There is a workaround though. http://www.supercluster.org/pipermail/torqueusers/2012-January/013974.html
On 22 November 2012 11:49, Clotho Tsang <[email protected]> wrote: > Seems it's the bug of maui 3.3.1, which is not found at maui 3.2.6 > http://www.supercluster.org/pipermail/torqueusers/2012-January/013974.html > > > On 22 September 2012 03:54, Daniel Davidson <[email protected]> wrote: > >> I am working on finalizing our cluster setup, and as part of that is >> nailing down the torque/maui config. >> >> I have been looking at what happens in maui when someone submits qsub -l >> procs=x blah.sh to their script. Right now, it looks like maui is >> ignoring the procs line. Here is an example: >> >> bash-4.1$ qsub -I -q test_queue -l procs=6 >> qsub: waiting for job 76338.biocluster.igb.illinois.edu to start >> qsub: job 76338.biocluster.igb.illinois.edu ready >> >> -bash-4.1$ >> >> However, when i do a tracejob: >> >> [root@biocluster init.d]# tracejob -v 76338 >> /var/spool/torque/server_priv/accounting/20120921: Successfully located >> matching job records >> /var/spool/torque/server_logs/20120921: Successfully located matching >> job records >> /var/spool/torque/mom_logs/20120921: No such file or directory >> /var/spool/torque/sched_logs/20120921: No such file or directory >> >> Job: 76338.biocluster.igb.illinois.edu >> >> 09/21/2012 14:47:16 S enqueuing into test_queue, state 1 hop 1 >> 09/21/2012 14:47:16 S Job Queued at request of >> [email protected], owner = >> [email protected], job name = STDIN, queue = test_queue >> 09/21/2012 14:47:16 A queue=test_queue >> 09/21/2012 14:47:17 S Job Run at request of >> [email protected] >> 09/21/2012 14:47:17 S Not sending email: User does not want mail of >> this type. >> 09/21/2012 14:47:17 A user=danield group=danield jobname=STDIN >> queue=test_queue ctime=1348256836 qtime=1348256836 etime=1348256836 >> start=1348256837 [email protected] >> exec_host=compute-0-1/0 >> Resource_List.mem=3gb Resource_List.ncpus=1 >> Resource_List.neednodes=1 Resource_List.nodect=1 Resource_List.nodes=1 >> Resource_List.procs=6 >> >> So it looks like only one processor is reserved. If I change procs=6 to >> nodes=1:ppn=6 then it works right: >> [root@biocluster init.d]# tracejob -v 76340 >> /var/spool/torque/server_priv/accounting/20120921: Successfully located >> matching job records >> /var/spool/torque/server_logs/20120921: Successfully located matching >> job records >> /var/spool/torque/mom_logs/20120921: No such file or directory >> /var/spool/torque/sched_logs/20120921: No such file or directory >> >> Job: 76340.biocluster.igb.illinois.edu >> >> 09/21/2012 14:50:12 S enqueuing into test_queue, state 1 hop 1 >> 09/21/2012 14:50:12 S Job Queued at request of >> [email protected], owner = >> [email protected], job name = STDIN, queue = test_queue >> 09/21/2012 14:50:12 A queue=test_queue >> 09/21/2012 14:50:13 S Job Run at request of >> [email protected] >> 09/21/2012 14:50:13 S Not sending email: User does not want mail of >> this type. >> 09/21/2012 14:50:13 A user=danield group=danield jobname=STDIN >> queue=test_queue ctime=1348257012 qtime=1348257012 etime=1348257012 >> start=1348257013 [email protected] >> >> exec_host=compute-0-1/5+compute-0-1/4+compute-0-1/3+compute-0-1/2+compute-0-1/1+compute-0-1/0 >> Resource_List.mem=3gb Resource_List.ncpus=1 >> Resource_List.neednodes=1:ppn=6 Resource_List.nodect=1 >> Resource_List.nodes=1:ppn=6 >> >> Can someone let me know why this would be, and why isnt ncpus set >> correctly in the lastjob. If I am mistaken about what the procs field >> mean, please let me know. >> >> Dan >> _______________________________________________ >> mauiusers mailing list >> [email protected] >> http://www.supercluster.org/mailman/listinfo/mauiusers >> > > > > -- > Clotho Tsang > Senior Software Engineer > Cluster Technology Limited > Email: [email protected] > Tel: (852) 2655-6129 > Fax: (852) 2994-2101 > Website: www.clustertech.com > > -- Clotho Tsang Senior Software Engineer Cluster Technology Limited Email: [email protected] Tel: (852) 2655-6129 Fax: (852) 2994-2101 Website: www.clustertech.com
_______________________________________________ mauiusers mailing list [email protected] http://www.supercluster.org/mailman/listinfo/mauiusers
