[torqueusers] Help with Torque routing queue configuration

Chris Samuel csamuel at vpac.org
Sat Feb 23 20:56:39 MST 2008


----- "Mike P Moore" <Mike.Moore at appliedbiosystems.com> wrote:

> We've found that with the particular application we are running,
> we can swamp the any given fileserver with ~20 jobs accessing
> that filesystem. Once we go past that point, the wall time on the
> job greatly exceeds the cpu time of the job (indicating that the job
> is being throttled by disk access).

Are these RHEL derived file servers (inc. CentOS and Rocks),
or others ?

I ask because we've found that RHEL ext3 doesn't
cut the mustard as an NFS server and long ago switched
to Debian and XFS for our NFS server for our clusters.

Not to say that pathological I/O patterns can't cause
it problems (we had one on Friday for instance), but
the RHEL box couldn't cope with our normal load..

-- 
Christopher Samuel - (03) 9925 4751 - Systems Manager
 The Victorian Partnership for Advanced Computing
 P.O. Box 201, Carlton South, VIC 3053, Australia
VPAC is a not-for-profit Registered Research Agency


More information about the torqueusers mailing list