[torqueusers] Configuration problems

Steve Young chemadm at hamilton.edu
Thu Jul 24 12:22:17 MDT 2008


I'm also wondering perhaps you might need to use mpiexec from OSC......

http://www.osc.edu/~pw/mpiexec/index.php

I needed that in order to get mpich2 and torque to work properly.  
Using this version then you don't have to use machinefile and torque  
maintains control of spawning the jobs on the nodes via pbs_mom which  
is faster.

Other than that like Justin pointed out you'd need to post more  
information about your configuration to figure it out.

-Steve


On Jul 24, 2008, at 1:38 PM, Justin Finnerty wrote:

>> From the symptoms it seems that only the 2 cpu front end node is a
> recognised execute host.  Hence only 2 serial jobs run and you  
> require the
> explicit mpi machinefile.
>
> It may seem obvious but have you added all the execute nodes to the  
> node
> file in [torque_dir]/server_priv?
>
> Are all the execute nodes listed when you run "pbsnodes"?
>
> If they are all listed then I think you need to post more  
> information such
> as your queue configuration etc.
>
> Cheers
>       Justin
>
>> Hi, everyone
>>
>>
>> I have a little cluster with 10 nodes Intel and 40 processors (10 x
>> Quad-Core).
>> My front-end have an Intel Dual-Core processor.
>
>
> -- 
> Dr Justin Finnerty
> Rm W3-1-218         Ph 49 (441) 798 3726
> Carl von Ossietzky Universität Oldenburg
>
>
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers



More information about the torqueusers mailing list