[torqueusers] curious qstat behaviour with Torque

[Chris Samuel]

On Tue, 24 Jul 2007, SCIPIONI Roberto wrote:

> I have a Linux Cluster with Xeon Intel CPUs and Intel compiler +
> LAM MPI
>
> It works fine but, I have noticed that for some codes when I use
> qstat I get a small Time Use while when I use
>
> qstat -u username
>
> it gives the right Elapsed times
>
> I know basically these are different times which are probably read
> in different places  but
>
> Why doesnt qstat update the time
>
> I am using parallel jobs but the cluster does NOT have a shared
> directory system

When you built LAM did you tell it to use Torque's TM API ?

If so, have you checked that your LAM jobs are using the TM API to 
boot and start ?

I believe that if you're not using that then it will fall back to 
ssh/rsh and then Torque cannot track the usage very well as it cannot 
track all the processes on the nodes correctly.

cheers,
Chris
-- 
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
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url : http://www.supercluster.org/pipermail/torqueusers/attachments/20070808/03b9a272/attachment.bin