Hi,

> Am 25.08.2014 um 23:29 schrieb Noah Knowles <noah...@gmail.com>:
> 
> Hi, a different question-- again, I am using OGS/GE 2011.11p1 on ROCKS. We 
> have a small cluster with a combination of 12- and 16-core blades. If I do:
>> qsub -pe orte 60
> with no jobs on the cluster, what gets allocated is usually sub-optimal; 
> instead of 3*16+12 (4 full nodes), I might get 2*16+2*12+4 (4 full nodes and 
> one partial).
> 
> Is there a general way to force allocation of the minimum possible number of 
> nodes, other than requesting the nodes explicitly?

No.

You could check in the beginning of the job script whether the granted 
allocation suits you, or otherwise reschedule the job (after a random wait 
maybe).

-- Reuti


> We're already using exclusive scheduling, so this would make a lot more 
> sense. This feels like a newb question but I haven't found an answer by 
> searching.
> 
>> $ qconf -sp orte
>> pe_name            orte
>> slots              9999
>> user_lists         NONE
>> xuser_lists        NONE
>> start_proc_args    /bin/true
>> stop_proc_args     /bin/true
>> allocation_rule    $fill_up
>> control_slaves     TRUE
>> job_is_first_task  FALSE
>> urgency_slots      min
>> accounting_summary TRUE
> Thanks,
> Noah
> _______________________________________________
> users mailing list
> users@gridengine.org
> https://gridengine.org/mailman/listinfo/users

_______________________________________________
users mailing list
users@gridengine.org
https://gridengine.org/mailman/listinfo/users

Reply via email to