Proxmox freeze without any error

[stich86]

Hi guys,

i've switched all my workload from old ESXi (on NUC 11th) to my new Proxmox 7.1 that is running on Lenovo ThinkCentre M70s (10th i7 with 40GB of RAM).
This is the 3rd time that the server freeze, no video output. The only way to recover it is to power cycle

Looking at /var/log/messages and /var/log/syslog there is no error, just logs that stop to write.

At this time is very difficult to troubleshoot. But I cannot found the root cause.

Now i've upgraded BIOS and PVE to version 7.2 to see if this problem occurs again, but may be I can boot the kernel with some parameter that enables debug

thx in advanced

EDIT: i've enabled netconsole https://pve.proxmox.com/wiki/Kernel_Crash_Trace_Log

 

[stich86]

UPDATE:

I run a memtest86+ for 1 days, no RAM problems.
Also the Lenovo UEFI Diagnostic Tools has run for 18 hours (extended one) and no HW error.

Now i've changed the cmdline adding these parameters "processor.max_cstate=5 intel_idle.max_cstate=5", and disabled memory baloon on all VMs

I'll try to revert all configuration back on the Friday, so I can monitor it for all the weekend

 

[Spoonman2002]

The official Lenovo specs on their website says the unit supports a maximum of 32 GB ram...
(I know you did a memtest, but maybe you can try with 32 GB and see what happens with PVE).

 

[stich86]

i've removed the 8GB stick, doing some tests but still same freeze
looks like a problem when i'm using too much the network

 

[stich86]

i'm losting any hope..

What I've found is when there is an high network usage passing thru OPNSense VM, the server almost of the time freeze.
My first doubt was related to the BCM57810s card. But using it to pass on my internal LAN and OPNSense on ESXi (running on NUC 11th), after a lot of speedtest I cannot freeze the host.

I'll format everything and put ESXi on this machine trying to simulate same configuration to see if it's an HW problem (may be PSU..) or a Proxmox VE issue.

I'm hating this situation, but at the moment there is no other way
Proxmox was a very nice product, but may be is not compatible with my ThinkCentre...