SUMMARY: CAAD forks defuncts under heavy load

From: Juergen Wolf <juergen.wolf_at_idmt.fraunhofer.de>
Date: Fri, 06 Aug 2004 08:57:19 +0200

Thanks for all the help.

Kevin Raubenolt pointed out, that this is a bug in the Tru64 Cluster
System. The HP Engineering was able to work out a Patch for me. After
patching the cluster all seems to be fine for now.

Regards,

  Juergen


--On Tuesday, May 25, 2004 03:30:11 PM +0200 Juergen Wolf
<wolf_at_idmt.fhg.de> wrote:

> Hi all,
>
> currently we are experiencing problems with the caa daemon process on
> our Tru64 5.1b Rev 2650 PK3 cluster. When one node is coming under heavy
> load during nightly incremental backup runs of some file systems (around
> 7TB used space); the caa daemon process on this node starts to produce
> lots of <defunct> processes. At the same time the EVM sends out a warning
> mail ("EVM ALERT [600]: EVM daemon: High event activity - exceeds 500 in
> 10 minutes"). After doing "/sbin/init.d/clu_caa stop" a new caa daemon
> starts and all the <defunct> processes are gone.
>
> Does anybody know how to prevent the caa from forking so many processes
> or even better from not defuncting them ?
>
> Regards,
>
> Juergen
>

-- 
email: Juergen.Wolf_at_idmt.fraunhofer.de
gilb:  Fraunhofer-Institut fuer Digitale Medientechnologie IDMT
       98693 Ilmenau, Langewiesener Straße 22
Tel.:  +49 3677 467-234				Fax:   +49 3677 467-467

Received on Fri Aug 06 2004 - 06:59:15 NZST