[Moabusers] No effect when changing JOBNODEMATCHPOLICY

[Dave Jackson]

Jason,

  EXACTNODE was the correct setting.  Can you confirm that Moab is
properly parsing JOBNODEMATCHPOLICY using 'mschedctl -l | grep
JOBNODEMATCH'.  Also, are you using qsub or msub?  Finally, what version
of Moab are you running?

Thanks,
Dave

On Mon, 2006-10-30 at 10:27 -0600, Jason urizen wrote:
> The results are identical with EXACTPROC and EXACTNODE.  The allocated
> resources show up as [<node>:2] in both cases.
> 
> On 10/30/06, Chris Vaughan < chris at clusterresources.com> wrote:
>         Hi Jason,
>         
>         Try EXACTPROC instead of EXACTNODE.  Let me know if that
>         solves it for you.
>         
>         
>         Jason urizen wrote:
>         > I am trying to get Moab to interpret "-l nodes=2:ppn=1" as
>         two
>         > distinct nodes with one processor each, where nodes have 2 
>         > processors.  Setting JOBNODEMATCHPOLICY to EXACTNODE seems
>         to have no
>         > effect.  Jobs submitted with this resource string are going
>         to a
>         > single node regardless of this setting.
>         >
>         > I'm new to the list but found some talk on this subject in
>         the 
>         > archives.  Everything reads like this parameter should
>         invoke the
>         > desired behavior.  Am I misinterpreting the documentation or
>         is there
>         > more to be changed, perhaps in the torque configuration?
>         >
>         ------------------------------------------------------------------------
>         >
>         > _______________________________________________
>         > moabusers mailing list
>         > moabusers at supercluster.org
>         > http://www.supercluster.org/mailman/listinfo/moabusers
>         >
>         
>         
>         --
>         Chris Vaughan
>         Cluster Resources, Inc. 
>         Direct - UK Office:  +44 1223 437 132
>         US Headquarters:  +1 801 717 3700
>         www.clusterresources.co.uk
>         
> 
> _______________________________________________
> moabusers mailing list
> moabusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/moabusers