[LU-5684] conf-sanity test_56: 'Stripe count not two: 1'

• Status: Resolved
• Project: Lustre
• Fix Version/s: Lustre 2.7.0
• Type: Bug
• Priority: Critical
• Reporter: Maloo
• Assignee: Li Wei (Inactive)
• Resolution: Fixed
• Severity: 3

Description

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

This issue relates to the following test suite run: https://testing.hpdd.intel.com/test_sets/0d86a994-4878-11e4-b83b-5254006e85c2.

The sub-test test_56 failed with the following error:

Stripe count not two: 1

Please provide additional information about the failure here.

Info required for matching: conf-sanity 56

Comments

[John Hammond]

The stripe count check was introduced by:

commit b5485d307568af92e1a940fa4a7859e6db5b7a97
Author: Li Wei <[e-mail]>
Date:   Wed Sep 24 12:12:37 2014 +0800

    LU-5654 osd-ldiskfs: Handle holes in osd_ldiskfs_read()

    Current osd_ldiskfs_read() incorrectly returns zero and leaves the
    corresponding portion of the buffer untouched when a block to be read
    is not allocated.

    Change-Id: Idfd441656b99aa039a6bb4f7141b5407553855da
    Signed-off-by: Li Wei <[e-mail]>
    Reviewed-on: http://review.whamcloud.com/12035
    Tested-by: Jenkins
    Tested-by: Maloo <[e-mail]>
    Reviewed-by: Liang Zhen <[e-mail]>
    Reviewed-by: Johann Lombardi <[e-mail]>
    Reviewed-by: Oleg Drokin <[e-mail]>

Li Wei can you comment?

[Jodi Levi (Inactive)]

Di,
Li Wei is out this week. Would you be able to have a look and give an initial assessment of this?
Thank you!

[nasf (Inactive)]

another failure instance:
https://testing.hpdd.intel.com/test_sets/9c7219da-48f1-11e4-be93-5254006e85c2

[Li Wei (Inactive)]

Could it be that the setstripe I added is racy against the MDT's discovery of all two OSTs?

[Li Wei (Inactive)]

I think that's indeed what happened. The MDS debug log has

00000100:00080000:0.0:1412036924.035595:0:5165:0:(import.c:899:ptlrpc_connect_interpret()) ffff88006d760800 lustre-OST2710_UUID: changing import state from CONNECTING to FULL

for OST2710, but does not have the same for OST03e8.

[Li Wei (Inactive)]

http://review.whamcloud.com/12145 is an updated version of the reverted 12035.

[Andreas Dilger]

The patch that introduced this failure was reverted from master, but any later patches that depend on it need to be rebased.

[Li Wei (Inactive)]

The updated version has re-landed to master.