0

I have several Win10 VMs set up in Hyper-V on Server 2012 R2. The server participates as a node in a failover cluster. The VMs are Roles in the cluster. These are new VMs with fresh installs of Windows 10 Enterprise 1909.

I've observed that when a user uses Remote Desktop (RDP) to connect to one of the VMs, then disconnects the session (by closing the RDP client; Microsoft RDP client on both Windows and MacOS), then attempts to reconnect - the VM spontaneously reboots.

I've scoured the Google and found older forum posts making references to services that no longer exist (or are no longer configurable) and registry keys that also do not exist in my environment. I can not locate any material published by Microsoft relating to this.

My Windows 7 VMs running on the same Hyper-V server do not exhibit this behavior - but I wasn't around when they were built. I don't know what customizations may have been applied to them.

Is this normal behavior? Is there documentation of this behavior? Is it adjustable?

I'm happy to provide any additional information that may be required.

Sam Axe
  • 163
  • 6

1 Answers1

0

Turns out this is a change in the way Win10 1903 handles invalid data from video drivers, such as that produced by the Remote Desktop WDDX driver. The invalid data somehow causes a single CPU core to spike to 100% usage. My VMs we set up as single core machines - so when a new session tried to connect the VM stopped responding and Hyper-V rebooted it.

Microsoft issued a patch in KB4522355. Seems to work fine on Win10 1909.

Sam Axe
  • 163
  • 6