Windows KVM guest hangs after short period

[cr08]

I have a 64 bit Windows 7 KVM guest running on my Proxmox host. This guest was originally built last December and has been offline for about the past 8 months. Brought it back up today and have been running issues with it hanging repeatedly and reliably. Initially it was brought back up near stock with no windows updates installed. Started the process of doing those. After about 75% through the updates it completely froze. Hard rebooted the guest, came up to the Configuring Windows screen and froze completely there as well after a short period.

So I decided to reinstall Windows. Left the VM instance intact but booted up the Windows ISO and started that process. About 15% through the file copy process it freezes once more. Tried about 5 times.

VM is set up for virtio disk and net. I have both the 1.126 and 1.141 virtio isos and tried both. 1.126 was the original version used when I built the VM. Host is a bog standard Dell Poweredge T20 with a Xeon E3-1225 v3 and the VM is set for 'host' on the CPU side.

Any ideas? Back during the original operation of the VM it ran fine for about a month before I offlined it. Other than regular updates to the host between then and now, not sure what is going on.

 

[GadgetPig]

I would probably start at the hardware first. Have you tried smartctl long test per drive, and/or run memtest for about 1hour? CPU fan/case fan is clean/working, and not clogged up? What version proxmox you running, and have you tried VM CPU set for KVM64?

 

[cr08]

Drives are fine. I have other LXC guests on this host which have no issues and the host logs aren't reporting anything unusual. Haven't had a chance to do a memtest but may be a good idea. I added some extra RAM a few months back that wasn't officially 'tested' but the system seems ok. Mind you this isn't a super mission critical system and has data that is easily replaced.

I actually went through and just rebuilt the VM from scratch and it seems to be ok. Kept the CPU as 'host' and everything else the same as it was originally. I was waiting to make sure I could do so as I had to try and track down my config file mods (mainly bind mounts and a USB passthrough, the latter which seems to finally have a GUI option which is nice).

I want to say back when I built this VM I was on either VE 4.3 or 4.4. Was up to 5.0 when this arose and I did a quick update to 5.1 to see if it would help resolve it but that didn't change anything.

 

[GadgetPig]

There's been some talk lately about certain Intel CPUs having issues with more recent KVM; See here about a test kernel:

https://forum.proxmox.com/threads/pve-5-1-kvm-broken-on-old-cpus.37666/

This is another post about BSOD with VM on SSD:

https://forum.proxmox.com/threads/blue-screen-with-5-1.37664/#post-185356

What happens if you ran Proxmox backup on the original Win7KVM guest, and then restore it in place, with a different VMID?

 

[cr08]

Sorry for the late reply. Unfortunately I did not keep a copy of the prior VM instance. Just blew it away and started fresh. Wasn't too much of a hassle outside of getting my USB devices attached and doing a few bind mounts. Otherwise I didn't have anything of importance on the VM disk itself.

I did run into a new BSOD issue (unrelated to the one you linked to) which was the classic IRQL_NOT_LESS_OR_EQUAL which seemed to be a daily occurrence. Found another thread referencing it which suggested installing an intel microcode update package on the host and reboot. Did so and the VM's uptime is going on 4 1/2 days now with no issue. My host is running a Haswell based Xeon, btw.

 

[GadgetPig]

Thanks for the update. In case you interested, there are some test 4.10/4.13 kernels posted with updated zfs 0.7.3

https://forum.proxmox.com/threads/zfs-cannot-import-rpool-after-reboot.37611/page-2