[GE users] auto rerun lower priority jobs when higher are waiting?

reuti reuti at staff.uni-marburg.de
Tue Aug 25 21:28:23 BST 2009

Am 25.08.2009 um 16:49 schrieb jesperkrogh:

>>> http://krogh.cc/~jesper/all.q.txt
>>> http://krogh.cc/~jesper/rerunnable.q.txt
>>> Can you see the misconfiguration?
>> Did you restrict somewhere the number of total slots to 32 on this
>> machine or per machine?
> The largest machine has 32 slots .. but they range in between 2-32,  
> so I
> configured the queue to 32 and setting it individually per.  
> execution node
> in the complex-attributes to the actual number.

Well, then the new jobs can't start, as all slots are already used up  
per node. SGE will not killl/remove and job before the new one had  

But if you suspend the subordinate queue rerunnable.q already when  
one slot in all.q is used, you could set the limit also in the queue  
definition. There you can specify the number of slots also per  
hostgroup like:

slots 2,[@amd2384=8],[@amd270=4]

Of course, then you can have only these two queues per node to avoid  
oversubscription by a third one.

-- Reuti


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

More information about the gridengine-users mailing list