topic Re: cmlvmd: Could not read messages from /usr/lbin/cmcld in Operating System - HP-UX

cmlvmd: Could not read messages from /usr/lbin/cmcld

Hans Hendriks_1 — Tue, 02 Apr 2002 07:40:30 GMT

Hello

In my syslog.log I got the next message:

cmlvmd: Could not read messages from /usr/lbin/cmcld: Software caused connection abort.

cmlvmd: CLVMD exiting
cmsrvassistd[1496]: Lost connection to the cluster daemon
cmsrvassistd[1496]: Lost connection with ServiceGuard cluster daemon (cmcld): Software caused connection abort
cmtaped[1501]: Lost connection to the cluster daemon.
cmtaped[1501]: cmtaped terminating

The machine is in a two node cluster and this machine reboots too often. the last weekend 2 times.

I hope someone can help me.

Thanks.

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

Ceesjan van Hattum — Tue, 02 Apr 2002 08:07:40 GMT

Looks to me like a result of incorrectly switched IPadress(es) to a primary lan for a different subnet as the result of a local failover. In case it is, a patch can be used: PHSS_17581.
It is a non-critical an no reboot is needed. Can you check if you have this one installed (swlist) ?

Regards,
Ceesjan

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

Clemens van Everdingen — Tue, 02 Apr 2002 08:08:11 GMT

Hans,

Looks like a problem with the machine.
Is it generating a dump in /var/adm/crash ?
If so is ther a file in /var/tombstones called ts99.

I would suggest if you have a software contract to log a call to the response center.

Might be some software causing the reboots, or maybe a hardware issue.

C.

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

U.SivaKumar_2 — Tue, 02 Apr 2002 08:10:18 GMT

hi,
what setup do you use for heartbeat monitoring
serial or lan ?.
check the lan connection between the two nodes
with continous ping from one node to another
for a period of i hour and see any timeouts are
there. any abnormal syslog messages on the other node ?.
regards,
U.SivaKumar

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

melvyn burnard — Tue, 02 Apr 2002 08:20:44 GMT

Couple of questions here:

1) what OS version?
2) what SG version?
3) what SG patch?
4) are you perhaps using sar?
5) has anything changed to start this behaviour?

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

Hans Hendriks_1 — Tue, 02 Apr 2002 09:44:26 GMT

the os version is HPUX 11.00
the sg version is 11.12
what sg patch. We have the patch bundel from december 2001 on the machine. I do not know whitch patch is a ms/sg patch. Maybe you have a suggestion.

I think their is nothing changed since this behavior has started.

Thanks and hope to hear from you again.

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

Ceesjan van Hattum — Wed, 03 Apr 2002 21:05:30 GMT

Hi Hans,
Here i am again..
PHSS_17581 is from march 1999,
so your bundle of dec 2001 makes my suggestion obsolete.
But did you find something in /var/tombstones (see Clemens) or at /var/adm/crash?
And do you see ping-timeouts in the heartbeat? (see our new member SivaKumar..).
We see no assigned points, so we do not know how valuable our comments are...
Please do not make the mistake to install a newer patchlevel (like june 2001) without thorough research.
Hope to hear from you again.

Re: cmlvmd: Could not read messages from /usr/lbin/cmcld

melvyn burnard — Wed, 03 Apr 2002 21:20:43 GMT

The latest patch for Sg 11.12 is PHSS_26270.
Be aware, Sg patches are NEVER included in a standard patch bundle, so you will not have installed one with the bundle.
to check, do:
what /usr/lbin/cmcld | grep PHSS

see what you get back.

Are you using sar(1m) at all??
If so, there is a known issue with sar causing these symptoms, you need to patch sar with the latest sar patch PHCO_25174