Re: [HTCondor-users] Upgrading from 7.5.5

[Brian Bockelman <bbockelm@xxxxxxxxxxx>]

Hi Szabolcs,

Wow, that's an old version!

In general, the HTCondor team keeps around compatibility fixes for what feels like forever.  So, there's a possibility this will work.  With certainty, you can at least test the upgrade before moving over the production hosts.

Here's what I would do:
1) Upgrade a single worker node and only allow it to accept jobs from a test user.  See if this works.
2) If it does, do the upgrades across the rest of the worker nodes.  You can drain them one-at-a-time (or a rack at a time, if necessary) to avoid impacting users.
3) Setup a test schedd on 7.5.5 and practice the schedd upgrade from 8.0.x.  In current versions, this should not disrupt running jobs (shadows should restart and reconnect).  I'm not sure if that extends to an upgrade from 7.5.5.
4) If the test upgrade goes well, do the upgrade on the production schedd(s).
5) Finally, upgrade the collector and negotiator.
6) Review all the configuration knobs you use.  Some of them are likely no longer necessary and some of them are likely deprecated.

For 8.0.0, HTCondor did a 'list of compatibility gotchas' for upgrades; however, I don't think they did a list for prior releases.

Best of luck,

Brian

On Mar 18, 2014, at 7:35 AM, Szabolcs Horvátth <szabolcs@xxxxxxxxxxxxx> wrote:

> Hi,
> 
> We have a Condor pool that still uses an old 7.5.5 version for all deamons. I'd like to upgrade the pool to a more up to date version, but it seems that we won't have the time to stop submitting new jobs, wait for the queue to drain and simply switch versions, so we'll have to do it by keeping the content of the queue valid if its possible. I have the following questions:
> - Is it safer to switch to the stable branch for this?
> - Are there any known issues that make the switch impossible? Are there any significant changes that make jobs or dagman stop working with old job submit files?
> - Can an older scheduler work with a 8.x startd? Can I first upgrade the worker machines and upgrade the scheduler/negotiator later? Or is it more safe to upgrade both of them at the same time?
> 
> Thanks in advance for any help; it looks like a delicate issue and any help is much appreciated!
> 
> Cheers,
> Szabolcs
> _______________________________________________
> HTCondor-users mailing list
> To unsubscribe, send a message to htcondor-users-request@xxxxxxxxxxx with a
> subject: Unsubscribe
> You can also unsubscribe by visiting
> https://lists.cs.wisc.edu/mailman/listinfo/htcondor-users
> 
> The archives can be found at:
> https://lists.cs.wisc.edu/archive/htcondor-users/