[torquedev] Torque Semantics
cfr100 at psu.edu
Tue Jun 15 06:18:20 MDT 2010
On that note, what about compute node OS as a resource? Work is being
done on light weight compute node kernels (eg Catamount, Kitten, etc),
and it would be nice to try/use them under certain jobs. However, it
is unclear that every job will run under a one of these (almost
certainly will not).
It seems to me that the scheduler should manage the compute node OS as
a resource. This might mean starting and stopping services or it might
mean rebooting it to a different kernel.
2010/6/11 Ken Nielson <knielson at adaptivecomputing.com>:
> Yes. It seems over time keywords like procs and ncpus have been added to
> help schedulers better utilize the cluster but TORQUE has not been
> modified to keep up. We need to examine the other resources such as mem
> to make sure they are uniformly interpreted as well.
More information about the torquedev