[LU-13402] sanity test_252: Invalid number of mdtlov clients returned by /usr/sbin/lr_reader

• Type: Bug
• Resolution: Fixed
• Priority: Critical
• Fix Version/s: Lustre 2.14.0
• Affects Version/s: Lustre 2.14.0
• Severity: 3
• Assignee: Alex Zhuravlev
• Reporter: Maloo

Description

This issue was created by maloo for S Buisson <[e-mail]>

This issue relates to the following test suite run: https://testing.whamcloud.com/test_sets/59cb919c-f797-4928-908e-b8cc42e8b5bd

test_252 failed with the following error:

Invalid number of mdtlov clients returned by /usr/sbin/lr_reader

mdtlov for MDT0002 seems to be missing, as lr_reader output only shows lustre-MDT0001-mdtlov_UUID and lustre-MDT0003-mdtlov_UUID.
So MDT target local to MDT0 is missing?

VVVVVVV DO NOT REMOVE LINES BELOW, Added by Maloo for auto-association VVVVVVV
sanity test_252 - Invalid number of mdtlov clients returned by /usr/sbin/lr_reader

Issue Links

• is duplicated by: LU-13408 tgt_cancel_slc_locks()) ASSERTION( lock->l_client_cookie != 0 )
• is related to: LU-13469 MDS hung during mount
• is related to: LU-13379 interop b2_12/master: lustre-initialization failed: Unrecognized mount option "localrecov"

Comments

[Peter Jones]

Landed for 2.14

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/38138/
Subject: LU-13402 target: never exclude MDT/OST from last_rcvd
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 6682b74280ed778f8668f942c808eb70ed7bc67f

[Gerrit Updater]

Alex Zhuravlev ([e-mail]) uploaded a new patch: https://review.whamcloud.com/38138
Subject: LU-13402 target: never exclude MDT from last_rcvd
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 7962e141c6c6468527b9c329eeaa81b2f8a8f325

[Andreas Dilger]

This looks to be only failing for the LU-13379 patch.

[Andreas Dilger]

Alex, any thoughts on this? The 38022 patch is failing consistently due to moving the "-o localrecov" flag definition, so it is definitely the fault of the patch.

[Andreas Dilger]

This is seen during testing of patch https://review.whamcloud.com/38022 "LU-13379 tests: don't use localrecov for older servers" which just moves the "-o localrecov" setting later in the test setup so that we can check the server version before adding the mount optoin.

I thought that this patch is identical to the original code, but I'm wondering if somehow "localrecov" is causing the local MDT not to register itself in the last_rcvd file? In the failure case, lustre-MDT0002 is missing from the log on lustre-MDT0000, so it would be a "local" client in that respect.