Re: [Condor-users] Mixed 6.8.6 and 6.9.5 installation

["Peter Myerscough-Jackopson" <peter.myerscough-jackopson@xxxxxxxxxx>]

Alternatively you can create a DAG file with all the jobs in an tell
dagman to limit the number of idle jobs. This will allow your cluster to
be fully loaded, but to keep the queue length down and manageable for
6.8.6 


Dr Peter Myerscough-Jackopson 
Senior Engineer, MAC Ltd 

phone: +44 (0) 23 8076 7808  fax: +44 (0) 23 8076 0602 
email: peter.myerscough-jackopson@xxxxxxxxxx  web: www.macltd.com 

Multiple Access Communications Limited is a company registered in
England at Delta House, Southampton Science Park, Southampton,
SO16 7NS, United Kingdom with Company Number 1979185


-----Original Message-----
From: condor-users-bounces@xxxxxxxxxxx
[mailto:condor-users-bounces@xxxxxxxxxxx] On Behalf Of Dan Bradley
Sent: 04 December 2007 16:42
To: Condor-Users Mail List
Subject: Re: [Condor-users] Mixed 6.8.6 and 6.9.5 installation


Yes, you should be able to run 6.9.5 on your submit node and/or central
manager while running 6.8.6 on some/all execute nodes.

If you want to upgrade without killing any running jobs, it is possible,
at least for vanilla universe configured with a long enough
JobLeaseDuration to let you get the upgrade done.  However, if that is
not hugely important, it is simpler to just shut down your submit node
normally (which will stop all jobs in 6.8.6), do the upgrade, and then
start it back up.  I'd recommend upgrading the central manager first and
then doing the submit node once the new central manager daemons are up
and running.

--Dan

Robert E. Parrott wrote:

>Hi folks,
>
>We have a cluster of 240 processors, managed by condor v6.8.6, and have

>some users who would like to schedule 1000 jobs at once. At full 
>throttle I can imagine seeing 5000 jobs in under condor_q if we let the

>users run free.
>
>I understand that schedd and other daemons are single threaded, and 
>with that many jobs can show significantly reduced performance, but 
>that in v6.9.5 this has been refactored, which is a significant 
>attraction to the development series.
>
>My question is how one might make use of the updated condor daemons on 
>the frontend and submit machine, without having to upgrade all the 
>compute nodes. Would this be possible?  What would be needed to do so 
>in a graceful manner?
>
>thanks,
>rob
>
>_______________________________________________
>Condor-users mailing list
>To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with

>a
>subject: Unsubscribe
>You can also unsubscribe by visiting
>https://lists.cs.wisc.edu/mailman/listinfo/condor-users
>
>The archives can be found at: 
>https://lists.cs.wisc.edu/archive/condor-users/
>  
>
_______________________________________________
Condor-users mailing list
To unsubscribe, send a message to condor-users-request@xxxxxxxxxxx with
a
subject: Unsubscribe
You can also unsubscribe by visiting
https://lists.cs.wisc.edu/mailman/listinfo/condor-users

The archives can be found at: 
https://lists.cs.wisc.edu/archive/condor-users/