[LU-8586] pios_ssf returning ENOSPC due to mixed OST size.

• Type: Bug
• Resolution: Fixed
• Priority: Major
• Fix Version/s: Lustre 2.10.0
• Affects Version/s: Lustre 2.5.1
• Labels: patch
• Severity: 3
• Assignee: James Nunez (Inactive)
• Reporter: Arshad Hussain

Description

test(pios_ssf) failing intermittently is that the two OST’s are setup with different OST size by TB. The size difference is significant (~200MB). This leads to smaller OST filling up much faster during the pios run. Therefore, any subsequently write to FS now returns and ENOSPC.

Fail case.

/dev/vdc                                     688976      34196    618444   6% /mnt/ost1
/dev/vdd                                     468680      30108    412640   7% /mnt/ost2

Pass case. OST with identical space definition.

/dev/vdc                                     688976      34196    618444   6% /mnt/ost1
/dev/vdd                                     688976      34196    618444   6% /mnt/ost2

Comments

[Peter Jones]

Landed for 2.10

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/22343/
Subject: LU-8586 test: Fix failure due to mixed OST size.
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 6e6a3ac4ef3231463e05935ee68a1508b3a5d8d4

[Arshad Hussain]

Update:

For Fail case the error seen is ...

stdout.log
Run	Test	Tstamp		T	N	S	C	Aggregate	Lowest	Highest	Run_time
-----------------------------------------------------------------------------------------------------------
1	Write	1452330509	1	37	8MB	1MB	205MB/s		0.036s	0.041s	1.442s
2	Write	1452330510	8	37	8MB	1MB	191MB/s		0.201s	0.410s	1.545s
Error: IO error while doing pwrite64, aborting.
 sanity-benchmark test_pios_ssf: @@@@@@ FAIL: test_pios_ssf failed with 24 
  Trace dump:
  = /usr/lib64/lustre/tests/test-framework.sh:4672:error()
  = /usr/lib64/lustre/tests/test-framework.sh:4932:run_one()
  = /usr/lib64/lustre/tests/test-framework.sh:4968:run_one_logged()
  = /usr/lib64/lustre/tests/test-framework.sh:4774:run_test()
  = /usr/lib64/lustre/tests/sanity-benchmark.sh:324:main()
Dumping lctl log to /tmp/test_logs/1451120908/sanity-benchmark.test_pios_ssf.*.1451120914.log
Resetting fail_loc and fail_val on all nodes...done.
FAIL pios_ssf (7s)

OST setup is

/dev/vdc                                     688976      34196    618444   6% /mnt/ost1
/dev/vdd                                     468680      30108    412640   7% /mnt/ost2

[Gerrit Updater]

Arshad Hussain ([e-mail]) uploaded a new patch: http://review.whamcloud.com/22343
Subject: LU-8586 test: Fix failure due to mixed OST size.
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 2b4310be272eb6dba3dbfd2a2aeaf5e3986a6ac5