[LU-2054] Test failure on test suite sanity, subtest test_76

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.3.0
• Fix Version/s: Lustre 2.3.0
• Type: Bug
• Priority: Minor
• Reporter: Maloo
• Assignee: WC Triage
• Resolution: Fixed

Environment

Server: b2_3-RC1 RHEL6
Client: b2_3-RC1 FC15

• Severity: 3

Description

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

This issue relates to the following test suite run: https://maloo.whamcloud.com/test_sets/22afcdae-0a7d-11e2-9fff-52540035b04c.

The sub-test test_76 failed with the following error:

test_76 returned 1

== sanity test 76: confirm clients recycle inodes properly ====== 14:22:45 (1348953765)
/usr/lib64/lustre/tests/sanity.sh: line 4590: echo: write error: Input/output error
before inodes: 104
after inodes: 104
/usr/lib64/lustre/tests/sanity.sh: line 4590: echo: write error: Input/output error
 sanity test_76: @@@@@@ FAIL: test_76 failed with 1 
  Trace dump:
  = /usr/lib64/lustre/tests/test-framework.sh:3643:error_noexit()
  = /usr/lib64/lustre/tests/test-framework.sh:3665:error()
  = /usr/lib64/lustre/tests/test-framework.sh:3907:run_one()
  = /usr/lib64/lustre/tests/test-framework.sh:3937:run_one_logged()
  = /usr/lib64/lustre/tests/test-framework.sh:3759:run_test()

Comments

[Jian Yu]

Lustre Tag: v2_3_0_RC2
Lustre Build: http://build.whamcloud.com/job/lustre-b2_3/32
Distro/Arch: RHEL6.3/x86_64(server), FC15/x86_64(client)
Network: TCP
ENABLE_QUOTA=yes

The same issue occurred: https://maloo.whamcloud.com/test_sets/749f719a-15cd-11e2-9c65-52540035b04c

[Jian Yu]

Lustre Tag: v2_3_0_RC3
Lustre Build: http://build.whamcloud.com/job/lustre-b2_3/36
Distro/Arch: RHEL6.3/x86_64(server), FC15/x86_64(client)
Network: TCP
ENABLE_QUOTA=yes

The same issue occurred: https://maloo.whamcloud.com/test_sets/3eebf870-1609-11e2-9c65-52540035b04c

[Oleg Drokin]

The problem here is that writes to /proc/slabinfo are only supported if you have SLAB allocator as default.
If you have SLUB allocator configured, writes to /proc/slabinfo are not permitted.

Since this write attempt is the last at the end of test_76, the bad status becomes that of the test.
We probably should just stick return 0 at the end as the workaround.

Whenever the test is robust enough without slab parameters changing is a different question that should be separately investigated too.

[Andreas Dilger]

This bug is no longer being hit.