[LU-11226] sanity-flr test 200 fails with 'checksum error for mirror 3'

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Fix Version/s: Lustre 2.12.0
• Affects Version/s: Lustre 2.12.0
• Labels: FLR2
• Severity: 3
• Assignee: Zhenyu Xu
• Reporter: James Nunez (Inactive)

Description

sanity-flr test_200 started failing on August 6, 2018. The last lines seen in the test_log is

/mnt/lustre/f200.sanity-flr: found 6 stale components
/mnt/lustre/f200.sanity-flr: resyncing mirror: 1, components: 65537 65538 65539 65540 
components synced: 65537 65538 65539 65540 
/mnt/lustre/f200.sanity-flr: resyncing mirror: 3, components: 196609 196610 
components synced: 196609 196610 
3
 sanity-flr test_200: @@@@@@ FAIL: checksum error for mirror 3

There’s nothing interesting in the node console logs for this test.

Looking at all the patches that landed around August 6, https://review.whamcloud.com/32808 looks like a possible candidate for a change in the mirror resync behavior.

Here are links to a few recent failures:
https://testing.whamcloud.com/test_sets/71ef290a-9a2d-11e8-8ee3-52540065bddc
https://testing.whamcloud.com/test_sets/5ee037ee-99d0-11e8-87f3-52540065bddc
https://testing.whamcloud.com/test_sets/4cb6a618-99bf-11e8-a9f7-52540065bddc

Issue Links

• is related to: LU-12022 sanity-flr: test_200 'checksum error for mirror 3'
• is related to: LU-10916 Improve lfs mirror resync performance