Intel NUC kit NUC7i7DNHE freezes after update to Proxmox 8.0.3

[John245]

Updated my cluster (3 Nodes) to Proxmox 8.0.3.

One of the Nodes is an Intel NUC kit NUC7i7DNHE. This Node dies within a couple of minutes. Than it will restart a number of times and in then it is fully dead.

I updated the BIOS to the latest version but the behavior did not change.

When I pin the kernel to 5.15.107-2-pve the system functions properly.

How can I solve this issue.

And what is the impact of running the 5.15.x kernel on one (1) Node while the other Nodes are on 6.2.16-3.

Will I still receive kernel updates?

---
John

 

[John245]

@t.lamprecht

Adding a HDMI dummy solves the issue so it seems a kernel issue.

 

[t.lamprecht]

And what is the impact of running the 5.15.x kernel on one (1) Node while the other Nodes are on 6.2.16-3.

It might have some impact on live-migration, but depends a lot on how different those two nodes are, e.g., in terms of CPU model.

Will I still receive kernel updates?

Not for 5.15 on Proxmox VE 8, there 6.2 will get upgrades.

Adding a HDMI dummy solves the issue so it seems a kernel issue.

You mean plugging an HDMI cable without a monitor connected?

Without a specific (kernel) log error it's really fishing in the dark, so maybe check older journactl entries for errors.

If nothing is found there maybe you could try to reproduce it again while connecting from another host via SSH and run journalctl -f to extract log messages even if the NUC is already going down.

 

[beisser]

hdmi dummy is a plug that emulates a display. basically a fake display. some computers need this to boot. havent seen that one any of my recent harwdare though.

something like this: https://www.amazon.de/Headless-Emulator-Passform-headless-1920-1080-3840/dp/B06XT1Z9TF

 

[stuartbh]

Beisser,

I was using an HDMI dummy plug on my NUCs for a while but they were replaced with a PiKVMs, so that works far better for me now!

Stuart