I have a Windows 10 VM that keeps becoming unresponsive. I pass through a Radeon RX 470 and a USB 3.0 card. Sometimes it goes for quite a while without problems but the last few days it's become unresponsive several times and needs to be force stopped. Any ideas?

unraid-diagnostics-20200517-1548.zip

Have you tried i440fx rather than Q35 (as per SIO's video guide)??

10 hours ago, methanoid said:

Have you tried i440fx rather than Q35 (as per SIO's video guide)??

What would be the advantage to this? Wouldn't the Q35 template be more compatible with a UEFI based VM?

 

I've tried upgrading the VM from Q35-3.1 to Q35-4.2 to see if that helps.

7 hours ago, Taddeusz said:

What would be the advantage to this? Wouldn't the Q35 template be more compatible with a UEFI based VM?

 

I've tried upgrading the VM from Q35-3.1 to Q35-4.2 to see if that helps.

I have no idea to be honest.. I have failed to find authoritative text that convinces me anyone really KNOWS which is best for VMs or when to use. I'm having issues myself with a simple Win10 VM (the one that was always easy a year ago when I used unRAID last) but my Ubuntu and Macos VMs are fine...

Well, my change didn't help. I then tried turning disabling the power option to turn off hard disks after inactivity. That also didn't help. Now I've taken out the USB Bluetooth adapter to see if that is causing a problem. I don't really have anything else to try.

 

I can try swapping out USB cards but that will leave things incompatible with my macOS VM.

It almost feels like it's freezing after inactivity. Right now I'm working from home so I can't keep my monitor attached all the time to my VM. I have an HDMI switch to switch between my work laptop and VM. I have the second HDMI port on my monitor hooked to the integrated video on my Unraid server. I may need to figure something else out. I have to keep the Unraid integrated video directly connected to the monitor or else on reboot my BIOS makes the Radeon card the primary video card and won't pass-through properly.

 

The HDMI switch disconnects the monitor from the other devices when not switched to that device. I can tell this because of the way my laptop behaves. It does not duplicate the monitor information across the inactive ports so I wonder if the system doesn't like this state for long periods of time?

I swapped out my USB card for the other one I have that only has 2 ports and that's incompatible with macOS and so far it's been running fine. Been running for nearly 3 days without issue. So I guess I have a USB card that's going bad.

Well, I'm still having problems. If I shut down my Unraid server it seems to work fine until the first time I have to reboot the vm. After that the vm will eventually lock and one of the CPU cores will be stuck at 100%. I've recently switched from using vdisk images for the drives to installing a PCIe NVME drive to pass through. The one thing I haven't tried is to remove and redo my vm template to rule out anything in the UEFI saved settings.

 

I've even tried the AMD reset script to see if that would correct the problem with no success.

Is your VM trying to hibernate or go into standby?  I've had issues with VMs waking back up when they do.

1 minute ago, toastman said:

Is your VM trying to hibernate or go into standby?  I've had issues with VMs waking back up when they do.

No, it's not set to sleep, I can double check that though. I do have it set to sleep the display but that's never been a problem.

@toastman Verified. Sleep and hybernation are set to "never". I've changed the screen to "never".

 

I'm really at a loss with this. There doesn't seem to be any logs to point to the issue.

So apparently I blew out one of the ports on my 2-port USB card trying to use a USB extension cable. It provides power but can't talk to any device. So I had to put the 4-port card back in. I'm thinking things will go fine as my vm hasn't gone unresponsive since setting the screen sleep to "never". Crossing my fingers.

Ok, so finally I've determined that if monitor sleep in Windows is enabled and no monitors are connected the VM will freeze. I've recently changed my setup to have two monitors. One has two HDMI inputs and the other has an HDMI switch. That way the one with two inputs can always be connected to the VM. I've re-enabled monitor sleep and haven't had a lock since.

 

Now my problem is the second monitor's color is different. 🙁

3 hours ago, Taddeusz said:

Now my problem is the second monitor's color is different. 🙁

Spray paint?

I have the same problems. Actually my Windows VM is running clean. But if I don't use it for some time then I can't connect via VNC or Remote Desktop. The VM is not in standby because I can access it via UPnP. Also other services in the network are accessible. Standby and Hibernate are deactivated. Also turning off the monitor is disabled. 

There is no monitor connected to my unRaid server so the Windows VM is running headless.

 

Currently, the VM is only accessible again via VNC or RD after a force stop. Does anyone have any idea how I can solve this problem?

Edited August 11, 20205 yr by Pixelpaule

2 hours ago, Pixelpaule said:

I have the same problems. Actually my Windows VM is running clean. But if I don't use it for some time then I can't connect via VNC or Remote Desktop. The VM is not in standby because I can access it via UPnP. Also other services in the network are accessible. Standby and Hibernate are deactivated. Also turning off the monitor is disabled. 

There is no monitor connected to my unRaid server so the Windows VM is running headless.

 

Currently, the VM is only accessible again via VNC or RD after a force stop. Does anyone have any idea how I can solve this problem?

You should start a new topic and attach your Diagnostics (Tools -> Diagnostics -> attach full zip file) but note to attach one right after you experience the freezing since that would contain the most useful info).

Also elaborate what you mean by "access it via UPnP".

hey, I figured since this thread fits, I'd answer with the same problem. I will open a new thread tomorrow and attach the data.

31 minutes ago, Pixelpaule said:

hey, I figured since this thread fits, I'd answer with the same problem. I will open a new thread tomorrow and attach the data.

May seem similar but my problem is related to the hardware pass through.