[LU-1100] metabench failed on NFSv3/v4 over Lustre

• Type: Bug
• Resolution: Duplicate
• Priority: Minor
• Fix Version/s: Lustre 2.1.2, Lustre 1.8.8
• Affects Version/s: Lustre 2.1.1

Environment

Lustre Tag: v2_1_1_0_RC2
Lustre Build: http://build.whamcloud.com/job/lustre-b2_1/41/
Distro/Arch: RHEL6/x86_64 (kernel version: 2.6.32-220.el6)
Network: IB (in-kernel OFED)
ENABLE_QUOTA=yes

• Severity: 3
• Assignee: WC Triage
• Reporter: Jian Yu

Description

Lustre Client/NFS Server: fat-intel-2-ib
NFS Clients: client-1-ib,client-4-ib

The metabench test failed on NFSv3/v4 clients as follows:

[02/13/2012 22:19:44] Entering par_create_multidir to create 4343 files in 1 dirs
Removed 1095 files in     25.466 seconds
[02/13/2012 22:20:09] FATAL error on process 7
Proc 7: Cant remove directory [/mnt/lustre/d0.metabench/TIME_CREATE_007.000]: Directory not empty

Dmesg on NFS client client-4-ib showed that:

NFS: directory d0.metabench/TIME_CREATE_007.000 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 304394735309107493
NFS: directory d0.metabench/TIME_CREATE_007.001 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 309878210750602462
NFS: directory d0.metabench/TIME_CREATE_007.000 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 73053889
NFS: directory d0.metabench/TIME_CREATE_007.000 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 73053889
NFS: directory d0.metabench/TIME_CREATE_007.000 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 76752781
NFS: directory d0.metabench/TIME_CREATE_007.000 contains a readdir loop.  Please contact your server vendor.  Offending cookie: 76752781

Maloo report: https://maloo.whamcloud.com/test_sets/3586a652-56d5-11e1-8ba4-5254004bbbd3

The issue is related to CentOS 6.2 with kernel 2.6.32-220.el6:
https://www.centos.org/modules/newbb/viewtopic.php?topic_id=35387&forum=55&post_id=152924#forumpost152924

Issue Links

• duplicates: LU-1191 Test failure on test suite parallel-scale-nfsv3, subtest test_metabench.
• Trackbacks: Lustre 2.1.1 release testing tracker Lustre 2.1.1 RC2 Tag: v2110RC2 Build:
• Trackbacks: Changelog 2.1 Changes from version 2.1.1 to version 2.1.2 Server support for kernels: 2.6.18308.4.1.el5 (RHEL5) 2.6.32220.17.1.el6 (RHEL6) Client support for unpatched kernels: 2.6.18308.4.1.el5 (RHEL5) 2.6.32220.17.1....

Comments

[Li Wei (Inactive)]

Yu Jian, thanks for the info.

[Jian Yu]

This is a duplicate of LU-1191, which is fixed by the kernel update tracked under LU-1424 and landed for 1.8.8, 2.1.2 and 2.3.

[Li Wei (Inactive)]

https://maloo.whamcloud.com/test_sets/fbe67f94-a599-11e1-8432-52540035b04c

[Sarah Liu]

2.1.55 NFS v3/v4 also has this issue.

Lustre: DEBUG MARKER: == parallel-scale test metabench: metabench == 17:26:11 (1329355571)
NFS: directory d0.metabench/TIME_CREATE_003.000 contains a readdir loop. Please contact your server vendor. Offending cookie: 233958805
NFS: directory d0.metabench/TIME_CREATE_003.000 contains a readdir loop. Please contact your server vendor. Offending cookie: 233958805
Lustre: DEBUG MARKER: parallel-scale test_metabench: @@@@@@ FAIL: metabench failed! 1

[Peter Jones]

As this is an NFS issue rather than a Lustre issue, dropping as a blocker

[Peter Jones]

Lai

Could you please look into this one?

Thanks

Peter