Well, I have my script that submits 25 jobs to PBS, each a little different,
but, they can all run independently, and not clobber each other.
However, when I look at qstat, it seems to be executing the jobs in serial
anyway by assigning every node to the first job, waiting for completion,
then the next job, etc. I've set the queue max_cpus to 2 and the NCPUS in
the script is only 1. But it doesn't seem like PBS is scheduling each job on
a seperate machine like it should. I can't seem to find which queue property
or server property is doing this. or maybe I'm just missing something. I get
qstats that look like this:
Job ID Username Queue Jobname SessID NDS TSK Memory Time S
Time
--------------- -------- -------- ---------- ------ --- --- ------ ----- -
-----
107.athisl.quan testbed workq PseDistrib 3956 1 1 -- 10000 R
--
beo9+beo8+beo7+beo6+beo5+beo4+beo3+beo21+beo20+beo2+beo19+beo18+beo17+beo16
+beo15+beo14+beo13+beo12+beo11+beo10+beo1
108.athisl.quan testbed workq PseDistrib -- 1 1 -- 10000 Q
--
--
109.athisl.quan testbed workq PseDistrib -- 1 1 -- 10000 Q
--
--
110.athisl.quan testbed workq PseDistrib -- 1 1 -- 10000 Q
--
--
111.athisl.quan testbed workq PseDistrib -- 1 1 -- 10000 Q
--
--
...
All the jobs except for the first one get Q'd and just wait to be executed
one at a time.
Thanks again for everyone's help with this project.
Brian
Brian E. Williams
Software Developer and Systems Administrator
Quantum Leap Innovations
(302)894-8036 [EMAIL PROTECTED]
http://copland.udel.edu/~brianw
-------------------------------------------------------
This SF.NET email is sponsored by:
SourceForge Enterprise Edition + IBM + LinuxWorld = Something 2 See!
http://www.vasoftware.com
_______________________________________________
Oscar-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/oscar-users
