This is happening basically daily, for the last 10 days at least. Server works well, then all of a sudden in becomes unreachable, even using the monitor attached via hdmi. In one of the previous crashes, I was able to see the mode after a long freeze, opened the terminal in the ui, but it was not working. I’m pretty sure i also had a crash when the os was not in ui mode and the command line was not responding either. I also tried SSH, but it is not working. My only option is a hard reset at that point, but this already created a lot of problems with storage becoming unmountable.

 

 I’ve attached diagnostics taken seconds after the restart, I’m not sure Unraid saves data from previous the crash.

nickfury-diagnostics-20241214-1503.zip

26 minutes ago, Dreadino said:

I’m not sure Unraid saves data from previous the crash.

It doesn't, update to latest stable, then enable the syslog server and post that after a crash.

I did and tonight, seemingly after the CA Appdata Backup run, the server became  unresponsive. No web ui, no ssh, no video output when using its monitor (which is not always the case).

 

Note: I wanted to upload the backup.log for CA Appdata Backup, but it was not created and only a handful of apps were backed up, despite the backup being marked as successful (or at least, not being marked as failed). I guess the crash occurred during the process.

 

Edit: I'll throw in diagnostics taken right now.

 

Edit2: I'll add that the docker page is acting up now. It take a loooong time to load, there are no folders (I use the folder plugin), it shows "ghost" containers and when trying to update a container it fails telling me

 

docker: Error response from daemon: Conflict. The container name "/qbittorrent" is already in use by container "b3726b12ae7aace567cce16d0fbbe34a69c3f2d817e3610136557834b2104f9f". You have to remove (or rename) that container to be able to reuse that name.
See 'docker run --help'..

 

I had this 2 days ago and even after trying all the tricks I found to stop the array, something was still clinging to /var/lib/docker/graph, not allowing the array to stop.

syslog-192.168.0.30.log

nickfury-diagnostics-20241218-0659.zip

Edited December 18, 20241 yr by Dreadino
Added diagnostics

There are multiple zfs related call traces, I would start by running memtest, and if nothing is found, backup and recreate the zfs pool(s).

The test is still running, but it doesn't look good. What would be the next steps in this situation? New RAM? ECC?

First step is to fix the RAM issue, you can try running memtest with one stick at a time to see if you can find the bad one.