Hi all, this issue has some history:
My original problem: http://www.supercluster.org/pipermail/mauiusers/2009-February/003711.html Roy's solution: http://www.supercluster.org/pipermail/mauiusers/2009-February/003732.html Briefing: maui never schedules all IDLE queue. The scenario: you have 2 queues, quque A looking for nodes with priority A (PA) and queueu B lookinf for nodes with priority B (PB). Nodes with PA are all busy, but you have many jobs in queue A. Nodes with PB are free, and many jobs in queue B. So, if jobs in queueA are in top of global IDLE queue, jobs from queue B, which are in bottom, have resources available but they are never scheduled. Maui says: # checkjob 14611086 checking job 14611086 State: Idle Creds: user:lhpilot002 group:lhpilot class:test qos:lhclhcb WallTime: 00:00:00 of 8:00:00 SubmitTime: Wed Dec 22 11:07:07 (Time Queued Total: 00:39:32 Eligible: 00:39:32) StartDate: -00:36:30 Wed Dec 22 11:10:09 Total Tasks: 1 Req[0] TaskCount: 1 Partition: ALL Network: [NONE] Memory >= 0 Disk >= 0 Swap >= 0 Opsys: [NONE] Arch: [NONE] Features: [test] IWD: [NONE] Executable: [NONE] Bypass: 0 StartCount: 0 PartitionMask: [ALL] PE: 1.00 StartPriority: 7 job can run in partition DEFAULT (120 procs available. 1 procs required) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ This problem was solved by Roy with a path, but this patch has not been introduced into prod branch and I see no solution to this problem in maui's doc. So now I'm wondering if we're doing something wrong or if it's real a big maui's bug which need urgent solution (meanwhile I'll use Roy's patch again). So, what are other admins with similar scenario solving this issue? maui-server-3.3-1.x86_64 TIA, Arnau _______________________________________________ mauiusers mailing list [email protected] http://www.supercluster.org/mailman/listinfo/mauiusers
