On Thu, 2010-04-29 at 14:51 -0500, Josh Patten wrote:
> Already did those things you recommended with the "First Test System".
> That doesn't "mirror" my current production environment. It only makes
> a separate environment that I can play around in.
> 
> What I need to be able to do is have an exact replica of my 4.0.4
> production environment installed and upgraded to 4.2 on a separate
> server, then dropped in place of the production system for about a
> half hour so I can do extensive testing on a 4.2 system that mimics
> our configuration while not causing basic calling on the redundant
> proxy and all phones connected to it to go down. I just need to make
> sure that If I have the redundant proxy running 4.0.4 and the replica
> of the main production server running 4.2 that I won't see any
> problems when I unplug the replica 4.2 server and put the 4.0.4 main
> production server back in place.
> 
> I know this sounds redundant and unnecessary and quite possibly
> confusing but this is the only way I can think of that will not cause
> any downtime for basic calling to users, and allow me to do FULL 4.2
> testing.

There are at least two tricky things about the test you describe
(upgrading only one of an HA pair and then testing it):

      * Unless you _very_ carefully analyze where the packets went in
        your test calls, the one you upgraded could be badly broken and
        the fact that you have the old one still there in HA mode could
        cause all the calls to work anyway.
      * The upgraded master will try to reconfigure the distributed
        server, but that configuration won't take because the software
        version won't match (but the configuration could be corrupted
        such that on reboot the service won't come up - only bringing
        the original master back and pushing profiles will fix it).

What you really want to do is create an entire HA pair of new systems by
cloning your existing servers onto them with the old software, then
upgrade the clones, then at some safe time swap the clones bringing all
the old servers off line.   Test the clones, and when you've satisfied
yourself that they work, do the upgrade on the originals and swap them
back in.  If you have spare hardware, you can just use the clones and
repurpose the originals; if not, you can use VMs to create the clones
and you end up doing the upgrades twice.


_______________________________________________
sipx-users mailing list [email protected]
List Archive: http://list.sipfoundry.org/archive/sipx-users
Unsubscribe: http://list.sipfoundry.org/mailman/listinfo/sipx-users
sipXecs IP PBX -- http://www.sipfoundry.org/

Reply via email to