[Mauiusers] jobs not running when procs available.

Marc Noguera marc at klingon.uab.es
Thu Feb 19 08:14:32 MST 2009


Hello everyone,
I am facing this strange problem with maui-3.2.6p21/torque 2.3.3 on a
fedora core 6 x86_64 system.

qstat -na1 transmet4 results in:
quanta.uab.es:
                                                                        
Req'd  Req'd   Elap
Job ID               Username Queue    Jobname          SessID NDS   TSK
Memory Time  S Time
-------------------- -------- -------- ---------------- ------ ----- ---
------ ----- - -----
45216.quanta.uab     sergi    transmet fos-cpenta-H15-d   7313     1 
--    --  59:59 R 52:30   borg46
45225.quanta.uab     sergi    transmet fos-cpenta-C2-di   7332     1 
--    --  59:59 R 52:31   borg46
45566.quanta.uab     oleg     transmet FeH+2xHFIP_MHB_I   2695     1 
--    --  59:59 R 14:23   borg47+borg47+borg47+borg47
45585.quanta.uab     joaquin  transmet irtp_ph_h_ph_alk   3616     1 
--    --  59:59 R 00:00   borg47+borg47+borg47+borg47
45612.quanta.uab     max      transmet ph-cf3-cat-int-4  24166     1 
--    --  59:59 R 00:00   borg46
45613.quanta.uab     max      transmet ph-cf3-cat-int-4    --      1 
--    --  59:59 Q   --     --
45622.quanta.uab     max      transmet ph-NMe2-cat-int-    --      1 
--    --  59:59 Q   --     --
45623.quanta.uab     max      transmet ph-NMe2-cat-int-    --      1 
--    --  59:59 Q   --     --


borg47 has 8procs(np=8) which are all occupied. borg46 has also 8
procs(np=8) but only three are running.
Jobs 45613, 45622 and 45623 all are asking for 1 proc. Checkjob results in

checking job 45613

State: Idle
Creds:  user:max  group:transmet  class:transmet4  qos:DEFAULT
WallTime: 00:00:00 of 2:11:59:59
SubmitTime: Thu Feb 19 15:30:41
  (Time Queued  Total: 00:38:19  Eligible: 00:38:19)

Total Tasks: 1

Req[0]  TaskCount: 1  Partition: ALL
Network: [NONE]  Memory >= 0  Disk >= 0  Swap >= 0
Opsys: [NONE]  Arch: [NONE]  Features: [transmet4]


IWD: [NONE]  Executable:  [NONE]
Bypass: 12  StartCount: 0
PartitionMask: [ALL]
PE:  1.00  StartPriority:  38
job can run in partition DEFAULT (5 procs available.  1 procs required)

So job are in state idle although the job can run. This is for all jobs
queued.

Anyone experienced this problem?

Thank you
Marc



More information about the mauiusers mailing list