Plex inside VM causes Proxmox host to freeze

[kprox1]

Hello,

I'm encountering this weird problem where my entire Proxmox host freezes shortly after I start the plexmediaserver Service inside a VM. (It's reproducible)
Sometimes the host freezes after a few minutes, sometimes seconds thereafter. Without the plex service running, everything runs smoothly without crashes whatsoever.

By freezing, I mean that the host becomes completely unreponsive (inlcuding all VMs of course).
Its not responding over the network and when I plug in an HDMI cable to access the console, my monitor does not detect anything.
I tried plugging in the cable before a crash occures (which works) to see whats going on in dmesg, but the screen just becomes green when the system freezes.
Even weirder, the CPU fan is usually continously adjusting its speed (managed by the bios?), but when the freeze occurs, its also "stuck" at the same speed.
So i figure the entire system gets stuck somehow.
Only a hard reset by pressing the power button reboots the system.

I already tried the steps described here (setting aio=none and installing the amd64-microcode on the host), but it didn't help.

System Specs:
GigaByte Brix BRR7H-4800 (rev 1.0) (latest BIOS update is installed)
AMD Ryzen 7 4800U
Proxmox VE 7.0-13
Kernel: Linux brix01 5.11.22-5-pve #1 SMP PVE 5.11.22-10 (Tue, 28 Sep 2021 08:15:41 +0200) x86_64 GNU/Linux

I don't really know how to troubleshoot the problem to figure out what's actually causing the freeze/crash,
but I feel like starting a Service in a VM should not be capable to cause this.

 

[kprox1]

Changing the virtual hard drive's cache to "writeback" seems to have solved the problem.

Described here:
https://forum.proxmox.com/threads/kernel-panic-whole-server-crashes-about-every-day.91803/page-2#post-403130

EDIT: It only worked for about 20minutes until it crashed again :-(

 

[kprox1]

I tried to get information about the Kernel panic (if it is one?) with netconsole as described here:
https://forum.proxmox.com/threads/proxmox-host-random-freeze.54721/post-252772

When I produce a test kernel panic, the reason for the kernel panic is written into the logfile. The kernel panic my host produces when plex is running is however not received by the target host.

 

[fsc]

Hola @kprox1 !
I am having the same problem with GB-BRR7H-4800 (rev. 1.0). Did you get a solution?

 

[wimgoyv]

Hi @kprox1 & @fsc

I am also having the same problem, also with the BRR7H 4800. Did you guys ever find a solution to this?

Thanks!