[LU-12691] obd_max_recoverable_clients is not atomic

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.13.0, Lustre 2.12.4
• Affects Version/s: Lustre 2.10.8
• Severity: 3
• Assignee: Tatsushi Takamura
• Reporter: Tatsushi Takamura

Description

Originally obd_max_recoverable_clients is not increased at the same moment.
But because of LU-3540, it will be increased by multiple processes.
This causes LBUG.

The type of obd_max_recoverable_clients should be atomic_t and be handled by atomic operations.

Issue Links

• duplicates: LU-11001 insanity test_2: :check_for_recovery_ready()) ASSERTION( clnts <= obd->obd_max_recoverable_clients )

Comments

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/36878/
Subject: LU-12691 ldlm: obd_max_recoverable_clients is not atomic
Project: fs/lustre-release
Branch: b2_12
Current Patch Set:
Commit: ccd89f519d4cdb3ebf971edf178f9cd2e664beec

[Gerrit Updater]

Minh Diep ([e-mail]) uploaded a new patch: https://review.whamcloud.com/36878
Subject: LU-12691 ldlm: obd_max_recoverable_clients is not atomic
Project: fs/lustre-release
Branch: b2_12
Current Patch Set: 1
Commit: fe66543cb7e0abdd0daaf8c4be0fccf41f299787

[Peter Jones]

Landed for 2.13

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/35914/
Subject: LU-12691 ldlm: obd_max_recoverable_clients is not atomic
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 01261e7b563adc97899d962f0ba2d1b430894bf7

[Tatsushi Takamura]

We hit this LBUG when running insanity test4. (According to the crash dump, LU-11001 seems to be the same issue)

 

00010000:00040000:0.0:1520239440.432805:0:10096:0:(ldlm_lib.c:2206:check_for_recovery_ready()) ASSERTION( clnts <= obd->obd_max_recoverable_clients ) failed:
00010000:00040000:0.0:1520239440.432814:0:10096:0:(ldlm_lib.c:2206:check_for_recovery_ready()) LBUG

After applying this patch, this LBUG have not been reproduced.

 

 

[Gerrit Updater]

Tatsushi Takamura ([e-mail]) uploaded a new patch: https://review.whamcloud.com/35914
Subject: LU-12691 ldlm: obd_max_recoverable_clients is not atomic
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 8772d09ae7ed23f8d29a8f3404c42ae095c53431