[GE users] Newbie questions

Schmitz Dale M Contr 20 IS/INPTG Dale.Schmitz at offutt.af.mil
Thu Apr 7 16:21:25 BST 2005


I see your other e-mail and I'll respond to it here.  First off, I
qsub'd a number of sleeper.sh jobs like you suggested and here's the
qstat on that:

queuename            qtype used/tot. load_avg arch       states
------------------------------------------------------------------------
----
host150.q            BIP   1/8       8.16     solaris64  
     33     0 QLOGIN     dschmitz     r     04/07/2005 09:26:40 MASTER

------------------------------------------------------------------------
----
host01.q             BIP   2/2       0.03     solaris64  
     34     0 Sleeper    dschmitz     r     04/07/2005 10:13:39 MASTER

     36     0 Sleeper    dschmitz     r     04/07/2005 10:13:39 MASTER

------------------------------------------------------------------------
----
host02.q             BIP   2/2       0.55     solaris64  
     37     0 Sleeper    dschmitz     t     04/07/2005 10:13:39 MASTER

     39     0 Sleeper    dschmitz     t     04/07/2005 10:13:39 MASTER

------------------------------------------------------------------------
----
host03.q             BIP   0/2       1.17     solaris64  
------------------------------------------------------------------------
----
host04.q             BIP   2/2       0.24     solaris64  
     35     0 Sleeper    dschmitz     t     04/07/2005 10:13:39 MASTER

     38     0 Sleeper    dschmitz     t     04/07/2005 10:13:39 MASTER

------------------------------------------------------------------------
----
host05.q             BIP   1/2       1.13     solaris64  
     40     0 Sleeper    dschmitz     t     04/07/2005 10:13:40 MASTER

------------------------------------------------------------------------
----
host06.q             BIP   0/2       1.57     solaris64  

As for your other questions, I have a network of 7 systems, one with 8
CPUs and 6 with two CPUs.  We launch the big application on the 8-CPU
system to keep the execution time around 2 hours (wall-clock time).  I
was hoping that the grid engine software would allow me to use all 20
CPUs, but I've yet to learn just how to do that.  I've contacted the
software vendor about it and they tell me that the job is a good
candidate for this environment, but since I'm not a programmer myself,
nor am I anywhere near a programming environment, I lack the knowledge
and foresight to ask particular questions.  I'm learning them as I go
here.

Hope this helps.
Dale

-----Original Message-----
From: John Hearns [mailto:john.hearns at streamline-computing.com] 
Sent: Thursday, April 07, 2005 9:51 AM
To: users at gridengine.sunsource.net
Subject: RE: [GE users] Newbie questions

On Thu, 2005-04-07 at 09:34 -0500, Schmitz Dale M Contr 20 IS/INPTG
wrote:
> Thanks, but I think I've worked through the basic stuff.  I'm able to
> launch our application (which is interactive) but it's still bound to
> one machine.  

I suspect you haven't got the jobs starting on the exec hosts properly.

Try this test.
Look in $SGE_ROOT/examples and choose one of the example programs.
Lets' take sleeper.sh 

Run  'qsub sleeper.sh 600'  several times - in fact as many times
as you have compute nodes.
Now do 'qstat -f' and we can see if the jobs are being run on all the
nodes.





---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list