[GE users] Requested consumables in dynamic resource quota

reuti reuti at staff.uni-marburg.de
Fri Nov 28 13:29:10 GMT 2008


    [ The following text is in the "UTF-8" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some characters may be displayed incorrectly. ]

Am 28.11.2008 um 08:28 schrieb Henry Leyh:

> On Thu Nov 27 2008, reuti wrote:
>> Hi,
>>
>> Am 27.11.2008 um 17:25 schrieb henry.leyh at ipp.mpg.de:
>>> I am looking for a way to limit the requests of a consumable
>>> resource on a per-slot basis for certain hosts. Something like
>>>
>>>   limit users {*} hosts {@certain_hosts} to my_resource=
>>> $requested_slots*some_value
>>>
>>> where my_resource is the consumable resource and $requested_slots
>>> is the number of slots requested in the -pe statement in qsub.
>>
>> if you reqeust a PE for your job, all limits/resource requests are
>> multiplied automatically by the requested slot count.
>
> If I set up my RQS like this:
>
>   limit users {*} hosts {@certain_hosts} to my_resource=some_value
>
> then it looks like I am allowed to use no more than 'some_value' on  
> one host.
> That is, one slot seizes all of 'some_value' and I can run only in  
> one slot
> per host, no matter how many more slots are free. But what I would  
> like is
> more or less the opposite: that one slot can not request more than
> 'some_value' so that I can run in more than one slot with each  
> taking at most
> 'some_value' (in fact what you are saying: the resource limit  
> should be
> multiplied by the requested number of slots which, however, does  
> not appear to
> be the case).

A limit per slot is not implemented. You can multiply beforehand, e.g.

imit users {*} hosts {@certain_hosts} to virtual_free=800M

and submit with "-pe mpich 2 -l virtual_free=400M" - you should be  
able to get two slots on the machine. I.e. the request is 800M in  
total when run on one  machine.

-- Reuti


> Regards,
> Henry
>
> -- 
> Henry Leyh ------------ Software Development / System Administration
> Max-Planck-Institut für Plasmaphysik / Stellarator Theory Department
>
> /*   http://maps.google.com/maps/mm?t=k&ll=54.0743,13.4238&z=19   */
>
> ------------------------------------------------------
> http://gridengine.sunsource.net/ds/viewMessage.do? 
> dsForumId=38&dsMessageId=90192
>
> To unsubscribe from this discussion, e-mail: [users- 
> unsubscribe at gridengine.sunsource.net].
>

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=90240

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



More information about the gridengine-users mailing list