[LU-331] Test failure on test suite parallel-scale Created: 16/May/11  Updated: 24/Jun/11  Resolved: 24/Jun/11

Status: Resolved
Project: Lustre
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Minor
Reporter: Maloo Assignee: Hongchao Zhang
Resolution: Fixed Votes: 0
Labels: None

Attachments: Text File client-15-dmesg.log     Text File client-15-trace.log    
Severity: 3
Rank (Obsolete): 5190

 Description   

This issue was created by maloo for sarah <sarah@whamcloud.com>

This issue relates to the following test suite run: https://maloo.whamcloud.com/test_sets/957c1d18-8006-11e0-b5bf-52540025f9af.



 Comments   
Comment by Peter Jones [ 18/May/11 ]

HongChao

Could you please look into this test failure?

Thanks

Peter

Comment by Peter Jones [ 26/May/11 ]

HongChao

Could you please provide a status update on this issue?

Thanks

Peter

Comment by Hongchao Zhang [ 26/May/11 ]

the problem is in the test1 of connectathon of parallel-scale.sh, and there is a Segmentation fault in this test.
but it seems to be not related to Lustre for it occurs on the NFS4 client, and there are something confused in the logs of the test,
the nodes in this test is fat-intel-1(MDT), fat-intel-2(OST), Client(client-18), but the NFS4 server is at client-5 (10.10.4.5)

.........................
----============= acceptance-small: parallel-scale ============---- Mon May 16 14:50:51 PDT 2011
Loading modules from /usr/lib64/lustre
debug=0x33f0404
subsystem_debug=0xffb7e3ff
../lnet/lnet/lnet options: 'accept=all networks="o2ib0(ib0),tcp0(eth0)" accept_port=7988'
gss/krb5 is not supported
only running test connectathon
excepting tests: parallel_grouplock
NFSCLIENT mode: setup, cleanup, check config skipped
client-18.lab.whamcloud.com
10.10.4.5:/ /mnt/lustre nfs4 rw,vers=4,rsize=32768,wsize=32768,hard,intr,proto=tcp,timeo=600,retrans=3,sec=sys,addr=10.10.4.5 0 0
client-5.lab.whamcloud.com
Filesystem Type 1K-blocks Used Available Use% Mounted on
/dev/sda1 ext3 20315812 1312820 17954352 7% /
...

== parallel-scale test connectathon: connectathon ==================================================== 14:50:53 (1305582653)
OPTIONS:
cnt_DIR=/opt/connectathon
cnt_NRUN=2
/mnt/lustre/d0.connectathon: nfs4
tests: -b -g -s
./runtests -N 2 -b -f /mnt/lustre/d0.connectathon
... Pass 1 ...

Starting BASIC tests: test directory /mnt/lustre/d0.connectathon (arg: -f)

./test1: File and directory creation test
rm: cannot chdir from `.' to `/mnt/lustre/d0.connectathon': No such file or directory
./test1: (/opt/connectathon/basic) runtests: line 28: 5211 Segmentation fault ./test1 $TESTARG
basic tests failed
parallel-scale test_connectathon: @@@@@@ FAIL: connectathon failed: 1
.........................

Hi Sarah,
Is there a Lustre client mounted on client-5 and export the Lustre directory via NFS4?

Comment by Sarah Liu [ 26/May/11 ]

Hi Sarah,
Is there a Lustre client mounted on client-5 and export the Lustre directory via NFS4

yes. since I ran those tests on client-18 which is nfs client, I thought that was the reason client-15 was not listed in the node list

Comment by Hongchao Zhang [ 26/May/11 ]

the logs in client-5 (the nfs server exporting Lustre client) are needed for this issue, for the logs in client-18 can't give any trace except
the "Segmentation fault". Furthermore, it's only related to test tools of NFS4, not Lustre.

Comment by Sarah Liu [ 27/May/11 ]

please find the attached for logs from NFS server(lustre client)

Comment by Peter Jones [ 13/Jun/11 ]

Sarah

Oleg reports that all his NFS tests now pass cleanly. Could you please confirm whether this issue still exists after build 63 becomes available next week?

Thanks

Peter

Comment by Sarah Liu [ 13/Jun/11 ]

sure, will verify all nfs tests after the new build is available.

Comment by Peter Jones [ 24/Jun/11 ]

This issue did not occur during testing for build 63

Generated at Sat Feb 10 01:05:58 UTC 2024 using Jira 9.4.14#940014-sha1:734e6822bbf0d45eff9af51f82432957f73aa32c.