[torqueusers] Limiting max CPUs used by a queue

Peter van Heusden pvh at sanbi.ac.za
Wed Apr 23 08:15:41 MDT 2008


Ah, I'm using maui.

Thanks, I'll implement this.

Peter

Jerry Smith wrote:
> Peter,
> 
> Are you using the pbs_sched, or something like maui or moab for scheduling?
> 
> If you are using maui, have a look at :
> http://www.clusterresources.com/products/maui/docs/policynote-3.0.shtml
> 
> In moab you can set:
> 
> CLASSCFG[long]      MAXPROC=16  # where "long" is the name of your long
> run queue.
> 
> 
> Jerry
> 
> Peter van Heusden wrote:
>> Hi there
>>
>> We are running Torque 2.3.0 on an IBM p690 with 32 CPUs. We've got two
>> queues - the default batch queue (for jobs less with less than 2 hours
>> runtime) and the long run queue (no runtime limit). We would like to
>> limit the long run queue to never use more than 16 CPUs - I've set the
>> max_running to 16, the resources_max.ncpus to 16 and the
>> resources_available.ncpus to 16. I would think that this means that if I
>> submit a job using 16 CPUs, and then one using 2 CPUs, the 2 CPU job
>> would remain queued until the 16 CPU job is finished - but this is not
>> the case, both jobs run.
>>
>> Any idea why this is so?
>>
>> Peter
>> _______________________________________________
>> torqueusers mailing list
>> torqueusers at supercluster.org
>> http://www.supercluster.org/mailman/listinfo/torqueusers
>>
>>
>>   
> 



More information about the torqueusers mailing list