VM keeps freezing on PVE 9 kernel 6.14.11-2

[Advin]

Hello,

We recently deployed PVE 9 on some of our hypervisors running AMD Ryzen 9000 series of processors with the B650D4U motherboard. Recently, we started getting reports that 2 VMs on 2 separate nodes are freezing daily.

When the VMs are frozen, they remain in this odd state where the CPU and memory usages are nearly constant (with some small spikes). The IP of the VM is not pingable and the VM is completely frozen in the VNC console.

Based on this thread: https://forum.proxmox.com/threads/vms-hang-or-become-unresponsive.141320/

Here's the output of these commands when the VM is in the frozen/hung state:

echo '{"execute": "qmp_capabilities"}{"execute": "query-block", "arguments": {}}' | socat - /var/run/qemu-server/318845678.qmp | jq
https://fro-images.b-cdn.net/129945e6-b630-4338-8000-5cba2c602660.txt

gdb --batch --ex 't a a bt' -p $(cat /var/run/qemu-server/318845678.pid)
https://fro-images.b-cdn.net/548e1a04-3b82-4c0d-afd4-d3f22ced6ddf.txt

qm status 318845678 --verbose
https://fro-images.b-cdn.net/68f401dc-6483-4209-b768-92cac4ed3c74.txt

qm config 318845678
https://fro-images.b-cdn.net/a6e91f01-a96a-4b76-bea5-24be14db65ca.txt

pveversion -v
https://fro-images.b-cdn.net/fdaa35e5-61b4-4f69-9a4e-0487e7ed0b9e.txt

 

[abobakr]

Hello Advin,
Do you still got this problem? if you still have this issue then please provide pvereport for double checking.
On the command line run:

pvereport > $(hostname)-report.txt

The resulting TXT file can then be copied off the server, for example with WinSCP.

I would like to check the pressure:

root@pve:~# head /proc/pressure/*

Did you observe the same with older kernel? e.g kernel: 6.14.8-2-pve.