[cisco-voip] Questions regarding migration of CUCM 7.1.5.31900-3 to a UCS platform

[Matthew Saskin]

Fastest way to do it (per cluster) in my opinion:

1 - Acquire spare MCS server
2 - Implement change freeze on CUCM cluster in question, start cataloging
"emergency changes" that are made
3 - Perform DRS backup of Publisher, DRS restore to spare MCS server
4 - Upgrade spare MCS server 7x to 8x
5 - DRS backup of spare MCS server
6 - DRS restore of 8x backup to MCS server
7 - Rebuild subscribers from scratch on UCS
8 - Migrate IP phones from 7x cluster to 8x cluster on UCS
9 - remove change freeze, re-implement any emergency changes that were
required

There's some significant effort involved, in particular on the subscriber
rebuilds, ensuring any custom TFTP files (wallpapers, etc.) are in place.
That said, it's less work overall than doing a double upgrade.

Matthew Saskin
msaskin at gmail.com
203-253-9571



On Thu, Mar 31, 2011 at 2:01 PM, Peter Slow <peter.slow at gmail.com> wrote:

> Hello Everyone,
>
> I'd like your input on a project I'm working on. I have five clusters,
> all of which are currently running 7.1.5.31900-3. My goal is to
> upgrade each of them to 8.5 running in/on UCS.
>
> My _understanding_ is that I need to get a cluster to at least 8.0(2c)
> (which is what actual version, btw?), do a DRS backup, install the
> same version in UCS and then restore the backup into the VM.
>
> I was under the impression (until i read all of
> http://tinyurl.com/5nuf25 ) that on most I2 servers a Bridged upgrade
> is required and this would present an issue, because of time
> constraints - The smallest cluster is 9 servers, the largest is 17 or
> so and it can take upwards of 10 hours to upgrade a cluster, and about
> that long  to actually get everything switched over _after_ upgrading
> is complete. (I realize this doesn't need to be a serial process, but
> it mostly is at the moment, aside from subs having their versions
> switched in pairs.)
>
> The main goal here is to get to UCS, and the team I'm working with
> REALLY doesn't want to upgrade the MCS servers one weekend, then move
> to UCS servers a week later. (Thus far, I've been telling them that
> this is going to be necessary AND that there is NO way to install
> 7.1(5) in VMWare and _then_ upgrade to 8.x) Right now my only
> alternative _seems_ to be getting a CUCM in the lab running 7.1(5),
> doing a DRS export to that form the production cluster, upgrading
> that, then doing a DRS backup from _that_ to the UCS cluster in order
> to get the pre-staged environment up.
>
> There's some additional background below.
>
> Constraints:
> The first cluster to be upgraded is composed exclusively of 7845-I2
> IBM servers.
> In some cases clusters might have  7845-I1's functioning as TFTP or
> MoH servers, I am allowed to replace these with I2's prior to the
> upgrade if I need to.
> 76 GB drives and 4 GB RAM in the I2s
> The upgrades / reboots etc need to take place during a weekend outage
> window that provides me with enough time to roll back if something
> awful happens. (I'd like to get everything done in a day, so that I
> can spend Sunday troubleshooting if I need to.)
>
> Questions:
> * If a bridged upgrade is required, do _all_ servers in the cluster
> need to be upgraded, or can I just upgrade the publisher, export with
> DRS and proceed from there?
> * Can I run 8.5 on my existing cluster of 7845-I2's w/ 76 gig drives
> and 4 gigs of RAM? http://tinyurl.com/5nuf25 seems to indicate that I
> can, but if it's wrong, I get fired.
> * Am I correct about an upgrade from 7.1(5) to 8.x within VMWare not
> being supported / possible?
>
> -Pete
>
> Peter I. Slow
> AIM: humboldtbgp
> Email: peter.slow at gmail.com
> _______________________________________________
> cisco-voip mailing list
> cisco-voip at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-voip
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20110331/113cbc84/attachment.html>