You have to Stop & Start the VM after updating Qemu (in this particular case)
Windows VMs stuck on boot after Proxmox Upgrade to 7.0
- Thread starter lolomat
- Start date
Was that VM freshly booted or live migrated to another node after pve-qemu-kvm 6.2 has been installed?
That is how I see the problem. If the reboot is issued from within the VM, it will hang on either a black screen, the "Guest has not initialized the display (yet)" or for Windows it could also be the endless spinning circle. Stopping it and doing a clean start then works fine.
That would have been a surprise
. The hope is that the VMs will not get into that state with pve-qemu-kvm 6.2.
We have a PVE 7.1-12 running here in our Office which has an Uptime of 150 days now.
It has all Enterprise-Repository-Updates installed, except it has Kernel "Linux 5.13.19-1-pve #1 SMP PVE 5.13.19-3 (Tue, 23 Nov 2021 13:31:19 +0100)" running. Cause we did not do any reboot, only installed every update which was offered.
But this system only has planned Reboots every 180 days. Guess what... every single Windows 10/Windows Server VM and also an BSD12 OS rebooted inside the guest coming up without ANY issues. So maybe there is some correlation with Kernel AND QEMU-Version? Just a guess....
It has all Enterprise-Repository-Updates installed, except it has Kernel "Linux 5.13.19-1-pve #1 SMP PVE 5.13.19-3 (Tue, 23 Nov 2021 13:31:19 +0100)" running. Cause we did not do any reboot, only installed every update which was offered.
But this system only has planned Reboots every 180 days. Guess what... every single Windows 10/Windows Server VM and also an BSD12 OS rebooted inside the guest coming up without ANY issues. So maybe there is some correlation with Kernel AND QEMU-Version? Just a guess....
Last edited:
@tkffaul have you had this happen more than once to each VM? We have not recorded the same VM having the problem more than once. However, we are not 100% confident that is the case.
I havent tried it a second time since the 7.2 upgrade. But prior to 7.2 yes I can replicate this issue over and over, mind you only since moving from 6 to 7. I'll do some more testing offline and let you know what I find.
@tkffaul you could replicate it?! Interesting. We have not been able to replicate it other than to upgrade a 6.x cluster to 7.x.
After the upgrade, all (most?) Windows VMs will hang on the first reboot. After they have had the failed reboot once, it does not appear to happen again to the same VM.
If you have had the same VM hang on reboot, I would really be interested to know what you have done to do that or your setup.
Clearly this is a big issue affecting lots of users. Is there any official word from Proxmox on an upcoming fix or anything? I was hopeful that 7.2 would address this. I think we all desperately want to get this resolved. We had issues last night with 2 clients that resulted in of course early AM support tickets.
Yes it is a big problem. Some time ago we opened some official tickets to Proxmox, but it all faded into a ... "we can't replicate the problem" "try using the 5.15 kernel" "try using qemu 6.2". No deterministic answer.
Yes, it happend again with the latest 05-2022 Windows Updates
I am not sure if it helps but maybe all of you should also follow the reported bug here:
https://bugzilla.proxmox.com/show_bug.cgi?id=3933
Maybe it gets some more "attention" by the proxmox-developers. Currently it is not even assigned to someone.
Maybe also someone from Proxmox-Support can remote into a system to view what is happening when reboot hangs.
We have around 5 to 10 VMs where we can reproduce without any "interruption" for our customers.... they just need to ask.... ;-)
https://bugzilla.proxmox.com/show_bug.cgi?id=3933
Maybe it gets some more "attention" by the proxmox-developers. Currently it is not even assigned to someone.
Maybe also someone from Proxmox-Support can remote into a system to view what is happening when reboot hangs.
We have around 5 to 10 VMs where we can reproduce without any "interruption" for our customers.... they just need to ask.... ;-)
@itNGO how do you reproduce it? We have some VMs too but have not found a way to reproduce the hang.
Will upload a video where I show behavior and make some predictions when this happens....
I can with about 90% chance say which of our VMs will have this next. For me it looks that it is always with migrated HyperV-VMs and when these have at least 4 to 12 days uptime....
Video: https://owncloud.it-ngo.com/index.php/s/atbRmsY3dd3K6Km
Hey @itNGO thanks for the video.
Most of our VMs were built in PVE and have only experienced the hang once each (that we know of).
I just rebooted one that had 49 days of uptime and it rebooted fine. It had been one of the ones that had a problem earlier. Earlier today, rebooted another one with more than three weeks of uptime that had also had an issue.
What is interesting is the VMs we see hanging are always at the spinning dot screen. Yours was on a blank black screen (other people have reported that too).
@itNGO are all the VMs that repeatedly freeze using version 6.0 or later of i440fx?
Code:
# qm showcmd <vmid> --pretty | grep machine
-machine 'type=pc-i440fx-6.1+pve0' \I noticed that in your video. Also, in PVE Roadmap I noticed under 6.4:
- Support pinning a VM to a specific QEMU machine version.
- Automatically pin VMs with Windows as OS type to the current QEMU machine on VM creation. This improves stability and guarantees that the hardware layout can stay the same even with newer QEMU versions.
pc-i440fx-5.1 but yours appears to be on version 6.0 or later.The problem occurs with any version of qemu 6.x I do not use versions 5.x and the problem occurs anyway. On different clusters and with different storage systems, in the same way. Strangely I have never and again never had any problems with Windows 2008r2.