cmclconfd: Failed to release volume group xxxxxx

[Roberto Gallis]

Hi all,
I've a 4 cluster node on hp-ux 11.00, SG A.11.09.
Last week we restarted 2 nodes running oracle.
Now, on a node I've this message on syslog.log
"cmclconfd: Failed to release volume group xxxxxx"
and cmclconfd now occupies one cpu entirely.
What can I do?
Regards
Roberto

[Carsten Krege · Accepted solution]

The cmclconfd message "failed to release VG" does usually have no effect. This is logged due to an incomsistency in the LVM code that cmclconfd uses (when trying to deattach a physical volume from a volume group) and the LVM lab is currently working on this. This happens when cmclconfd queries the VG config on the node (e.g. when doing cmquerycl, cmcheckconf or cmapplyconf) that does not have the VG activated.

I don't think that the high CPU usage of cmclconfd is related to this. You can safely kill the cmclconfd process when it is NOT the parent process of the cmcld process. Note: If you issued a SG command before that waits for this cmclconfd to return data (e.g. cmviewcl), this command will fail.

Carsten

[melvyn burnard]

The message you see in syslog is nothing more than an informational message when someone does something using SG commands that tries ot look at the VG's as if they were exclusive.

As for the cmclconfigd using 1 cpu, have uyou patched SG 11.09 with the latest patch?
PHSS_24536 is the latest.
I vaguely recall ther ebeing a fix for something like this.

[Roberto Gallis]

Hi,
... after some minutes the percentage of cpu went down... maybe after cmcheckconf done by a collegue...
I will install the patch mentioned

Best regards

Roberto