Since the release of Horizon version 2006 it’s possible to upgrade your Connection Servers in parallel. Instead of upgrading them one by one, you can now upgrade all Connection Servers (up to a maximum of three pods) simultaneously. Which can save you a huge amount of (down)time!
In this blog post I’ll quickly share my experience and show you how the process works.
Table of Contents
Prerequisites
- Verify there are no issues with Horizon LDAP replication
- To check the replication status of the local LDAP instance, run the following command:
repadmin.exe /showrepl localhost:389
- To check the replication status of the global LDAP instance in a Cloud Pod Architecture, run:
repadmin.exe /showrepl localhost:22389
- To check the replication status of the local LDAP instance, run the following command:
- Upgrade one Connection Server manually to determine there are no issues with Horizon LDAP replication. After you resolve any errors during the upgrade process, you can proceed to upgrade multiple Connection Servers in parallel.
- Bring up all the LDAP nodes in the Connection Server cluster before the upgrade. This ensures that the schema master node is available on the cluster.
- All other prerequisites for upgrading a Connection Server still apply, as described here.
Upgrading the first Connection Server
Before upgrading the first Connection Server I made sure it was disabled from the Horizon Console and my load balancer.
The upgrade of a Connection Server itself is fairly straight forward, as you can tell from the screenshots below
(click to enlarge)
Upgrading the remaining Connection Servers in parallel
Before starting the parallel upgrades, I verified if LDAP replication was still successful. As this was the case I could now disable the remaining Connection Servers and start the upgrade(s).
The upgrade process of the remaining Connection Servers remains the same, the only difference is we can now run them simultaneously.
(click to enlarge)
Conclusion
I wanted to test the parallel upgrading ever since I saw this mentioned in the release notes of version 2006. It’s a great way to save time, especially when running a large(r) amount of Connection Servers across multiple pods.
After upgrading the first Connection Server it took me approximately 15 minutes to upgrade the remaining Connection Servers. With this feature I can see the duration of maintenance windows being reduced drastically.
I would however love to see the upgrade process moved to a central location, as you see with other VMware products, to make the process even easier.