[SOLVED] Some VMs in any node stop responding to console on web gui

[Gusimelo]

Hi all,

I have a 4 node cluster running the latest 8.4.5 pve, with diverse, mostly old, Intel CPUs:

16 x Intel(R) Xeon(R) CPU E5620 @ 2.40GHz (2 Sockets)
24 x Intel(R) Xeon(R) CPU E5645 @ 2.40GHz (2 Sockets)
32 x Intel(R) Xeon(R) CPU E5-2640 v2 @ 2.00GHz (2 Sockets)
64 x Intel(R) Xeon(R) Gold 6226R CPU @ 2.90GHz (2 Sockets)

All runninng Linux 6.8.12-12-pve (2025-07-14T13:20Z)

Some time ago a strange issue started to happen, I think since pve 8.4.1

Some VMs, mostly almalinux 9, in any node, start showing all of these symptoms:

- After some time running, the console on proxmox web gui for that VM stops responding. The cursor is not blinking, and doesn't respond to key presses, just shows the login prompt.
- I can't reboot or halt the machine, just STOP or RESET through the web gui.
- I can't connect using spice, returns a "can't connect" error

The machines are running fine, and can login through ssh, but can't reboot or halt.
If I reboot, I loose the ssh connection, but the machine never reboots or stops, or at least I can't reconnect to it.

Thank you for any Help!

Gustavo Melo

 

[Gusimelo]

Update:

This only happens to VMs with Display explicitly set to spice (qxl) and running Almalinux 9.
With Display on Default none of the VMs running almalinux 9 show the symptons.

I changed the display type, and will get back with the results

 

[Gusimelo]

Update:

This only happens to VMs with Display explicitly set to spice (qxl) and running Almalinux 9.
With Display on Default none of the VMs running almalinux 9 show the symptoms.

I changed the display type to Default, and will get back with the results

EDit: confirmed, the problem is the spice (qxl) display

 

[warner_k]

I have the same problem. Is there a solution in sight?
I'm running several VM's with Debian 11,12 and 13 and they all show the same behaviour.
On the latest Proxmox.