Torque Users,
I'm running torque 2.1.2 with the pbs_sched scheduler.
I have some users who would like to get in with a high priority
job , having it start running immediately even when no nodes are
free.
I was able to do this on a shared memory machine by suspending the
lower priority job and starting the new job, then unsuspending
after the high priority job completes.
All nodes are 4 processors, with ppn=4.
I can get the older jobs to suspend, but how do I get
the new job to start on a node with all cpus already
reserved?
Thanks,
- Jim Coyle
|