Debian 13 guests freezing on reboot

pringlestuffs

Member
Apr 7, 2022
16
7
8
Is anybody else experiencing fairly frequent freezes from Debian 13 vms upon issuing a 'reboot' or 'shutdown -r now' command?

If I'm ssh'd in to the guest upon issuing the command, the ssh session freezes (and eventually experiences 'connection reset by peer') and the console stops responding while displaying the login prompt. Sending ctrl-alt-del at the console is not effective. The guest is still pingable but ssh connections are refused. I've let them sit for 10 or 20 minutes but they never reboot on their own. CPU activity is low. I have to hit the reset button at which point they come back up, usually with some borked inodes during fsck.

This is a very persistent problem for me with Proxmox VE 9 across two hosts, 8th and 9th gen Intel Skylake. Not sure if it's happening on my AMD host as I recently moved a bunch of guests off that one because I'm reorganizing my setup, thus I'm not updating Debian to 13.3 (and experiencing these freezes) on guests on the AMD box today.
 
Hi,
please share the VM configuration qm config ID with the numerical ID as well as the system logs/journal from within the guest around the time of the issue.
 
Is anybody else experiencing fairly frequent freezes from Debian 13 vms upon issuing a 'reboot' or 'shutdown -r now' command?

If I'm ssh'd in to the guest upon issuing the command, the ssh session freezes (and eventually experiences 'connection reset by peer') and the console stops responding while displaying the login prompt. Sending ctrl-alt-del at the console is not effective. The guest is still pingable but ssh connections are refused. I've let them sit for 10 or 20 minutes but they never reboot on their own. CPU activity is low. I have to hit the reset button at which point they come back up, usually with some borked inodes during fsck.

This is a very persistent problem for me with Proxmox VE 9 across two hosts, 8th and 9th gen Intel Skylake. Not sure if it's happening on my AMD host as I recently moved a bunch of guests off that one because I'm reorganizing my setup, thus I'm not updating Debian to 13.3 (and experiencing these freezes) on guests on the AMD box today.
Yes, I experienced the exact same issue in my (two node) homelab. My guess is: This happens every time I reboot a guest while the node is low on RAM. Maybe you can try again with this in mind?
 
Last edited:
Hmm, interesting point. I hadn't noticed that pattern and don't reboot my guests often, which is why I haven't responded to provide logs yet.

I'll have to keep an eye on this... the node where I run the most guests is usually above 80% memory usage.
 
Ok, my guess with low RAM doesn't seem to be a that hot.
Instead I recognized that all affected VMs

a) the consoles are frozen even before I try to reboot the VMs
b) have a lot of "[TTM] Buffer eviction failed" errors in their journals

So I searched for the error mentioned above and found this:

I didn't read this thread completely, but the error seems to be related to the display setting "SPICE".
So I just changed it to Virtio-GPU on all VMs. Next weekend I will check if that changes have an effect.
 
Ha, I just came back to this thread to report the same thing, but you'd beaten me to it :)

So far I don't have much data, but the trend with just a handful of VMs changed to a different display adapter is that they no longer have this behavior.