Extended Unstun Times with VVOLs and Veeam Proxy Fixed in 9.5 Update 3

Recently Veeam released Veeam Backup and Replication 9.5 Update 3″ This update has brought a number of fixes and additional features that you can read about in Anthony Spiteri’s post VEEAM BACKUP & REPLICATION 9.5 UPDATE 3 – TOP NEW FEATURES
This particular release brings a welcomed fix for backing up VVOL backed VMs when using a proxy server. The symptoms occur when you backup a VM that is utilising VVOL storage and a proxy server with hotadd. The snapshot attempts to remove too soon before the HotAdded disk finishes its unbind process. When this occurs the VM can freeze anywhere from a number of seconds up to 80+ seconds.  These issues were not present when the backup proxy was on the same host as the VM that was backing up. The workaround prior to this release was to run in NBD mode which uses the host as a proxy and is a slower method.
So, what am I looking for? The most obvious symptom is when your VM freezes and can not perform any actions, however performance graphs, etc all should a healthy VM. The other is in your VM log file, you will find a line similar to below. this is a standard line in your log, the difference is the the length of time the process runs for.  In this sample: 56 seconds

Checkpoint_Unstun: vm stopped for 56223314 us

In Veeam B&R 9.5U3, you can now add a registry value to set a wait time to allow the unbind from the proxy to complete before the snapshot is removed. to do this, open up your Veeam B&R server -> Open RegEdit -> navigate to:

HKEY_LOCAL_MACHINE\SOFTWARE\Veeam\Veeam Backup and Replication\

Create a new REG_DWORD: HotaddTimeoutAfterDetachSec
Using decimal set your wait time (value) in seconds for how long you require.
Once added, you can restart your server\services for the settings to take affect. After testing overnight with a few Backup jobs, I re-enabled all jobs to run through proxies and  have not seen any issues yet.