This an imporatant piece of information. However I'm not sure if that leads me to a solution for the problem I asked first: some nodes have vacant processors doing nothing at the moment, but jobs stay in Q status with the "insufficient idle procs available" notification. These jobs are not asking for multiple procs, but only one proc., which is available at the moment.
I'll probably continue with Jan Ploski's proposal (Jan 24, 2008) and see how it's going - and keep you updated! On Jan 25, 2008 12:58 AM, Garrick Staples <[EMAIL PROTECTED]> wrote: > > The "busy" node state in PBS is based on the node's load average. It has > nothing to do with what jobs are assigned, running, supposed to be running, or > any config in pbs_server or the scheduler. > > See the pbs_mom manpage on $ideal_load and $max_load. > > > _______________________________________________ > mauiusers mailing list > mauiusers@supercluster.org > http://www.supercluster.org/mailman/listinfo/mauiusers > > _______________________________________________ mauiusers mailing list mauiusers@supercluster.org http://www.supercluster.org/mailman/listinfo/mauiusers
