I'm hoping that somebody might have an answer to this mystery. We are seeing a random issues across different pools where uses are getting "Devices Getting Ready" message when they log in. Sometimes it wouldn't go past that stage and sometimes it does but then they have no input devices like mouse and keyboard. Sometimes they will reboot their client or disconnect and will get back to that machine with no problems but sometimes they will not be able to log in until that machine is removed. We do use AppVOlumes for appstacks and writable profile for personalization but it does appear to be mostly Horizon or Image related issue.
One thing I will admit to is that the image we currently use is completely not optimized besides couple of basic settings.
Here is an example:
I had one user today with no mouse or keyboard in her session. When I looked at the events she got assigned to her clone at 8:02:45 am. I have looked at the events in vshpere and see that this VM was created at 8:01:37 am so again I believe it wasn't fully available even though agent was ready when she got on it. At that time there should have been 20 VMs available because this is what I have set for up front number of VMs and 5 spares, during the day there is average of 23 sessions. Increasing the up front VMs helped us as a workaround and getting way less calls about the same problem. However, it looks like even though there have been VMS that were available for much longer time she got assigned to the one that got barely built. Is there anyway to force Horizon to use the oldest available VM in the pool? I believe if we were to do that the problem would be solved
Horizon version is 7.13.1 both server and agent, Appvolumes 2103, WIndows 10 SAC 20H2
Let me know if you have a any additional questions,
Is the host hardware the same in each pool? EVC enabled?
EVC is enabled and VM hardware is exactly the same. I can reproduce it on other pools too if I synchronize logon with timing when machine completes customizing and becomes available in Horizon. If I wait for that VM to sit there for a bout a minute and then logon everything seems fine
Were there any changes made to the gold image after which the problem started?
No major changes besides normal monthly patching cycles. FOr now we have increased the minimum of instant clones in the pool and that seems to be working well. We are already in process of prepping for major host upgrade and that wil linclude Horizon agent and nvidia drivers so I'll see if maybe after those updates the issue goes away