John, good day.
Thu, Jul 31, 2008 at 07:52:03AM +0100, Gordon, JC (John) wrote:
> > -----Original Message-----
> > From: Sergio Maffioletti [mailto:[log in to unmask]]
> > Sent: 30 July 2008 18:25
> > To: [log in to unmask]
> > Subject: Maui error
> >
> > since upgrading our lcg_CE to release 3.1.16-0
> >
> > we've been observing malfunctioning of torque + maui maui
> > register occasioanl errors ( extracted form the logs )
> > ERROR: cannot get node info: Premature end of message
> >
> >
> > showq command ran constantly into timeouts.
> > The maui service could not be shut down regularly.
> >
> > we have already applied the solution proposed in
> >
> > http://www.clusterresources.com/pipermail/torqueusers/2007-May
> /005613.html
> > but does not help
> >
> > does anyone registered a similar anomaly ?
May be you're observing the bug described in
http://www.clusterresources.com/pipermail/torquedev/2008-June/001111.html
Try to set 'RMCFG[servername] TYPE=PBS TIMEOUT=30' in the
maui.cfg to make larger timeout. If this will help then you seem
to face the mentioned bug. I had no time yet to return to this
bug, but I'll try to beat it to the end in the way that will be
acceptable to the Torque/MAUI developers.
--
Eygene Ryabinkin, Russian Research Centre "Kurchatov Institute"
|