after adding some debug statements to this function, i figured it out: the job was asking for more vmem then was available on the nodes (but less then maximum). this is a clear error, yet it's impossible to figure that out without adding extra debug code.
stijn > hia ll, > > i'm struggling with following issue (maui 3.3.1): > node with 32cores per node, job in torque that request nodes=1:ppn=32 > > jobs don't get started for no apparent reason. > > when i set LOGLEVEL to 8, i get > 08/03 16:49:40 MNodeGetTC(node,32,32,0,1,0) > 08/03 16:49:40 INFO: 31 tasks located > > and a bit later > 08/03 16:49:40 MNodeGetTC(node,32,32,0,1,0) > 08/03 16:49:40 INFO: 32 tasks located > > (with node the same node). > > what is going on here? the first one blocks the start of the job, but > why does maui think the node can only take 31 tasks? > > i'm trying to make sense out of MNodeGetTC, but i don't make much out of > it. (setting LOGLEVEL higher doesn't help much either) > > > > stijn -- http://hasthelhcdestroyedtheearth.com/ _______________________________________________ mauiusers mailing list [email protected] http://www.supercluster.org/mailman/listinfo/mauiusers
