0

I have several servers from Hetzner AX and EX series (AMD & Intel), that I installed with Centos 8, then migrated to Centos Stream, but I get a Kernel Panic every time I try to boot with any image from Stream.
Of course, they are not very helpful with suggestions on potential kernel configuration for it to work except claiming that there are no know issues (HA!). As it doesn't even get to logging, I'm pretty much helpless.
I have done 10ish migrations to Stream on various PCs, but I only have problems with HZ servers.

Anyone have an idea about this?

screenshot from the Robot

kenlukas
  • 3,101
  • 2
  • 16
  • 26
Istvan Prosinger
  • 3
  • 2

1 Answers1

0

The error occurs in the amd64_edac_mod module.

You can disable the module by adding it to a module blacklist

IE in /etc/modprobe.d/edac_blacklist.conf

blacklist amd64_edac_mod

Then attempt to reboot and see if the problem re-occurs.

This disables machine check exception support for ECC memory and might prevent knowing if an ECC error has occurred.

Given this occurs on multiple hetzner hosts, its probably a software bug to do with the motherboard or memory control unit on the board.

Newer kernels might actually fix it, but I suspect blacklisting the module should prevent the panics.

Any hosts still online that have loaded but not died, you might be able to modprobe -r the module to prevent issues there too but I'd suggest testing that first on a non-important host.

Matthew Ife
  • 23,357
  • 3
  • 55
  • 72
  • Well I can boot the hosts with the old Centos 8.x kernel. This does make sense, despite that I have implied that one server is an AX (AMD) and one is EX (Intel) - but I made a mistake, both that go panic are AMD. I'll try this and report back. Thanks a lot – Istvan Prosinger Apr 07 '22 at 14:15
  • That worked, thanks. This is very inportnat info. I would applaude you and buy you a beer if I could. – Istvan Prosinger Apr 07 '22 at 18:54