kernel panic not syncing fatal exception in interrupt

[timmjones]

Proxmox keeps crashing and not recovering form kernel panic.
Currently running 5.4.78-2-pve but it has crashed on previous two kernels that I booted it with.
MSI B450 tomahawk max
Ryzen 3600x
gskill ripsaw V 2x16GB(tried ballistic 2x8gb) not ECC
Nvidia gtx 1070(tried with old amd firepro v3900)
1TB nvme m.2 boot drive(inland)(tried with 256GB version too)
I have tried a number of different hardware except for the motherboard and CPU. I have the most updated BIOS installed.
Attached is a picture from screen when it goes unresponsive, but keeps this up on the screen. Since it's a kernel panic I don't think it records it unelss I don't know where to look. I have looked a kernel.log and I can post if it might help.
It seems to only happen when running a KVM. It crashes in a few minutes to a few hours. When running just LXC it seems fine. No ZFS running on machine.
My converted desktop is running into a deal breaking issue. Please help.

 

[Stefan_R]

A fatal kernel exception in a default IRQ like the APIC timer would point to either a firmware (BIOS or microcode) or simply hardware error. There should not be a way for bad software to cause such a scenario.

Maybe try older kernels, or boot a different linux distribution (Ubuntu or what have you), start a VM with KVM there and see if it also happens. Otherwise check your hardware again, also see if all RAM sticks are seated properly and working (maybe run a memtest to confirm).

In general, your hardware configuration should be supported pretty well, I've seen quite some people in this forum using a 3000 series Ryzen now without problems.

 

[timmjones]

Thanks Stefan,
I have tried a lot of different hardware configs(same MB and CPU) and did a memory test. I was hoping that someone had something like, "make sure something is set to this in the BIOS," and it would be fixed. I had issues with ubuntu installing but windows 10 seems to run fine. I also tried 3 different builds of the 5.4 kernel, I guess I could go to different versions as well. I've tried assigning less cores to VMs and it has been more stable but still crashing after quite a few hours instead of the minutes it sometimes takes. I'll keep trying things and try to save up for some new hardware.

 

[Marko68]

Thanks Stefan,
I have tried a lot of different hardware configs(same MB and CPU) and did a memory test. I was hoping that someone had something like, "make sure something is set to this in the BIOS," and it would be fixed. I had issues with ubuntu installing but windows 10 seems to run fine. I also tried 3 different builds of the 5.4 kernel, I guess I could go to different versions as well. I've tried assigning less cores to VMs and it has been more stable but still crashing after quite a few hours instead of the minutes it sometimes takes. I'll keep trying things and try to save up for some new hardware.

Hi Timm,
any luck so far?

Struggling with similar issue on a Ryzen 2600 + 48GB RAM + Asus B450-F + ZFS. Kernel Panic after some hours.

 

[timmjones]

I didn't have any luck with this. I ended up building a Broadwell xeon machine. It was very frustrating and I spent a lot of time without ever figuring it out. Good luck

 

[boopzz]

I seem to be hitting this issue also. I've also got a an MSI B450-F Tomahawk, 32GB RAM, Ryzen 3900X. I was running an LXC with an NVMe as a mountpoint and was hitting issues when copying over to a spinning disk in another VM. I then had a VM with the NVMe and seemed fine to put data on it, reattached the NVMe to a VM to copy over to a disk attached to that VM and it kernel panicked within 5 mins.

I assume its a BIOS issue with the B450 and NVMe as that seems to be the common factor among us? Slightly worrying that an error simply copying files on an LXC or VM would down the host

 

[boopzz]

Ive come across this post https://forum.proxmox.com/threads/kernel-panic-whole-server-crashes-about-every-day.91803 which has some options, I ended up adding the amd64-microcode package and added a line to the drives in storage.cfg, seems to have fixed my issue which was linked to IO on my disks

 

[kinetickudu]

Anyone know where can I check this logs?