[LU-8972] conf-sanity test_101: File hasn't object on OST

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.12.0
• Affects Version/s: Lustre 2.9.0
• Severity: 3
• Assignee: Alex Zhuravlev
• Reporter: Maloo

Description

This issue relates to the following test suite run: https://testing.hpdd.intel.com/test_sets/57cc8712-c9db-11e6-af18-5254006e85c2.

The sub-test test_101 failed with the following error:

File hasn't object on OST

Info required for matching: conf-sanity 101

 

This is quite new issue which starts to happen since yesterday, it can be related to some recent commits.

Issue Links

• is related to: LU-9285 revert LU-8367 and LU-8972
• is related to: LU-8562 osp_precreate_cleanup_orphans/osp_precreate_reserve race may cause data loss

Comments

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/25079/
Subject: LU-8972 osp: skip subsequent orphan cleanups
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 6f56f71b407a8c14db4c2accd37da5b4feecde1a

[Joseph Gmitter (Inactive)]

Reassigning to Alex since he has a patch in flight for the issue.

[Gerrit Updater]

Alex Zhuravlev ([e-mail]) uploaded a new patch: https://review.whamcloud.com/25079
Subject: LU-8972 osp: skip subsequent orphan cleanups
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: c2c5dcb106efdf290a93e7c91dcf48a09f5512b6

[Joseph Gmitter (Inactive)]

Assigning to James to reproduce in manual mode and validate where the test is going wrong to identify next steps.

[Ned Bass (Inactive)]

I'm disappointed that https://review.whamcloud.com/24517/ was fast-tracked through the landing process with no explanation or discussion. The lack of transparency is a bit alarming given that the failing test is designed to catch filesystem data loss. To make matters worse, the test results are hidden behind a login screen so there is no public way to discern why you might have decided to disable the test.

[Bob Glossman (Inactive)]

another on master:
https://testing.hpdd.intel.com/test_sets/08853232-cc1d-11e6-bb5d-5254006e85c2

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/24517/
Subject: LU-8972 tests: Disable conf-sanity test 101
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: fa722ccf04fe6c386628ea5a575165c173643800

[Gerrit Updater]

Oleg Drokin ([e-mail]) uploaded a new patch: https://review.whamcloud.com/24517
Subject: LU-8972 tests: Disable conf-sanity test 101
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 12a9eee4334df25f1445a53e5b544de561c33a1b

[Mikhail Pershin]

failed test 101 was introduced in LU-8562

[Mikhail Pershin]

I see that this issue happens quite often right now in Maloo. Failure rate is about 50% right now