Re: [condor-users] flocking between 6.4.7 and 6.6.0

[Zachary Miller <zmiller@xxxxxxxxxxx>]

On Thu, Feb 12, 2004 at 06:18:59PM -0600, David A. Kotz wrote:
> I have a production cluster running Condor 6.4.7 and I'm currently
> building a new cluster with Condor 6.6.0.  I've configured two-way
> flocking between the clusters, and the new cluster is running jobs from
> production.
[snip]
> 2) Are there any known gotchas to worry about in this situation?

yes.  there are a few bugs in the session handling code in 6.4.7.  however,
it is disabled by default in 6.4.X.

most of those bugs have been fixed in 6.6.0, and a couple more in 6.6.1.  the
problem is that 6.6.X will request to have sessions enabled by default.  so
when it talks to a 6.4.7 daemon, you may expose those bugs.

for maximum backwards compatibility, your best bet is to tell 6.6.X not to use
sessions by default.  to do so, add this line to your condor_config file:

  SEC_DEFAULT_NEGOTIATION = OPTIONAL

then your 6.4.X and 6.6.X pools should play nicely together.  however, the 
downside is that you cannot use any strong authentication or crypto since that
requires sessions.


cheers,
-zach

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>