Re: Why resources_used.cput=00:00:00 ???

On Tue, 2007-05-15 at 21:24 +0200, Filippo Spiga wrote:
>    one examples from accounting logs...
> 
> 05/11/2007 15:41:59;E;1745.scilx.disco.unimib.it;user=bernasconi
> group=mater jobname=cp2k_on_dolphin queue=sci_queue ctime=11
> 78869212 qtime=1178869212 etime=1178869212 start=1178869280
> exec_host=node20/1+node20/0+node07/1+node07/0+node15/1+node15/0
> +node06/1+node06/0+node14/1+node14/0+node05/1+node05/0+node08/1
> +node08/0+node18/1+node18/0 Resource_List.neednodes=8:ppn=2:sci
> Resource_List.nodect=8 Resource_List.nodes=8:ppn=2:sci
> Resource_List.walltime=06:00:00 session=8354 end=1178890919
> Exit_status=271 resources_used.cput=00:00:00
> resources_used.mem=20056kb resources_used.vmem=52356kb
> resources_used.walltime=06:00:39 
> 
> Why resources_used.cput=00:00:00 ?

Was this job started by a TM-aware parallel startup mechanism such as
OSC's mpiexec or the built-in launchers in LAM and OpenMPI, or using so
rsh/ssh-based method?  TORQUE can only do CPU time accounting for
processes that are children of pbs_mom; anything started using a
rsh/ssh-based scheme isn't going to show up in the PBS accounting logs.

Also, I notice that the job requested walltime=6:00:00 and died after 6
hours plus epsilon.  Are you sure this job wasn't terminated for
exceeding its walltime limit?

        --Troy
-- 
Troy Baer                       troy@xxxxxxx
Science & Technology Support    http://www.osc.edu/hpc/
Ohio Supercomputer Center       614-292-9701