Under what circumstances Proxmox will completely freeze with a full blank green screen in local console?

habitats-tech

habitats-tech

Member
Proxmox Subscriber
Jan 19, 2022
34
13
8
Dubai
portal.habitats.tech
I have an issue with a test Proxmox 7.2-7 no-subscription system. The system would freeze every few hours, displaying a blank green screen on local console with nothing recorded in the syslog. The system just dies with no reported errors.

I am still testing but the problem seems to have something to do with the Eclipse MQTT broker Mosquitto. It seems every time I bring a Mosquitto instance up in a CT or VM or VM Docker, Proxmox after a few hours, usually three, will freeze. As soon as I stop the Mosquitto CTs/VMs Proxmox has no issues.

As I have spent countless hours trying to troubleshoot, I was wondering if any of the Proxmox staff or anyone else can assist me home on what could be causing such behaviour.
 
done that recently but no crash yet as I changed things around trying to understand where the issue is. Will update as soon as I have something concrete.
 
You could enable SysReq in order to crash your system manually if the "green screen freeze" occurs, that way, you'll have a crashdump that can be analysed further.
 
have tried SysRq c but does not produce a green screen, just crashes the system
 
when I get the green screen the system is completely unresponsive either locally or remotely; not sure if you mean something else I do not understand though.
 
FYI. I recently had the same green screen problem and I found my own fix.
The problem I have was, proxmox, and all services running in any CT and VM, went completely unresponsive to web requests, and the hardware that ran proxmox outputted a pure green screen. It happened sometimes when VMs and CTs were under load, sometime when there was no load, or even sometimes right after proxmox bootup.
My root cause was that, I attached a external NAS to proxmox and used it as a proxmox storage for VM disks, CT disks, and images because it was a lot larger than the internal drive. That NAS was a old product so it offered low performance, like 15MB/s, and only SMB2.0 for protocol.
As soon as I completely removed this NAS from proxmox and moved all disks to the internal drive, the green screen problem never returned.
Hope it could help you fix your problem.
Scratch all above. My fix failed today. I had the green screen again today. I guess the problem is not my old NAS but a VM in proxmox that runs OpenMediaVault as a NAS. I will continue my test, and come back to update after it is REALLY fixed.
 
Last edited:
After a long time struggling with the issue of randomly Proxmox becoming unresponsive and only a hard reset would resurrect it, it has been just short of a month with no crashes. About two months ago I updated to kernel 6.x, which helped but did not solve the issue.

I then deleted the following additional software, which was running under the native host setup, and moved them to VMs.

- Cloudflare
- Cockpit
- Netdata
- Tailscale

Not 100% certain which of the above in combination or not with the kernel upgrade (from 5.x to 6.x) might have fixed the issue. My bet would be either Cockpit and/or Tailscale were the culprits, and/or possibly recent updates to kernel 6.x.

I was going to wait until mid Mar prior to updating this post, however I wanted to reply to thllxb for his kind consideration, with what I believe worked for me; fingers crossed! My issue is definitely related to the AMD Ryzen 9 4900H system as I have other AMD and Intel systems with different chipsets and they all work flawlessly.

This is what I am running on at the moment on this AMD Ryzen 9 4900H system: Linux 6.1.10-1-pve #1 SMP PREEMPT_DYNAMIC PVE 6.1.10-1

My other systems are all on 5.x kernel.
 
  • Like
Reactions: mow
habitats-tech said:
After a long time struggling with the issue of randomly Proxmox becoming unresponsive and only a hard reset would resurrect it, it has been just short of a month with no crashes. About two months ago I updated to kernel 6.x, which helped but did not solve the issue.

I then deleted the following additional software, which was running under the native host setup, and moved them to VMs.

- Cloudflare
- Cockpit
- Netdata
- Tailscale

Not 100% certain which of the above in combination or not with the kernel upgrade (from 5.x to 6.x) might have fixed the issue. My bet would be either Cockpit and/or Tailscale were the culprits, and/or possibly recent updates to kernel 6.x.

I was going to wait until mid Mar prior to updating this post, however I wanted to reply to thllxb for his kind consideration, with what I believe worked for me; fingers crossed! My issue is definitely related to the AMD Ryzen 9 4900H system as I have other AMD and Intel systems with different chipsets and they all work flawlessly.

This is what I am running on at the moment on this AMD Ryzen 9 4900H system: Linux 6.1.10-1-pve #1 SMP PREEMPT_DYNAMIC PVE 6.1.10-1

My other systems are all on 5.x kernel.
Click to expand...
Has the issue been already solved?

I have exact the same CPU and discovered the green screen crash lastly and a bunch of system crashes after upgrading to 6.x Kernel some months ago. With Kernel >= 5.14 all was fine.

AMD Ryzen 9 4900H Linux 6.2.16-19-pve
 
You must log in or register to reply here.
Top Bottom