[torqueusers] Difference Between Nodes and Nodesct

Si Hammond simon.hammond at gmail.com
Tue Feb 24 13:17:29 MST 2009


Hi guys,

I'm sorry to revisit this issue (I'm sure its driving you nuts) but  
can you explain how I could do the following...

I have three queues - small, medium and large and I want to make only  
jobs with say less than 8 processors go into small, 9-15 medium and  
16+ in large. I have set the nodect setting accordingly but this means  
that only 8 processors in the whole of small can run.

What I need is a way to apply limits to jobs only and not to queues as  
a whole. So, I'd like to run multiple jobs from small with each job  
being forced to use 8 processors or less but the queue as a whole able  
to use as many processors as are available.

Thanks for your help.



S.


On 23 Feb 2009, at 21:52, Garrick Staples wrote:

> On Mon, Feb 23, 2009 at 09:48:36PM +0000, Si Hammond alleged:
>>
>> On 23 Feb 2009, at 21:43, Garrick Staples wrote:
>>
>>> On Mon, Feb 23, 2009 at 08:45:44PM +0000, Si Hammond alleged:
>>>>>
>>>> Hi everyone,
>>>>
>>>> Firstly thanks for all of the help in various emails which have  
>>>> gone
>>>> through the mailing list (its been very useful).
>>>>
>>>> Can I ask what the difference is between the following two  
>>>> statements
>>>> please (I'm still not exactly clear on what the effect of these
>>>> is!) -
>>>>
>>>> set queue std resources_max.nodect = 64
>>>> set queue std resources_max.nodes = 32
>>>
>>> nodes is a string, max/min for a string don't mean anything.
>>
>> So this is a redundant configuration setting?
>
> Yes, nodes is only useful with resources_default to apply a default  
> resource to
> jobs that don't request the nodes resources.  Since it is a string,  
> it can be
> in the longer, complex form as defined in the pbs_resources manpage.
>
>
> -- 
> Garrick Staples, GNU/Linux HPCC SysAdmin
> University of Southern California
>
> See the Prop 8 Dishonor Roll at http:// 
> www.californiansagainsthate.com/
>
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers

Si Hammond

Performance Modelling, Analysis and Optimisation Team
High Performance Systems Group
Department of Computer Science
University of Warwick, CV4 7AL, UK







More information about the torqueusers mailing list