[torquedev] memory limit enforcement by pbs_mom - REQUEST FOR
troy at osc.edu
Thu Feb 2 10:25:05 MST 2006
On Tue, 2006-01-31 at 13:53 -0700, Dave Jackson wrote:
> TORQUE may have inherited this from the OSC patches back in the early
That would make sense, based on the behavior described.
> Did OSC see any issues with 'mem' enforcement on serial jobs?
Yes, but it was much more acute with parallel jobs. IIRC, vmem is
enforced on a per-node basis, whereas mem is enforced as an aggregate
across the entire job. That was exacerberated by PBS and Maui/Moab not
agreeing on what mem meant; the schedulers were assuming mem was a per-
Troy Baer troy at osc.edu
Science & Technology Support http://www.osc.edu/hpc/
Ohio Supercomputer Center 614-292-9701
More information about the torquedev