Incremental WVD deployments or redeploying persistent WVD hosts

When you add hosts to a WVD hostpool, and want to do so programmatically, this can be done by seperately deploying new hosts.

Redeploying an existing host manually is easy if it isn’t persistent, but if the disk should be retained, you’ll want to disconnect and reconnect it. This would require a lot of scripting (or third party tools).

As I wanted to keep our WVD solution contained to a single ARM template, some creativity was needed. For example, when deploying a new host that has a newer version of a gallery image, ARM would normally complain that the existing hosts don’t have the correct image version and it can’t change that for you since it is a read-only property (imageReference). Another example would be to move the VM between availability zones.

Linked WVD ARM template solves all these issues, but has a couple of important parameters to be aware of if you’re going to use it;

  1. redeploy: set to true in a scenario where you want to upgrade existing hosts, e.g. from a different VM type, or want to redeploy to another availability Zone. It will use the provided list of disk names to create new VM’s based on these existing disks
  2. existingDiskNames: if redeploy = true, supply the list of disk names of the persist VM’s you just deleted.
  3. vmNumberOfInstances: the number of VM’s you want to have (including existing VM’s)
  4. existingNumberOfInstances: the number of VM’s already present in the hostpool when running this arm template.
    #3 and #4 should be the same if redeploy = true, the template does not support redeploying and expanding the hostpool in the same deployment, this should be done in order.

After redeploying hosts, make sure to reconnect the users as they were assigned before you deleted them.

Download the ARM template here from GIT:

https://gitlab.com/Lieben/assortedFunctions/-/blob/master/ARM%20templates/wvdDynamicHostpool.json

Subscribe
Notify of
guest

This site uses Akismet to reduce spam. Learn how your comment data is processed.

0 Comments
Inline Feedbacks
View all comments