[Mauiusers] maui 3.2.6p11 HARD MAXPROC limit blocking job

Paul Van Allsburg vanallsburg at hope.edu
Mon Apr 30 07:29:06 MDT 2007


Hi,
I have 2 jobs waiting for 16 proc that will not start due to
 "violates active HARD MAXPROC limit of 22 for group"

The cluster has 32 processors,  the basic maui config is:
QUEUETIMEWEIGHT       1
BACKFILLPOLICY        FIRSTFIT
RESERVATIONPOLICY     CURRENTHIGHEST
NODEALLOCATIONPOLICY  CPULOAD
CREDWEIGHT            1
USERWEIGHT            1
GROUPWEIGHT           1
CLASSWEIGHT           1

USERCFG[DEFAULT]      MAXPROC=18
GROUPCFG[DEFAULT]     MAXPROC=22

CLASSCFG[normal]      MAXPROC=22
CLASSCFG[debug]       MAXPROC=30
CLASSCFG[admin]       MAXPROC=32

XFACTOR               1
XFMINWCLIMIT          1440


Both users below are in the same group, hence limited to 22 processors.  
The hinkle jobs each
use 4 proc, run for less than a day and qsub a new job upon their 
completion.   I
have been waiting for maui to hold two hinkle jobs and run the at least 
the 4hour request for
16 proc but it's not happening. 

                                                            Req'd  
Req'd   Elap
Job ID          Username Queue    Jobname    SessID NDS TSK Memory Time  
S Time
--------------- -------- -------- ---------- ------ --- --- ------ ----- 
- -----
9327.curie.chem vanallp  admin    qchempbs      --    8  --    --  10:00 
Q   --
    --
9354.curie.chem vanallp  admin    qchempbs16    --    8  --    --  04:00 
Q   --
    --
9372.curie.chem hinkle   normal   oligo1-7im    --    2  --    --  20:00 
R 08:23
   curie09/1+curie09/0+curie06/1+curie06/0
9373.curie.chem hinkle   normal   oligo1-2im    --    2  --    --  16:00 
R 05:05
   curie15/1+curie15/0+curie13/1+curie13/0
9374.curie.chem hinkle   normal   oligo1-9im    --    2  --    --  16:00 
R 01:55
   curie11/1+curie11/0+curie10/1+curie10/0

All I can find for an error is the HARD MAXPROC limit.  The priority for 
job 9354
cycles up in the log and gets reset back to 1 when a hinkle job finished 
and a new one
gets resubmitted.

Is there somthing obvious I'm missing that explains why the request for 
16 proc does not run
while three requests for 4 proc continue to cycle thru?  

Thanks!
Paul


-- 
Paul Van Allsburg  
Computational Science & Modeling Facilitator
Natural Sciences Division,  Hope College
Holland, Michigan 49423




More information about the mauiusers mailing list