proxmox ve 6.3.6 dies with kernel panic once a day

[Arnd]

Hi,

i am a new to this proxmox and I am using an old desktop pc.
My proxmox ve dies nearly once a day. so i connected a monitor an take a screenshot by mobile. How can I reserch deeper? Is this a hardware problem? Is this a configuration error? Is this a virtual machine problem?
Are there any ideas?

regards
Arnd

 

[Ramalama]

blk_mq has something todo with storage sleepstate or sth.
https://android.googlesource.com/ke...sm-triton-4.9-pie-wear-mr1/block/blk-mq-tag.c

my knowledge doesn't goes that deep, but probably if you use ssd's, you have a problem with lowest sleepstate or something like that.

This is a 100% guess.
however, it's the only idea i have, if you use an ssd/nvme etc, you could disable the lowest sleepstate via a grub cmdline edit. But again, that's only a guess.
Google about your error a bit, maybe you find something more concrete.

cheers

 

[Jeff Wadsworth]

It would be great to see your hardware for the hypervisor as well. Is this your computer? https://origin-www.asus.com/Tower-PCs/M51AC/

 

[Arnd]

@Jeff: The hardware is the M51AC, with i7 4770s, 24 GB DDR3 12800 RAM (2x4GB+2x8GB), Samsung EVO SSD SATA 500GB, onboard grafic (fancy GTX 650 removed for energy saving -> I hope greta will be thankful)

@Ramala: Found a suitible switch in BIOS: Aggressive LPM (Link Power Management). It was enabled. I switched it off now (disabled).

 

[Arnd]

It keeps crashing. But the screen looks different. Is this the same problem? I was able to get a stable system on an alternative hardware. But this system keeps crashing and I cannot trust this half of my cluster. But I can't read this linux-bluescreen. How do I get hints, whats going bad with this system? Is there a method to get additional information from inside the server besides this mobile-picture?

 

[0xd149e38e]

I would check if the RAM layout is OK: DIMM vs Channels, or just use 2 DIMMs insteat of 4.
Maybe run a memcheck on it.

 

[Arnd]

memcheck was happy with it. I thought memory is a bit more binary and not ... agile (aka "hmm, lets give it a try"). But I will remove 2 DIMM and wait for the next "bluescreen".

 

[0xd149e38e]

Random crashes are exactly what happens when you got mild memory problems.
I had this on my desktop with very slightly overclocked RAM. This gave me 1 - 2 freezes a day, as well as crash with stack trace.