[LU-9054] sanity test_312: FAIL: blksz error: , expected: 4096

• Type: Bug
• Resolution: Unresolved
• Priority: Minor
• Affects Version/s: Lustre 2.10.0
• Labels: always_except zfs
• Environment: FSTYPE=zfs
• Severity: 3
• Assignee: Jinshan Xiong (Inactive)
• Reporter: Jian Yu

Description

sanity test 312 failed as follows in review-zfs-part-1 test session on master branch:

== sanity test 312: make sure ZFS adjusts its block size by write pattern ============================ 21:36:27 (1485380187)
CMD: trevis-36vm4 zfs get -p recordsize lustre-ost1/ost1
CMD: trevis-36vm4 zdb -e -p /dev/lvm-Role_OSS -dddd lustre-ost1/ost1
CMD: trevis-36vm4 zdb -e -p /dev/lvm-Role_OSS -dddd lustre-ost1/ost1 149 172 227 259 291 323 355 387 419 451 483 515 547 580 594 630 662 694 709 741 764 774 838 870 902 934 966 979 998 1027 1059 1091 1123 1155 1187 1219 1251 1291 1407 1439 1471 1503 4128 8482 12131 12164 12196 12228 12260
trevis-36vm4: zdb: dmu_bonus_hold(149) failed, errno 2
1+0 records in
1+0 records out
4096 bytes (4.1 kB) copied, 0.155851 s, 26.3 kB/s
CMD: trevis-36vm4 zdb -e -p /dev/lvm-Role_OSS -dddd lustre-ost1/ost1
/usr/lib64/lustre/tests/sanity.sh: line 15331: 16K5125121K51251216K1: value too great for base (error token is "16K5125121K51251216K1")
/usr/lib64/lustre/tests/sanity.sh: line 15364: [: -eq: unary operator expected
 sanity test_312: @@@@@@ FAIL: blksz error: , expected: 4096 

Maloo reports:
https://testing.hpdd.intel.com/test_sets/0bd3c2a6-e35f-11e6-981b-5254006e85c2
https://testing.hpdd.intel.com/test_sets/2434ad12-e312-11e6-981b-5254006e85c2
https://testing.hpdd.intel.com/test_sets/7c8df6e6-c7d6-11e6-9ae2-5254006e85c2
https://testing.hpdd.intel.com/test_sets/ec2e0fb8-9676-11e6-bc10-5254006e85c2

Issue Links

• is related to: LU-18311 interop: sanity test_312: FAIL: blksz error, actual 4096, expected: 2 * 1 * 4096
• is related to: LU-15963 sanityn test_56b: OSS OOM with ZFS
• is related to: LU-14692 deprecate use of OST FID SEQ 0 for MDT0000

Comments

[Andreas Dilger]

The ZDB+MMP issue looks like it was fixed in ZFS Issue #6464 and ZFS Issue #7797, which were included into ZFS 0.7.12, so that should no longer be the issue here. The test is failing with:

ustre/tests/sanity.sh: line 24502: 16K51251251251251216K16K16K16K16K16K16K16K16K16K16K16K16K16K51251216K1: value too great for base (error token is "16K51251251251251216K16K16K16K16K16K16K16K16K16K16K16K16K16K51251216K1")
/home/green/git/lustre-release/lustre/tests/sanity.sh: line 24531: [: -eq: unary operator expected
 sanity test_312: @@@@@@ FAIL: blksz error: , expected: 4096 

so something about the zdb output is generating too many block size numbers (16K, 512 repeated). I'll push a debug patch shortly.

[Andreas Dilger]

+3 on Gerrit Janitor: https://testing.whamcloud.com/gerrit-janitor/25873/results.html

[James Nunez (Inactive)]

I'm opening this ticket again so we can remove the test from the ALWAYS_EXCEPT list.

[Gerrit Updater]

John L. Hammond ([e-mail]) merged in patch https://review.whamcloud.com/28480/
Subject: LU-9054 tests: disable test_312 due to zdb issue
Project: fs/lustre-release
Branch: b2_10
Current Patch Set:
Commit: 29b89d5689ea0a018aa05a6ea78bc90e5229ccff

[James Nunez (Inactive)]

Reopening this ticket until we support a version of ZFS that fixes this issue and we can run santiy test 312.

[James Nunez (Inactive)]

Why did we resolved this ticket as fixed? The only patch that has landed disables sanity test 312, but the problem still remains. Shouldn't we keep this ticket open until we have a solution in place and can remove the test from the ALWAYS_EXCEPT list?

[Gerrit Updater]

Minh Diep ([e-mail]) uploaded a new patch: https://review.whamcloud.com/28480
Subject: LU-9054 tests: disable test_312 due to zdb issue
Project: fs/lustre-release
Branch: b2_10
Current Patch Set: 1
Commit: 63e5ef7d684616d4493933d1bcd710c24fcc447d

[Minh Diep]

Landed in 2.11

[Gerrit Updater]

Oleg Drokin ([e-mail]) merged in patch https://review.whamcloud.com/28343/
Subject: LU-9054 tests: disable test_312 due to zdb issue
Project: fs/lustre-release
Branch: master
Current Patch Set:
Commit: 05ebd8218108dccaa31c1dbc97cd4cc90dfd3f8d

[Jinshan Xiong (Inactive)]

I have filed a ticket at https://github.com/zfsonlinux/zfs/issues/6464 so that zfs upstream community can take a look.

[Gerrit Updater]

Jinshan Xiong ([e-mail]) uploaded a new patch: https://review.whamcloud.com/28343
Subject: LU-9054 tests: make zdb work with zpool dataset
Project: fs/lustre-release
Branch: master
Current Patch Set: 1
Commit: 2dcb45d93787f5e7309395163aa76105775a1049