[Mauiusers] Maui killing job for resource violation, can't find violation

Jim Kusznir jkusznir at gmail.com
Thu May 29 09:35:52 MDT 2008

That's the behavior we were seeing.  The nodes have 8 cores each, and
just for clarity, we specified :ppn=1, but that didn't change things
at all (still scheduled on 8 CPUs).

We haven't resolved this yet, but the user backed up and is rebuilding
his app...It appears the binary he had was generated with pgif90
instead of mpif90, and there were some other questionable compilation
options.  The last point I saw, the newly-compiled app was segfaulting
(single-thread or multi-thread).

With respect to the original problem, when he ran it single-threaded
(nodes=1:ppn=1), it would run.  It was only when nodes=>1 that he had
problems.  We didn't try nodes=1:ppn=8.


On Thu, May 29, 2008 at 12:38 AM, Jan Ploski <Jan.Ploski at offis.de> wrote:
> mauiusers-bounces at supercluster.org schrieb am 05/28/2008 09:57:31 PM:
>> I would agree on the cpu time, but also, how many processors are there
>> per node? When you say nodes=8, you get 8 nodes, not necessarily 8
>> cpus.
> With Maui it is the other way around... you get 8 cpus, not necessarily 8
> nodes (machines).
> Without Maui (just TORQUE) you would get 8 nodes, AFAIK.
> Regards,
> Jan Ploski
> _______________________________________________
> mauiusers mailing list
> mauiusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/mauiusers

More information about the mauiusers mailing list