the following post is just a short one out of my learnings during my tests with rolling cluster upgrade.
In the first place, I think many of you already noticed the new failover cluster feature. It enables you to migrate clusters deployed on Windows Server 2012 R2 to Windows Server 2016 without building a new cluster and migrating the cluster roles to it. Currently there is only a validation for clusters running Hyper-V and Scale out Fileserver but as soon as I have some more time I will also try to Upgrade some of my Virtual Machine Manager and Fileserver Clusters and report back to you.
The way how you migrate the cluster is already very well documented on technet.
For those of you who are familiar with Active Directory Migrations, the way a Failover Cluster is Upgrade looks very familiar. At first you have three phases like shown in the figure below.
Preparations before you start with the migration.
- Check if your Servers are compatible with Windows Server 2016. Run the new build only on supported environments.
- Ensure that you have always enough compute resources during the whole time upgrade process. Normally you run a cluster with a minimum of n+1 cluster nodes. During the cluster upgrade, I would suggest to add another node to the cluster and run with a minimum of n+2 nodes. That would prevent you from any resource shortages during the upgrade.
In the first Phase with nativ Windows Server 2012 R2 you have the following tasks to perform:
- Run Cluster Aware Update on your Cluster and Update it to the lates patchstate
- Backup your Cluster Database and Cluster Configuration
- Install the first 2016 node, add the server role and failover cluster feature and features like MPIO (if needed). Please note inplace upgrades of nodes are not supported, so please reinstall the nodes.
In the second Phase, you will run in cluster mixed mode:
Please notice that the mixed mode is only supported for 4 weeks and you should get out of it as soon as possible. Anyway, you should take your time to check if the new hosts and the cluster runs stable. As soon as you are on Windows Server 2016 native mode there is no way back.
- Add the first 2016 node to the cluster
- when the node is added properly and runs fine, migrate to cluster role over to the new role
- if the migration runs fine and for example the VMs are working, set the first Windows Server 2012 R2 node in maintenance mode and drain the roles. After that evict the Windows Server 2012 R2 node
- Now you can install the second node and redo the steps 1. to 3. until you have removed all Windows Server 2012 R2 nodes
At this point as, long as you still have one Windows Server 2012 R2 node left in the cluster you can go back if anything goes wrong.
At the end, you have a native Windows Server 2016 cluster node running in functional level Windows Server 2012 R2. Like an active directory with Windows Server 2012 R2 and running on forest function and domain level Windows Server 200 R2 before you raised the level.
Now we enter the third stage, here we need to raise the Cluster Function Level. For that we need to run a powershell command.
So please open the PowerShell Commandline on one of your new cluster nodes as administrator.
Afterwards you can start your backup again and restart the cluster aware update service.
Now the last point, housekeeping. That means, update the virtual machine versions of you VMs and install the new version of the virtual machine management tools or what ever need to be done for the cluster roles.
So that’s all from my site today. I will write a much more detailed post, as soon as Windows Server 2016 reaches RTM.