On Wed, 18 Feb 2015 12:13:04 +0000
Kevin Taylor <[email protected]> wrote:

> 
> I have several groups of machines that have infiniband on them and due to 
> history and physical locations, these groups of machines have individual 
> infiniband domains.
> 
> What I've done right now (not in production) is create a boolean complex for 
> 'ib' and identify all of the nodes that contain infiniband. I've also created 
> a string complex called 'ibdomain' that has a name to uniquely identify which 
> systems connect to each other with IB.
> 
> Is there a way that a user could just ask for 'ib' when submitting a parallel 
> job (I don't care where it goes as long as it has infiniband), and have the 
> grid engine tell the job the value of 'ibdomain'? Or keep the job within 
> systems on the same ibdomain?
> 
> 
What you can do is setup multiple PEs with names in a pattern like ib-a ib-b 
ib-c etc one for each infiniband domain and associated only with queue 
instances on nodes in that domain. The user can then request a wildcard  ib-* 
or, on recent grid engine versions ib-[ab] and grid engine will choose one of 
the matching parallel environments to run in.  If you are wedded to the name ib 
being what the user requests then use a jsv to rewrite the requested PE to a 
wildcard.

-- 
William Hay <[email protected]>

Attachment: pgpRtqSZE6hHK.pgp
Description: PGP signature

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

Reply via email to