[LU-8482] auster run of sanity trying to use deprecated lctl conf_param in 2.8

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.8.0
• Type: Bug
• Priority: Major
• Reporter: John Salinas (Inactive)
• Assignee: James Nunez (Inactive)
• Resolution: Won't Fix
• Labels: LS_RZ
• Severity: 1

Description

1. export FSTYPE=zfs; ./auster -c cfg/multinode.sh -v -k sanity --only 0
   Started at Thu Aug 4 15:55:55 PDT 2016
   simics2: Checking config lustre mounted on /mnt/lustre
   Checking servers environments
   Checking clients simics2 environments
   Logging to local directory: /tmp/test_logs/2016-08-04/155555
   running: sanity ONLY=0
   run_suite sanity /usr/lib64/lustre/tests/sanity.sh
   ----============= acceptance-small: sanity ============---- Thu Aug 4 15:55:58 PDT 2016
   Running: bash /usr/lib64/lustre/tests/sanity.sh
   simics2: Checking config lustre mounted on /mnt/lustre
   Checking servers environments
   Checking clients simics2 environments
   Using TIMEOUT=100
   seting jobstats to procname_uid
   Setting lustre.sys.jobid_var from disable to procname_uid
   lustremgs: warning: 'lctl conf_param' is deprecated, use 'lctl set_param -P' instead
   lustremgs: error: conf_param: Invalid argument
   pdsh@simics2: lustremgs: ssh exited with exit code 22
   sanity : @@@@@@ FAIL: conf_param lustre.sys.jobid_var failed
   Trace dump:
   = /usr/lib64/lustre/tests/test-framework.sh:4670:error_noexit()
   = /usr/lib64/lustre/tests/test-framework.sh:4701:error()
   = /usr/lib64/lustre/tests/test-framework.sh:3776:set_conf_param_and_check()
   = /usr/lib64/lustre/tests/test-framework.sh:3799:init_param_vars()
   = /usr/lib64/lustre/tests/test-framework.sh:3988:check_and_setup_lustre()
   = /usr/lib64/lustre/tests/sanity.sh:108:main()
   Dumping lctl log to /tmp/test_logs/2016-08-04/155555/sanity..*.1470351363.log
   sanity returned 0
   Finished at Thu Aug 4 15:56:05 PDT 2016 in 10s
   ./auster: completed with rc 0

Fails (what it tries to run)
[root@simics2 tests]# /usr/bin/pdsh -S -Rssh -w lustremgs '(PATH=$PATH:/usr/lib64/lustre/utils:/usr/lib64/lustre/tests:/sbin:/usr/sbin; cd /usr/lib64/lustre/tests; LUSTRE="/usr/lib64/lustre" sh -c "/usr/sbin/lctl conf_param lustre.sys.jobid_var='\''procname_uid'\''")'
lustremgs: warning: 'lctl conf_param' is deprecated, use 'lctl set_param -P' instead
lustremgs: error: conf_param: Invalid argument
pdsh@simics2: lustremgs: ssh exited with exit code 22

Works — what it should run
[root@simics2 tests]# /usr/bin/pdsh -S -Rssh -w lustremgs '(PATH=$PATH:/usr/lib64/lustre/utils:/usr/lib64/lustre/tests:/sbin:/usr/sbin; cd /usr/lib64/lustre/tests; LUSTRE="/usr/lib64/lustre" sh -c "/usr/sbin/lctl set_param -P lustre.sys.jobid_var='\''procname_uid'\''")'

This is with the following RPMs on the client node where the run is started from (latest 2.8 build):
lustre-client-tests-2.8.0-3.10.0_327.3.1.el7.x86_64.x86_64.rpm
lustre-iokit-2.8.0-3.10.0_327.3.1.el7.x86_64.x86_64.rpm

Issue Links

• is related to: LU-7004 fix "lctl set_param -P" to allow depr...

Comments

[Oleg Drokin]

The eprecated message is no longer valid and we reinstated it.

Now the real problem is the parameter it's giving is somehow wrong and gets rejected.
So that's somethng that should be figured out for hte failure, I think?

Or if this just purely about the deprecated message - we can resolve this ticket or somebody might backport the patch to stop the complaints to 2.8 as well I guess.

[John Salinas (Inactive)]

No, this is not just about the message sanity.sh cannot be run.

It was pointed out to us that we should be using MDSZFSDEV instead of MGSZFSDEV let me try that and see if it makes a difference.

[Andreas Dilger]

There is no need to quote the "procname_uid" parameter name.

[Andreas Dilger]

The patch to remove the conf_param deprecation is http://review.whamcloud.com/20573 and could be backported.

[John Salinas (Inactive)]

We have a work around that we are using for now: lctl set_param -P jobid_var=procname_uid on mgs

[James Nunez (Inactive)]

The team that opened this ticket will be using a later version of Lustre than 2.8, they will use 2.11, and should not see this issue. Please reopen or create a new ticket if you experience this issue again.