNE
• Severity: 3
• Assignee: James Nunez (Inactive)
• Reporter: Maloo

Description

test_160g failed with the following error:

'mds2: User cl9 still registered'

This issue relates to the following test suite run: https://testing.whamcloud.com/test_sets/d2eee862-8ad3-11e8-9e83-52540065bddc

This issue was created by maloo for James Nunez <[e-mail]>

<<Please provide additional information about the failure here>>

The patch for LU-10734, https://review.whamcloud.com/#/c/31604/, recently landed to master. It modifies sanity test 106g and removes that test from the ALWAYS_EXCEPT list. There seems to be an issue with the test since it fails in DNE testing when there is more than two MDSs.

Looking at the MDS console logs for the test session mentioned above, we see the following. On the first MDS with MDT0 and MDT2 we can clearly see that the changelog user cl9 is deregistered

[ 6260.597934] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0002.changelog_users
[ 6260.922657] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0002.changelog_users
[ 6264.392672] Lustre: 8849:0:(mdd_trans.c:187:mdd_chlg_garbage_collect()) lustre-MDD0000: Force deregister of ChangeLog user cl9 idle with more than 4 unprocessed records
[ 6264.600644] Lustre: DEBUG MARKER: ps -e -o comm= | grep chlg_gc_thread
[ 6264.928866] Lustre: DEBUG MARKER: ps -e -o comm= | grep chlg_gc_thread
[ 6265.932195] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0000.changelog_users
[ 6266.259457] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0000.changelog_users
[ 6266.586909] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0000.changelog_users

On the other MDS with MDT1 and MDT3, we don't see the same user deregistered. In fact we don't see any users deregistered

[ 6261.587345] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0003.changelog_users
[ 6261.913956] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0003.changelog_users
[ 6262.237009] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0003.changelog_users
[ 6265.271455] Lustre: DEBUG MARKER: ps -e -o comm= | grep chlg_gc_thread
[ 6265.602819] Lustre: DEBUG MARKER: ps -e -o comm= | grep chlg_gc_thread
[ 6266.927027] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0001.changelog_users
[ 6267.246822] Lustre: DEBUG MARKER: /usr/sbin/lctl get_param -n mdd.lustre-MDT0001.changelog_users
[ 6267.685450] Lustre: DEBUG MARKER: /usr/sbin/lctl mark  sanity test_160g: @@@@@@ FAIL: mds2: User cl9 still registered 

VVVVVVV DO NOT REMOVE LINES BELOW, Added by Maloo for auto-association VVVVVVV
sanity test_160g - 'mds2: User cl9 still registered'

Issue Links

• is related to: LU-10734 sanity test_160g: User cl8 still found in changelog_users

Comments

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/34297/
Subject: LU-11161 tests: start running sanity 160g again
Project: fs/lustre-release
Branch: b2_12
Current Patch Set:
Commit: 5eb52e556975def830fbe0a8c323bff09690b16a

[Gerrit Updater]

Minh Diep ([e-mail]) uploaded a new patch: https://review.whamcloud.com/34297
Subject: LU-11161 tests: start running sanity 160g again
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: 6455dce5440c587aa495acf27226c1e16144ff74

[James Nunez (Inactive)]

Landed for 2.13.0