[LU-7072] sanityn test_78: Expected set_param to return 0 or EAGAIN

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.8.0
• Affects Version/s: Lustre 2.8.0
• Severity: 3
• Assignee: Henri Doreau (Inactive)
• Reporter: Maloo

Description

This issue was created by maloo for Andreas Dilger <[e-mail]>

This issue relates to the following test suite run: https://testing.hpdd.intel.com/test_sets/7cec4ac6-4f04-11e5-bc53-5254006e85c2.

The sub-test test_78 failed with the following error:

CMD: shadow-21vm3 lctl set_param ost.OSS.ost_io.nrs_policies=orr
CMD: shadow-21vm3 lctl set_param ost.OSS.*.nrs_orr_quantum=1
shadow-21vm3: error: set_param: setting ost.OSS.ost_io.nrs_orr_quantum=1: No such device
ost.OSS.ost_io.nrs_orr_quantum=1
 sanityn test_78: @@@@@@ FAIL: Expected set_param to return 0 or EAGAIN

This may relate to the recently landed patch:
http://review.whamcloud.com/15104 "LU-6673 ptlrpc: Forbid too early NRS policy tunings"

This just started on Aug 30, and all patches failing so far are based on either:
LU-6667 llite: improve ll_getname
LU-6903 lov: call lov_object_find_cbdata() inside lock

Info required for matching: sanityn 78

Issue Links

• is related to: LU-7105 sanityn test_28 fails with 'error() without useful message, please fix'
• is related to: LU-7096 Unprotected critical section in nrs_policy_stop
• is related to: LU-7310 sanityn test_39a fails with 'mtime is not updated on write: 1444890978 <= 1413354977 <= 1444890979'
• is related to: LU-6673 NRS crash when applying tunings

Comments

[Andreas Dilger]

Patch landed to master for 2.8.0.

[Gerrit Updater]

Andreas Dilger ([e-mail]) merged in patch http://review.whamcloud.com/16178/
Subject: LU-7072 tests: Fix and re-enable test 78 in sanityn
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: b8d51c44f20c2a3decd4230b7ec656727ea813a1

[Henri Doreau (Inactive)]

I have updated the check to make it accept ENODEV as well, see: http://review.whamcloud.com/16178

Testing reveals that there are still critical race conditions when applying NRS tunings. This is out of the scope of this ticket though, I'll open separate ones.

[Henri Doreau (Inactive)]

I see what's going on (tuning applied too early). I need to evaluate how to fix it in the cleanest way and will try to propose a patch very soon.

[Oleg Drokin]

While I landed a patch disablign the test for now, we still need to fix it.

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch http://review.whamcloud.com/16164/
Subject: LU-7072 tests: disable sanityn test 78
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: fb8e349de2c11082b7e4de83146c0058e221fa37

[Gerrit Updater]

Oleg Drokin ([e-mail]) uploaded a new patch: http://review.whamcloud.com/16164
Subject: LU-7072 Disable sanityn test 78
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 83ef21f64b227003f4aa41f0d1abb6f5530e9acd

[Andreas Dilger]

Henri, it looks like test_78 that was recently added for http://review.whamcloud.com/15104 has started failing regularly. Could you please take a look the test failures:

https://testing.hpdd.intel.com/test_sets/7cec4ac6-4f04-11e5-bc53-5254006e85c2 (first failure)
https://testing.hpdd.intel.com/test_sets/2754282e-50ce-11e5-95a9-5254006e85c2 (most recent failure)