Showing results for 
Search instead for 
Did you mean: 

AMA-CUCM Troubleshooting: Best Practices for Reading Trace Files


CUC 9.1 migration to VM process

I am working on migrating our CUC 8.5 physical cluster to 9.1.1 on VMware, however the documentation doesn't seem to have a clear path for a clustered CUC installation.

I am referring to the following document:

I have upgraded the physical servers to 9.1 per the document, however where the document falls down is how to then migrate to a virtual environment (unless I am missing it).  The document gives directions on how to migrate for a standalone installation, but not a cluster.

Is it possible to remove one physical server from the cluster, then add a new VM to the cluster, switch roles after replication has completed (publisher/subscriber), remove the old physical publisher(now subscriber) and add a new vm subscriber?  Then swap roles again if you want to maintain subscriber/publisher locations?

I would like to keep the existing IP addresses if possible, but changing IP addresses isn't a deal breaker if I have to.

Or is it best practice to break the cluster down to standalone, then follow the document for migrating a physical standalone to virtual standalone?

As seems common in this section, please don't suggest calling the PDI helpdesk as I am not a partner.

Cisco Employee

Re: CUC 9.1 migration to VM process

If you want to keep your IPs the same then DRS is your game. DRS and voicemail can be tricky because of the size of the back and amount of time a back can take. If you are looking for minimum down time, then you would do this:

1) Take a backup of secondary

2) Shutdown secondary

3) Install new Secondary on VM

4) Restore Backup

5) Verify cluster is operational (in sync, not in SBR)

6) Repeat 1-5 for Primary

If downtime is of no concern (maintenence window) then you could do both servers at the same time. But probably best to do one at a time. DRS backups of CUC can be very time consuming though so be patient.

If you ask me, what might be easier is to build a new cluster on VMs and then use the COBRAs tool to migrate the data. You can leave the original cluster up, migrate the user data on your schedule and then cut over the SIP trunks after you have tested and verified the data is in place. The legacy cluster could also still be avaiable to move any user data over that a user thinks they might be missing. This would, of course, involve new IPs, but if its not a deal breaker, its probably the easiest way to go.

For information on COBRAS  please visit:

CreatePlease to create content
Content for Community-Ad
August's Community Spotlight Awards