On Mon, Mar 27, 2006 at 09:38:13AM -0600 or thereabouts, Gabe Turner wrote: > On Mon, Mar 27, 2006 at 02:23:39PM +0200, Gonzalo Merino wrote: > > Hello, > > > > We are running maui-3.2.6p11 on top of Torque-1.0.1p6 for managing a > > cluster of about 200 nodes at this moment. > > We have set up the scheduling interval to 2 minutes in the maui conf: > > > > # grep RMPOLLINTERVAL /var/spool/maui/maui.cfg > > RMPOLLINTERVAL 00:02:00 > > > > But on the maui.log I can see some messages that suggest that this is not > > being respected. It seems that maui scheduling cycles are happening much > > more often... several per minute! > > If you have scheduling = True in your torque server config, maui will poll > every time a job is submitted.
Hi Gonzalo, I was going to say have a look at the bit on JOBAGGREGATIONTIME on http://www.clusterresources.com/products/mwm/docs/a.ilargeclusters.shtml#jobs but then we have JOBAGGREGATIONTIME 00:00:00 but still see maui.log.2:03/27 17:23:52 INFO: scheduling complete. sleeping 30 seconds maui.log.1:03/27 17:23:57 INFO: scheduling complete. sleeping 30 seconds so it does not look to work as I had expected. The scheduling = False does do the job though. (See you next week I guess) Steve > > HTH, > > Gabe > > -- > Gabe Turner [EMAIL PROTECTED] > UNIX Systems Administrator, > University of Minnesota Supercomputing Institute > for Digital Simulation and Advanced Computation www.msi.umn.edu > _______________________________________________ > mauiusers mailing list > [email protected] > http://www.supercluster.org/mailman/listinfo/mauiusers -- Steve Traylen [EMAIL PROTECTED] http://www.gridpp.ac.uk/ _______________________________________________ mauiusers mailing list [email protected] http://www.supercluster.org/mailman/listinfo/mauiusers
