[GE users] qresub / scheduling of long jobs revisited

Patrice Seyed apseyed at bu.edu
Mon Oct 25 14:40:06 BST 2004


Sorry for the delay in the response.

Reuti - Are you insinuating the use of checkpointing in an effort to go
further down the path of subordinate queues and suspension of jobs?

I would be interesting in such a howto you mentioned.

Currently I am using the concept of express/subordinate queues with
suspension of course. So there are two slots in an express queue on each
host with two slots in the regular queue. The regular queue is subordinate,
but keeping in mind the express queue has a 2 hr hard limit. Also, once a
second express job is running on the same express queue, any regular queue
job is suspended. Once again, by this design you can have 3 jobs at once,
what you described as 3 queues looks cleaner.


-Patrice
-----Original Message-----
From: Reuti [mailto:reuti at staff.uni-marburg.de] 
Sent: Thursday, October 14, 2004 3:00 PM
To: users at gridengine.sunsource.net
Subject: RE: [GE users] qresub / scheduling of long jobs revisited

Hi,
 
> So qresub was suggested as a mechanism for breaking up long jobs that are
> iterative, into a bunch of smaller jobs. 

yes - no! It can be used in such a way. But another application will be the 
usage of checkpointing your job, and using qresub to continue where you
stopped 
during the next run.

I hope to find the time soon to finish my Howto of using the checkpointing 
interfaces and how to build it into your application.

> This was a proposed method for dealing with when a large number of jobs
are
> submitted with a long duration, and keep the job slots busy for a lengthy
> time keeping the slots from becoming available for the other users' job
> waiting in the queue.

For this I would use two different queues. The only problem may be, that 
suspended jobs a still use the resources.
 
Cheers - Reuti

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net


---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list