From: Brian Bennion (brian_at_youkai.llnl.gov)
Date: Wed Jan 12 2005 - 17:15:05 CST
Hi Cheri,
If the batch system is loading jobs on top of each other its broken or at
least its not set up correctly.
if you need a set number of nodes then
qsub -q "somequeue" -l nodes=6 somejobscript.in
if each has to procs then -l nodes=6:ppn=2
look in the man pages for pbs_resources for more "-l" commands
Regards
Brian
On Wed, 12 Jan 2005, Cheri M Turman wrote:
> Hi everyone,
> I am running into an issue with a scyld beowulf cluster. I read in the
> release notes that it is not good to run multiple namd jobs on the same
> nodes. Does this apply to other programs running on the same nodes as
> an namd job? I am experiencing a problem with my namd job hanging. It
> still says status "running" but I'm getting no output or logging data.
> Some days it runs fine but others it doesn't. Sometimes it hangs before
> entering startup phase and others end right as the startup phase ends.
> I think it may be an issue when the individual jobs are sent to
> different nodes. I share this 90 node cluster over the whole school so
> I am limited as to what I can do if it is a problem that other users are
> using the same nodes as I am. If this is the case, does anyone know how
> to choose what nodes namd runs on in PBS scripting?
> Thanks for any help,
> Cheri
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> Cheri M. Turman
> Graduate Student
> University of Texas-Houston Medical School
> 6431 Fannin
> Houston, TX 77030 USA
>
> e-mail: cheri.m.turman_at_uth.tmc.edu
> Ph.: 713-500-6126
> Fax: 713-500-0652
>
*****************************************************************
**Brian Bennion, Ph.D. **
**Computational and Systems Biology Division **
**Biology and Biotechnology Research Program **
**Lawrence Livermore National Laboratory **
**P.O. Box 808, L-448 bennion1_at_llnl.gov **
**7000 East Avenue phone: (925) 422-5722 **
**Livermore, CA 94550 fax: (925) 424-6605 **
*****************************************************************
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:39:06 CST