Windows 11 VM Doesn't Cleanly Shut Down

[dizzydre21]

Hello all,

My Hardware:

Proxmox 8.1.3
Motherboard - Asrock Rack Rome8d-2t
CPU - Epyc 7F72
RAM - 256GB 3200MHZ ECC
GPU - RTX-3080ti - passed through to Windows (tried a 2070 Super and a 2080 Super)
OS Drive - Samsung 980 Pro 500GB
VM OS Drives - ZFS Mirror 2x960GB Samsung P9A3
Windows VM Game Storage - ZFS Mirror 2x1TB WD SN850
HBA Card - LSI-9211-8i - passed through to TrueNAS
TrueNAS Drives - 6x6TB WD Ironwolf
NIC - 82599ES 10Gbe - passed through to TrueNAS


I have had several issues with this and other Windows VMs. I was getting a bunch of GPU/DirectX related BSODs at first. That forum post can be found here:
https://forum.proxmox.com/threads/w...nl-fatal-error-nvidia-gpu-passthrough.139323/

I basically tried so much testing and changing stuff that I don't really know what made the BSODs stop happening, but they did. Updating GPU drivers still seems to break something though and the BSODs will return.

Anyway, I have a backup of the VM that didn't have any BSODs for more than a week and seems stable unless like above I try to update the GPU driver. The issue is that whether I shutdown/restart the VM from within the guest or from Proxmox, it never shuts down correctly. You can see in the Event Viewer that it didn't shut down cleanly with a kernel power error.

What in the world is up with this VM?

 

[BobhWasatch]

Do you have the Guest Agent installed in Windows, and is it enabled in the VM Options?

 

[dizzydre21]

Do you have the Guest Agent installed in Windows, and is it enabled in the VM Options?

Yes, I do. Proxmox can see the Ip address on the summary page and I think it just says Qemu not installed or something if it isn't working.

 

[dizzydre21]

I seem to be answering my own posts here lately.

I believe disabling C-States in the BIOS has resolved this issue. It does use about 20 more watts at idle now, which I kind of dislike, but oh well.

I am no longer getting unexpected shutdown events after quite a few reboots and shutdowns between two different Windows 11 VMs. I have a sneaking suspicion that this may have resolved my BSODs that I mentioned above, but I need more time to tell. I don't think that Windows was shutting down correctly after driver installs or updates. I was also getting a weird issue where my GPU wouldn't clock down below 1755MHz despite being on normal power modes in Windows and Nvidia Control Panel. After using DDU to uninstall the drivers in safe mode, I reinstalled them in normal boot mode, succesfully rebooted, and now it idles at 210 MHz, as it should.

I will mark this as solved after some further testing.

 

[dizzydre21]

Another follow up:

The issues have returned. Every reboot or shutdown throws critical events in the Windows Event Viewer. I still have Global C-States disabled in the BIOS. My GPU is again stuck at 1755MHz and drawing a ton of wattage at idle.