Am 25.12.2012 um 23:30 schrieb Joseph Farran:

> Thanks for the explanation Reuti.
> 
> I have a lot of new users who keep sending jobs to a queue that has a max of 
> 2-cores per node requesting parallel environment of 32 cores.   So the jobs 
> can never start ( requesting 32 while nodes have 2 ).

The `qrsh` should timeout anyway.


> Is there another setting in Grid Engine to make it so that these types of 
> jobs can never be queued?    Some other kind of verification process?

You can setup a JSV (job submission verifier) and generate a proper error 
message in case you discover any improper request or you could even correct it 
therein.

There is an example in $SGE_ROOT/util/resources/jsv/jsv.sh and man pages "jsv" 
and "jsv_script_interface".

-- Reuti


> On 12/24/2012 8:02 AM, Reuti wrote:
>> Hi,
>> 
>> Am 24.12.2012 um 09:08 schrieb Joseph Farran:
>> 
>> maybe it's by design. From `man qsub` for the "-w" option: "It should also 
>> be noted that load values are not taken into account with the verification 
>> since they are assumed to be too volatile." It's of course funny that "-w p" 
>> will find suitable queues (as all load values are taken into account here 
>> [submit a too high value, and a correct output is created]), while "-w v" 
>> explains at least what is missing for "-w e" to succeed. So it looks like 
>> the mem_free value isn't reported to the verification process as it's a load 
>> value. -- Reuti
>>> _______________________________________________
>>> users mailing list
>>> [email protected]
>>> https://gridengine.org/mailman/listinfo/users
>> 
> 


_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to