[GE users] Script problem on SGE.

templedf dan.templeton at sun.com
Sun Mar 15 15:53:57 GMT 2009


First place to start is with qstat -j <jobid>.  If sched_job_info isn't 
turned off in your cluster, that should tell you why the job is still 
pending.  Another thing to check would be the lam PE definition.  Could 
the allocation rule be set to $pe_slots?  If you're using resource 
quotas, qqouta will tell you if you're running into a quota limit.  You 
could also try resubmitting the job with -w v. Rather than actually 
submitting the job, -w v tells qsub to just check that the job isn't 
asking for something impossible.

Daniel

bizon wrote:
> Hello!
>
> This is my queue script:
>
> #$ -q all.q
> #$ -cwd
> #$ -N ala-test
> #$ -j y
> #$ -V
> #$ -notify
> cd $SGE_O_WORKDIR
> #$ -pe lam 8
> export LAM_MPI_SOCKET_SUFFIX=$JOB_ID.$JOB_NAME
> lamboot
> /usr/bin/mpirun -np $NSLOTS \
>  cpmd.x ala.inp >&ala.out
>
> I have nodes with 8 processors. And when I try use 2 nodes per 8 processors I changed in script only this #$ -pe lam 8 to #$ -pe lam 16. I don't now what is wrong but my job still pending in queue. What should I do??
>
> Please help!
> Best.
>
> ------------------------------------------------------
> http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=132392
>
> To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].
>

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=132445

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list