Bridge Upgrade 7.1.5 to 8.6 Procedure all nodes

[shaftstri]

We have 5 servers and 1 dedicated bridge server .

NOTE: This is a parralel upgrade and the new UCS enviroment will be Isolated.

Do we need to bridge all the CUCM servers one at time using this process below

1. Build 7.1.5 on Bridge same IP and Hostname for Publisher.
2. Restore this from DRS
3. Upgrade to 8.6.2Su2
4. Take Back up
5. Build VM on UCS 8.6.2Su2 Same IP and hostname Retore from Backup.

Do we have to do this ,1,2,3,4,5 process above for the subscribers as well, is this the supported method.

Its going to be a longe process with one bridge server

Thanks

[Joseph Martini · Accepted solution]

No you do not need to do all servers, just the publisher is fine.  The only thing to remember is that you will have to install items on the subscribers that are not in the database which will be replicated from the publisher.  Things you'll want to remember to reinstall on the subscribers since they will be reinstalled on the new version are, MoH files and custom TFTP files such as ring lists, and firmware files.  Also make sure that the subscribers are not reachable from the bridge publisher or you'll run into http://tools.cisco.com/Support/BugToolKit/search/getBugDetails.do?method=fetchBugDetails&bugId=CSCtw75986.

[shaftstri]

Thanks Joe,

So apart from what you mention there are no gotchas

Noting: The Bridge server will be built on a dedicated Bridge HW for publisher node.

Once we have the Publisher running on a VM the what --> we just install the subscribers as VM using OVA and ISO from data store pointing to them to publisher and using the exact same infromation and IP addresses that they had before?

Is this supported  by Cisco the afore mentioned ? as the document suggest all servers had to be restored.

There will be no Data base inconsitancies letting the new subscribers sync rather than restore through DRS

As  for MoH and Custom files we can get them later. Not an Issue

Thanks

[Joseph Martini]

You're correct, there are no other gotchas really other than that one I pointed out.  Once the publisher is on VMware just use the install disc as though you are installing the subscribers and point them to the publisher, as you mentioned with the same IP/hostname/security password.  This is supported, and it actually very commonly done.

[shaftstri]

Thanks Joe

Got a query on just rebuilding subscribers as per below infromation.

When you say it is supported to just rebuild subscribers - can we get offical reference in Cisco document or from the BU

Simply installing the software, giving it the passphrase of the cluster,  and the same IP/hostname does not redeem it as the same node. Until you  do the DRS restore CUCM thinks you're adding a new node to the cluster.  If you were to look in the database without doing a DRS restore you  would see two CTI node IDs and a series of references to a new node in  RIS and other services. So, DRS restore is always a good idea. Another  hint that it is worth doing is that nowhere in the DRS guide does is  suggest you can get by with just reusing the hostname/IP. The proceedure  for replacing a subscriber explicitly says to do a DRS restore

[Joseph Martini]

I should have mentioned that re-installing the subscribers to not run into the above problem would include removing them from System > Server, and then adding them back, so that the node ID's and PKIDs are updated.

[shaftstri]

Joe,

would deleting all the nides be a problem noting one can potentially have 10000's of phones registered to these subscribers.

would you have to add the publisher node to all CUCM groups first so phones had a registration point.

The CUCM groups would need at least one node wouldnt they?