Tuesday, July 2, 2013

Renaming Lync Servers

No matter how well you discuss it with other stakeholders in advance, how many times you show it in design documents, and how many change requests you submit with it prominently displayed, you may find yourself having to rename Lync servers in a deployment.

I recently found myself in this unenviable position a few months after I built out a brand-spanking new Lync 2013 deployment for a customer. The server naming convention I proposed was accepted by all the important parties and was prominently displayed in design documents, presentations and change request orders.  Well into the deployment, an issue with the naming convention was brought up by someone important enough to force the server name change after the fact, but apparently not important enough to be included in the design process in the first place. Sigh...

At this stage, there were 3 Lync Server 2013 Enterprise Edition front-end servers, 2 edge servers, and a survivable branch server (SBS) all running Windows Server 2008 R2. Thankfully, they didn't have a problem with the pool names for the front-end and edge pools, so we didn't have to do a complete rip-and-replace, but that was minor consolation for the work ahead.

I managed to come up with a fairly streamlined process to do the renames, with minimal interruption to the users already piloting the deployment. The final result was a deployment with no errors relating to the name change.

I'm not saying there's not a faster way of doing this, by removing fewer components, but this process proved to work for me without issue.  If you've done the same thing by doing less, let me know in the comments.

Do each server one at a time, running through the complete process before starting the next one.

  1. Remove Lync server from topology
  2. Publish topology.
  3. Run Lync Server Deployment Wizard local setup on server to remove Lync components
  4. Uninstall SQL Server. Front-ends have LyncLocal and RTCLocal instances. Remove both, rebooting between instance removal.  Edge only has RTCLocal instance. 
  5. Remove SQL Server 2012 Management Objects (x64)
  6. Remove SQL Server 2012 Native Client (x64)
  7. Remove Microsoft System CLR Types for SQL Server 2012 (x64)
  8. Remove Microsoft Lync Server 2013, Front End Server
  9. Remove Microsoft Lync Server 2013, Core Components
  10. Delete C:\Program Files\Microsoft SQL Server
  11. Delete C:\Program Files\Microsoft Lync Server 2013
  12. Delete C:\CSData
  13. Rename server and restart
  14. Wait until AD replication completes with new server name.
  15. Open Topology Builder, add new server to existing pool and publish.
  16. Reinstall Lync components and all cumulative updates
  17. Generate new certificate with updated server name and assign to appropriate services using Lync Server Deployment Wizard.
  18. Restart all servers in pool at same time (only relevant for front-end servers in an Enterprise pool).




5 comments:

  1. Thanks Ken. This is useful. I just really hope I never have to do it.

    ReplyDelete
  2. yea no kidding, that's a though process and a lot of steps.
    There might also be some good stuff regarding this over at TechNet and the A B C restore http://technet.microsoft.com/en-us/library/jj945634.aspx

    ReplyDelete
  3. I think you could have saved some time with bootstrapper.exe /scorch in this case. It will handle removing all Lync components from the local server.

    ReplyDelete
  4. Even though FQDNs of the Front-End servers in the Lync Server Enterprise Pool should not be directly accessed (with the individual FQDNs), rather through the VIP of the pool itself via a Hardware Load Balancer, renaming the pool members' names to adhere to standard naming conventions goes a long way in simplifying administration, troubleshooting and support.

    Ken's post does a great job to cover the essential steps that a Lync administror should know and do in order to undertake such an important tasks in a production environment.

    http://www.leedesmond.com/

    ReplyDelete