[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [condor-users] Can I replace the Condor server in a live pool ?



Hi

This worked just dandy! My WinXP Pro central manager has been swapped for a Windows 2003 Server central manager with the same name/ip. I copied across the recommended accountant logs to keep my user priority.

	Many thanks
		Patrick.

--On 08 October 2003 11:33 -0500 Erik Paulson <epaulson@xxxxxxxxxxx> wrote:

On Wed, Oct 08, 2003 at 08:17:25AM -0500, Alain Roy wrote:

Yes, there will be no problems, you just won't start up new jobs while there isn't a central manager.

> Just by replacing the old server with the new will the nodes - both
> submitting and executing - cope, or do I need to refig them?

As long as the name/ip hasn't changed, you won't have a problem.
Otherwise,  you'll need to edit your configuration file(s) to point at
the new server  and do a condor_reconfig on the nodes.

You will probably notice that it will take about five minute for
"condor_status" to regain all of knowledge it used to have about the
pool,  but it will happen automatically.

> Will running jobs, queued jobs, etc continue as normal, or recover
> when  the matchmaking starts again

Running jobs will continue running. Queued jobs will remain queued. If
jobs  are interrupted during the outage, they will be queued then
restarted when  the central manager returns.

> Or do I need to copy some server data files across to ensure continued
> service?

There is no need to do anything like that.


Actually - there are some files that you may want to copy over - on the central manager, you probably want to save your Accountant.log and Accountantnew.log in the $(SPOOL) directory on your central manager. These are the files where the priority of the all of currently running users are found. There's nothing specific in those files about what central manager they came from, so the easiest way to migrate is to shut down the old CM, copy those files to the new CM, and then turn the new CM.

If you're not worried about user priorities, you don't need to copy them
to the new machine. If you don't copy them, there may be a bit of
preemption in your pool for the first few rounds of matchmaking as the
negoitiator tries to figure out what's going on.

-Erik

-alain


Condor Support Information: http://www.cs.wisc.edu/condor/condor-support/ To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with unsubscribe condor-users <your_email_address>

Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>



-------------------------------------
Patrick Townsend  -  Computer Support
Electrical  &  Electronic Engineering
MVB room 4.27   University of Bristol
www.bris.ac.uk     Tel: 0117 954 5288
email: patrick.townsend@xxxxxxxxxxxxx
Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>