[GE users] submitting parallel jobs

reuti reuti at staff.uni-marburg.de
Thu Dec 17 18:54:18 GMT 2009


Hi,

Am 15.12.2009 um 12:24 schrieb cgull:

> Hi
> Sorry to come back to this. But confused now with allocating memory  
> to a queue.
>
> I know submit to these three queues with qsub -pe orte*
> Which is working well.
>
> The problem we have is that the job size allowable can be different  
> for each of these machines.
>
> i.e. We know a job will only run on hill and bira, as job will take  
> 40GB which we know is too large for cevert.
>
> We could do different orte_"machine" but i know we will have more  
> machines in the future which will make this more complicated for  
> the users submitting.
>
> Would be nice if we could set the limit of the queue for cevert to  
> 35GB then if we submit with something like -l mem=40GB (or $# -l  
> mem=40GB in script) SGE would automatically know that it cannot  
> submit job onto cevert.
>
>
> If something like this was possible we could also assign hill and  
> bira to have limits 45GB and 80GB respectively, therefore if we  
> submitted a job with -l mem =75GB only bira would allow the job to  
> run on it.
>
> I am confused with how i really should be setting this as there  
> appear to be many different configurations. If you could shed any  
> light on how it should be configured it would be appreciated.
>
> Thanks for your time in advance

http://gridengine.info/2009/12/01/adding-memory-requirement-awareness- 
to-the-scheduler

-- Reuti

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=233969

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list