[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

great balls of fire



Jerry and Mike,

9:42pm....qstat -f shows:

Queue Name                       Queue Type    Quan  Load          State
----------                       ----------    ----  ----          -----
psc10_gsim                       batch         2/2   1.68  er      UP        
    freyberg gsim.batc              340   0:13  r       RUNNING   06/28/99 21:32:22 
    freyberg gsim.batc              347   0:20  r       RUNNING   06/28/99 21:32:25 
psc11_gsim                       batch         2/2   1.68  er      UP        
    freyberg gsim.batc              334   0:7   r       RUNNING   06/28/99 21:32:21 
    freyberg gsim.batc              335   0:8   r       RUNNING   06/28/99 21:32:21 
psc12_gsim                       batch         2/2   1.68  er      UP        
    freyberg gsim.batc              336   0:9   r       RUNNING   06/28/99 21:32:21 
    freyberg gsim.batc              337   0:10  r       RUNNING   06/28/99 21:32:22 
psc2_gsim                        batch         2/2   1.77  aer     ALARM     
    freyberg gsim.batc              348   0:21  r       RUNNING   06/28/99 21:32:25 
    freyberg gsim.batc              349   0:22  r       RUNNING   06/28/99 21:32:26 
psc3_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              343   0:16  r       RUNNING   06/28/99 21:32:23 
    freyberg gsim.batc              344   0:17  r       RUNNING   06/28/99 21:32:24 
psc4_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              338   0:11  r       RUNNING   06/28/99 21:32:22 
    freyberg gsim.batc              339   0:12  r       RUNNING   06/28/99 21:32:22 
psc5_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              332   0:5   r       RUNNING   06/28/99 21:32:20 
    freyberg gsim.batc              333   0:6   r       RUNNING   06/28/99 21:32:21 
psc6_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              328   0:1   r       RUNNING   06/28/99 21:32:20 
    freyberg gsim.batc              329   0:2   r       RUNNING   06/28/99 21:32:20 
psc7_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              341   0:14  r       RUNNING   06/28/99 21:32:22 
    freyberg gsim.batc              342   0:15  r       RUNNING   06/28/99 21:32:23 
psc8_gsim                        batch         2/2   1.70  er      UP        
    freyberg gsim.batc              330   0:3   r       RUNNING   06/28/99 21:32:20 
    freyberg gsim.batc              331   0:4   r       RUNNING   06/28/99 21:32:20 
psc9_gsim                        batch         2/2   1.68  er      UP        
    freyberg gsim.batc              345   0:18  r       RUNNING   06/28/99 21:32:24 
    freyberg gsim.batc              346   0:19  r       RUNNING   06/28/99 21:32:25 


----Pending Jobs ---------------------------------------------------------------
freyberg gsim.batc     350   0:23 gsim                     QUEUED 06/28/99 21:32:26 

This will continue to generate crap until I "qdel" the pending job.... A "top"
on the machines shows each gsim job occupying 50% of the CPU resources.  I hope
this means each gsim job has its own cpu....

HOW:  First I am only using PRODUCTION celeg and gsim_batch, second I have some
scripts in /home/freyberg/DQS_TEST/GSIM that run the job and make the next 
script.

In particular look at:  gsim.batch   and mk_gsim.batch

I first tuned up gsim.batch such that it works, and then tuned up the
perl script, mk_gsim.batch, so that when it reads in gsim.batch it
spits out a new version gsim.batch that works.  I will work on a more
object approach so that the user has a cleaner interface.

But this script appears to work and you can try to modify it for your
jobs.  To start the jobs one simply does "qsub -l gsim gsim.batch"
once, which will then automatically recreate a new gsim.bat and submit
that job...until all the queues are filled.

I'll let this run until morning to see if everything holds together.

Arne

-- 
Arne Freyberger         
JLAB (Mail Stop 12H)    Phone: (757) 269-6268  
12000 Jefferson Avenue  Pager: (757) 249-6268  freyberg@cebaf.gov
Newport News, VA 23606   Phax: (757) 269-5800  http://claspc1.cebaf.gov