[LU-4654] sanity test_129: current dir size 12288, previous limit 12288

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.6.0, Lustre 2.5.1
• Type: Bug
• Priority: Major
• Reporter: Maloo
• Assignee: Emoly Liu
• Resolution: Duplicate
• Labels: dne
• Severity: 3

Description

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

This looks a lot like LU-3909, already fixed. Not sure if this is a new bug or that same old bug popping up again. Not seen during version interop, as reported in LU-3909.

Seems like a high runner blocking tests. maloo reports:
Failure Rate: 19.00% of last 100 executions [all branches]

This issue relates to the following test suite run: http://maloo.whamcloud.com/test_sets/cf2c124c-9a26-11e3-baa9-52540035b04c.

The sub-test test_129 failed with the following error:

current dir size 12288, previous limit 12288

Info required for matching: sanity 129

Issue Links

• is related to: LU-2479 sanity.sh test_129: max dir size limi...

Comments

[Oleg Drokin]

Hm.
I noticed taht the patch is to rhel6.5 ,I wonder if in rhel6.5 something have changed again in this dir growth area?

[Bob Glossman (Inactive)]

another, in b2_5:
https://maloo.whamcloud.com/test_sets/c2381ac8-a4ac-11e3-a92a-52540035b04c

frequency is getting worse. maloo now says:
Failure Rate: 26.00% of last 100 executions [all branches]

Maybe this needs a raise in priority.

[Andreas Dilger]

I don't think the test is correct:

        if [ $(lustre_version_code $SINGLEMDS) -lt $(version_code 2.4.51) ]; then
                [ $I -eq $MAX ] && return 0
        else
                [ $I -gt $MAX ] && return 0
        fi
        error_exit "current dir size $I, previous limit $MAX"

The check for older <= 2.4.51 MDSes seems OK - if the directory size is equal to the limit then the test has passed.

I don't understand the else clause, however. That returns success (0) if the directory grew larger than the limit, which I thought the patch was trying to fix.

There are also a few other places in the test that could be fixed. "while [ ! $I -gt $MAX ]" should instead be "while [ $I -lt $MAX ]",

[Peter Jones]

Emoly

Could you please take care of this issue?

Thanks

Peter

[Bob Glossman (Inactive)]

Not sure if this is significant or relevant, but have been running into this a lot in b2_5 test runs lately. Blocking what I think are good mods from landing. more examples:

https://maloo.whamcloud.com/test_sets/03d5618e-a51b-11e3-9fee-52540035b04c
https://maloo.whamcloud.com/test_sets/dde0d8ba-a514-11e3-9e53-52540035b04c

[Jian Yu]

I found that the test failed on Lustre b2_5 branch with MDSCOUNT=4, but passed with MDSCOUNT=2.

MDSCOUNT=4:
https://maloo.whamcloud.com/test_sets/10b85614-a6b4-11e3-b6b0-52540035b04c

MDSCOUNT=2:
https://maloo.whamcloud.com/sub_tests/d9ac4fda-a6b4-11e3-9d0d-52540035b04c

[James Nunez (Inactive)]

Another instance with Lustre b2_5 and 4 MDTs: https://maloo.whamcloud.com/test_sessions/8045b1ee-b9ff-11e3-805d-52540035b04c

[Bob Glossman (Inactive)]

another in b2_5:
https://maloo.whamcloud.com/test_sets/667cdba2-c044-11e3-8c28-52540035b04c

[James Nunez (Inactive)]

Another failure with b2_5:
https://maloo.whamcloud.com/test_sets/e78c4e40-c048-11e3-8176-52540035b04c

[James Nunez (Inactive)]

Hit this again with b2_5 in review-dne-part-1
https://maloo.whamcloud.com/test_sets/bb581f36-abc4-11e3-a696-52540035b04c

[Bob Glossman (Inactive)]

another in b2_5
https://maloo.whamcloud.com/test_sessions/aa2bf6a6-c776-11e3-9a0e-52540035b04c

[James Nunez (Inactive)]

Another b2_5 in review-dne-part-1 at https://maloo.whamcloud.com/test_sets/3f63c60c-c765-11e3-9ec5-52540035b04c

[James Nunez (Inactive)]

Another b2_5 failure in review-dne-part-1 at https://maloo.whamcloud.com/test_sets/84240626-b9ff-11e3-805d-52540035b04c

[Emoly Liu]

Backport the patch http://review.whamcloud.com/8137 to b2_5 http://review.whamcloud.com/10043 to fix this problem.

[Jodi Levi (Inactive)]

Patches have landed to Master and b2_5.