[Mauiusers] Jobs apparently stuck in deffered state.

Chris Johnson johnson at nmr.mgh.harvard.edu
Tue Dec 6 06:08:50 MST 2005

On Mon, 5 Dec 2005, Michael Edwards wrote:

> Check out the DEFERTIME parameter.  I had a setup where if I didn't
> set DEFERTIME to zero jobs would stall indefinately.  Your problem
> doesn't sound that severe so you might not need to disable it, but it
> might be worth playing with and seeing if the default is, say, fifteen
> minutes or something.
> Just a thought.

      Very good thought.  Seems to have worked.  However ---

      I have another maui mini test cluster which, as far as I know, is
configured the same, and it doesn't have this problem.  The big 
difference is that the strangely behaving cluster is P-IIIs while one
that works is opterons.

      Also, I'm trying very hard to get maui to obey or at least 
simulate the node_pack = false parameter in the torque configuration. 
It ain't happening.  Jobs are being scheduled for the same node which 
causes some to wait so they don't get run in sequence the as 
they do with torque C scheduler.  I want one job per processor.

      I've tried playing with NODEALLOCATIONPOLICY but this doesn't 
seem to be helping much.

      Any suggestions?

Chris Johnson               |Internet: johnson at nmr.mgh.harvard.edu
Systems Administrator       |Web:      http://www.nmr.mgh.harvard.edu/~johnson
NMR Center                  |Voice:    617.726.0949
Mass. General Hospital      |FAX:      617.726.7422
149 (2301) 13th Street      |Doctors don't save lives.  The best they can hope
Charlestown, MA., 02129 USA |to do is save life.  Not the same thing.  Me

More information about the mauiusers mailing list