In the message dated: Tue, 01 Jan 2013 17:38:23 +0100,
The pithy ruminations from Reuti on
<Re: [gridengine users] How prevent abnormal nodes load using qsub?> were:
=> Am 31.12.2012 um 09:34 schrieb Semi:
=>
=> > The memory is not a problem, the problem CPU load,
=> > every python process runs 2 other processes and this stuck nodes.
=> > For examle: 16 CPU nodes run 48 python processes.
=>
[SNIP!]
=>
=> If all load is coming from his Python script, it needs to be adjusted to run
only serial and not
=> taking all available cores in the machine.
Assuming that there's a PE for running multi-threaded jobs already
configured in the cluster, the user could submit the job to that PE (named
"threaded" in the example below). If each python process has 2 children,
then the job submission might look like:
qsub -pe threaded 3 myjob.py
to tell SGE that each job uses 3 slots, so that SGE won't oversubscribe
a node.
Mark
=>
=> -- Reuti
=>
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users
