Which versions of Horizon Agent and VMware Tools are you running? I've seen this with older versions or if the VM is starved for CPU/memory.
We are currently using 7.8 in agent and 10.3.2.9925305, be we have see it since version 6.5. Happens with both win7 and win 10 desktop. Our win10 are assigned 2 CPU with 2 Core per Socket.
Did you try disabling the 3D Graphics Acceleration?
Yes, we tried that, but we are using vGPU cards. I happens with it on and off. It looks CPU driven. When we reconnect it spikes the CPU at 100%.
If you are using vGPU then maybe it is a NVIDIA Driver related issue.
I had some issues at Customers with this when Driver Versions did not match or NVIDIA Licensing Server had issues.
Could be worth checking this.
Unfortunately, it happens with the GPU disabled.
Do you use appvolumes, there was a version that caused this, which I think is fixed in one of the newer versions.
Think pre 2.13.2?
- A recalculation operation performed in a Microsoft Excel application (that was installed in base) takes twice the time when an Appstack is attached.
- On upgrading from App Volumes 2.12.1 to 2.13.1, machines created with the same name as machines that were marked as offline in the database prior to upgrade, throw a "virtualization disabled" error at login.
- The PATH environment variable does not retain all the path strings captured during provisioning.
- Writable volume space is getting filled up automatically even when users do install any applications or modify the user profile, leading to low writable disk space.
- Occurrence of blank screen upon disconnect and reconnect of View sessions.
- When there are agent machines online for more than 1 year, the message "Server error" is displayed on opening the Online Entities tab in App Volumes Manager and no Online Entities data is seen.
1 person found this helpful
We have the same issue with one of our Windows 10 Builds. The following resolved it for us.
Try reinstalling everything in the proper order:
- NVIDIA drivers and reboot
- VMware App Volumes agent and reboot
- VMware UEM agent and reboot
- VMware Horizon Agents and reboot
- VMware Tools and reboot
The following order can be used with a clean installation:
- VMware Tools
- VMware Horizon View Agent
- View Agent Direct-Connection
- VMware User Environment (UEM) agent
- VMware App Volumes (Agent)
- NVIDIA drivers
So we fought with this with NVIDIA for quite some time. What we found to help was before pushing your image out to a pool was to create a manual pool and break SSO on the VM so you don't log into it with an AD profile (or leave the gold image unjoined).
Connect to it with Blast and PCoIP and sit at the Windows account screen and make sure it comes up, resize the screen to force it to rescale, and then disconnect. RDP to the VM and do your cleanup/sealing prior to pushing it to a Linked-Clone/Instant Clone pool.
Seems to cut that time from 20/30 seconds to sub 5 seconds.
EDIT: Also you can pre-define the session resolution on your most common resolution/screen size so the VM doesn't need to perform as much resizing when the session starts. More prevalent in PCoIP.