[Swift-user] Coaster provider is not allocating dedicated nodes

Mihael Hategan hategan at mcs.anl.gov
Wed Jan 20 15:02:19 CST 2010


On Wed, 2010-01-20 at 11:01 -0600, Mihael Hategan wrote:
> On Wed, 2010-01-20 at 09:38 -0600, Michael Wilde wrote:
> > Using the sites entry below, I see that coasters is allocating 8 
> > *shared* nods rather than *dedicated* nodes; hence its running many more 
> > processes per node than it should, causing the jobs to run longer than 
> > expected and exceed their walltime.
> 
> Right. It looks like the pbs provider uses nodes= and doesn't mess with
> ppn=, which means it allocate nodes as defined by the local policy
> (which may mean cores instead of nodes).
> 
> I suggest setting workersPerNode to 1, but then you run into the
> previous problem, which I'm trying to fix now and for which I have an
> open ticket with PADS support.
> 

The ssh problem on PADS was fixed and I committed a patch to the branch
to start multiple instances of the app (cog r2683).

Mihael




More information about the Swift-user mailing list