Server crashes, freezes regularly: Bad rss-counter state

[joshkrz]

For the past few months my server keeps crashing and locking up completley every week or so, sometimes more often sometimes less.

 

It requires me to hold the power button to force shutdown, which has eventually led my cache pool to become corrupt.

 

The only two entries shown before the last crash were:

Nov  9 19:37:08 home-server monitor: Stop running nchan processes
Nov  9 19:37:08 home-server kernel: BUG: Bad rss-counter state mm:00000000084043e9 type:MM_SHMEMPAGES val:1

 

This issue seems to coincide with me upgrading from 6.11.5 to 6.12.4 but I cannot be 100% sure.

 

I've looked at this thread but as with the OP hugepages seems to be disabled on my system.

Thanks!

 

Edit: It also appears this person has the exact same issues as me including a corrupt cache pool, it was also posted around the same time I think my problems started.

 

home-server-diagnostics-20231110-1116.zip syslog-127.0.0.1.log

Edited November 10, 2023 by joshkrz
typo, added other examples, added logs, clarity

[joshkrz]

So after some testing by disabling docker for a couple of weeks I deduced that it was docker that was the issue.

 

I followed the instructuions outlined here: https://docs.unraid.net/unraid-os/release-notes/6.12.4/#fix-for-macvlan-call-traces in the section that allows you to keep macvlan enabled.

 

So far I have had no cashes and my Unraid machine is stable once again.

Edited December 14, 2023 by joshkrz