I've been dealing with random server hangups for a couple of months now - I primarily use my Unraid server as a media server and for general backups. I have Jellyfin installed and shared access with a buddy who has reached out several times because Jellyfin has become unresponsive while he was in the middle of watching something.

 

If I access the server immediately after he reports this, I can usually get logged in and view a couple things on the Dashboard and Main screens, but when I go to Docker it just hangs up and gets into a perpetual loading loop. The last time it occurred I noticed that the CPU and memory usage on the dashboard were both blank and the CPU usage showed something like "infinite" where the percent utilization normally shows. Usually shortly after this, the connection completely times out and I'm no longer to access the server whatsoever. Even if I go to the physical server and hook it up to a monitor I get no output. This results in having to do a cold reboot to get it running again.

 

I've noticed a couple of errors in the syslog that I think are pointing to the motherboard, but I'm not really sure how to read this to confirm. My parity checks have not found any errors, and I ran Memtest which also did not find any errors on the RAM. Can anyone help me with what my next troubleshooting steps should be?

tower-diagnostics-20231101-0919.zip

Docker image is corrupt, delete and recreate, also enable the syslog server and post that after a crash.

 

I deleted and recreated the Docker image about a month ago, and I've had the syslog server running since around July. Is there a good way to scrub that log of personal info before posting it?

Last week I ended up completely wiping my cache drive and rebuilding it from scratch. Luckily I had setup Appdata Backup a couple months ago, so I had backups of all my appdata.

 

I changed the file system type from zfs to xfs - I don't know enough about them to say that my decision was informed, but my main array is running on xfs, so I decided to mimic that. A month or so ago when I had recreated my docker image, I had also moved everything from my cache to the array, changed the file system from btrfs to zfs (again, not really an informed decision, more that I was having issues and figured people said good things about zfs so I thought that I might as well), then moved everything back from the array to the cache. That's when I started seeing massive issues with Docker crashing and hanging the entire system.

 

Because I had such issues with the move things to the array then back to the cache method, that's why I went with the full wipe approach. I ended up having to physically remove the cache drive from my system, disable auto start in the disk settings, then reinstall the cache drive and change the file system to xfs, then format (this was because the system was hanging when it was trying to mount the cache drive after a reboot). I also took the opportunity to check my Docker settings - I knew nothing about Unraid when I initially setup my server, so I followed guides for everything. This time, I set my network type to ipvlan, disabled host access to custom networks, and set preserve user defined networks to no (all recommended settings). I'm not sure if I just got lucky, but all of my docker containers still work just fine with these settings (jellyfin, nextcloud, nginx, pihole).

 

From there it was simple enough to reinstall my apps in docker (Apps --> Previous Apps) and restore the settings from the backup. So far I have not had any issues with Docker or with anything becoming unresponsive, so I'm cautiously optimistic.