[LU-7484] sanity test_225a and test_225b: mds-survey with zero-stripe failed

• Type: Bug
• Resolution: Fixed
• Priority: Minor
• Affects Version/s: Lustre 2.8.0
• Labels: SELinux
• Environment: 1 Client, 1 MDT, 2 OST (Master)
• Severity: 3
• Assignee: John Hammond
• Reporter: Saurabh Tandan (Inactive)

Description

sanity test_225a and test 225b failed for SElinux enabled client feature with following error:
Test log for 225a:

== sanity test 225a: Metadata survey sanity with zero-stripe == 09:17:00 (1448385420)
+ file_count=1000 thrhi=4 dir_count=2 layer=mdd stripe_count=0 rslt_loc=/tmp targets="eagle-52vm2:lustre-MDT0000" /usr/bin/mds-survey
Write failed: Broken pipe
Write failed: Broken pipe
Permission denied, please try again.
Permission denied, please try again.
Received disconnect from UNKNOWN: 2: Too many authentication failures for root
OST lustre-MDT0000 not setup
 sanity test_225a: @@@@@@ FAIL: mds-survey with zero-stripe failed 

Test log for 225b:

== sanity test 225b: Metadata survey sanity with stripe_count = 1 == 09:47:07 (1448387227)
+ file_count=1000 thrhi=4 dir_count=2 layer=mdd stripe_count=1 rslt_loc=/tmp targets="eagle-52vm2:lustre-MDT0000" /usr/bin/mds-survey
Permission denied, please try again.
Permission denied, please try again.
Received disconnect from UNKNOWN: 2: Too many authentication failures for root
Need obdfilter to test stripe_count
 sanity test_225b: @@@@@@ FAIL: mds-survey with stripe_count failed 

Comments

[John Hammond]

Need to do ssh-keygen and ssh-copy-id or set DSH=rsh then it works.

[John Hammond]

Can you setup your test nodes and ping me?

[Saurabh Tandan (Inactive)]

John, Yes I did.

[John Hammond]

Saurabh, did you try setting DSH=rsh in local.sh?

[Peter Jones]

John

Could you please advise on this one?

Thanks

Peter

[Saurabh Tandan (Inactive)]

/var/log/messages and /var/log/secure files are attached above.

[Oleg Drokin]

We probably need /var/log/messages and possibly /var/log/secure from this run to see what went wrong.
So far it looks like authentication failed either due to missing selinux label on .ssh or because of rlogin service not starting as in some another recent ticket