Re: Queuing System for Parallel Clusters
On Tue, Oct 24, 2000 at 08:28:59AM +1100, Drake Diedrich wrote:
> On Mon, Oct 23, 2000 at 04:57:51PM +0100, Tim Cutts wrote:
> > NQS derivatives are all hopeless, IMHO. They perform host selection at
> > job submission time rather than job execution time, which is a really
> > broken idea. They're fine for controlling queues on a single large
> > parallel machine, but they're no good for managing clusters.
> DQS determines nodes at execution time. GNQS (according to the
> documentation) does as well. Use -l, not -q, to specify node requirements.
> qsub -q xx will run a job only on the specified queue, but that's what it's
My apologies. My knowledge of GNQS must be hopelessly out of date. It
certainly didn't do that (or if it was meant do, it didn't work) when I
last tried it.