Hi,

> I would like to perform scheduling based on the 
> number of CPUs in use.  This section of the manual 
> is marked "(Under Construction)", so I'm a bit lost.  

Ah yes, this is a bit obfuscated. Maybe there is a key similar to the following:

NODEALLOCATIONPOLICY    PRIORITY
NODECFG[DEFAULT] PRIORITYF='-JOBCOUNT'

That way you define your own priority-function for the node allocations,
which is in this case inversely to the jobs on that node.
Else I am also at loss.

Cheers,
Ronny
_______________________________________________
mauiusers mailing list
[email protected]
http://www.supercluster.org/mailman/listinfo/mauiusers

Reply via email to