Trying to debug why UNRAID crashes approximately once per month

UNRAID Version: 7.1.4

Processor: Intel Core i7-14700K

Motherboard and Bios:

ASUSTeK COMPUTER INC. Pro WS W680-ACE , Version Rev 1.xx
American Megatrends Inc., Version 4302
BIOS dated: Mon 16 Jun 2025 12:00 AM

Steps previously taken

1. Ran 2 passes of Memtest86+ v6.20 : Pass

2. Removed usage of Intel Graphics SR-IOV

During this last crash which happened on

• Sep 9 between

  • 03:42:02 : last log written to log server

  • 06:43:08 : when I power cycled

The following errors were shown on the monitor of the server

Doing some ChatGPT questioning of lines in my log file. ChatGPT believes the following lines in the log file are the likely root cause of the crash

These lines occurred during mover running (Starts at 3:40:00). I am running a ZFS cache.

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs unmount 'cache_nvme/Downloads' 2>&1

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs destroy 'cache_nvme/Downloads' 2>&1

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs mount 'cache_nvme/Downloads' 2>&1

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs unmount 'cache_nvme/Downloads' 2>&1

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs destroy 'cache_nvme/Downloads' 2>&1

Sep 8 03:40:01 Tower shfs: /usr/sbin/zfs mount 'cache_nvme/Downloads' 2>&1

Is the unmount, destroy, mount in rapid successions normal when running mover on a ZFS cache? I didn't find any other mention of this in other places.

Other items on my list

• Bad USB boot drive

  • I've been using the same drive for many years. I have no other reason to think this but it's easy to try a new one

• Bad CPU

  • I do have one of the known problem Intel CPUs but I have kept up with the newest BIOS and only have ever run stock voltage and frequency.

Thanks for any help you can provide.

tower-diagnostics-20250909-2148.zip

Edited September 10, 2025Sep 10 by mwasserman

4 hours ago, mwasserman said:

Is the unmount, destroy, mount in rapid successions normal when running mover on a ZFS cache?

That's not a problem.

Enable the syslog server and post that after a crash, but plenty of confirmed cases of those CPUs being the problem, so it's a good suspect, especially if there's nothing relevant logged when it crashes.