System Hang When Searching on Web After Upgrading from 10.8.x to 10.9.x

[10rounds]

I recently upgraded my Jellyfin instance on a Raspberry Pi from version 10.8.x to 10.9.9 via apt, and then to 10.9.10 today. Post-upgrade, the Jellyfin web interface loads correctly, and I can play music without any issues. However, the system intermittently hangs completely. Despite waiting for about an hour, the system remains unresponsive, requiring a hard reboot.

Additionally, when the hang occurs, SSH becomes completely unresponsive, although the connection doesn’t drop. The console is also extremely slow, making even basic keyboard input difficult and almost impossible to execute any commands.

After experiencing this issue several times, I performed a backup and an SQLite database health check, both of which reported no problems. I then realized that the hang consistently occurs when I perform a search through the web interface, leading me to believe that this may be the trigger after the upgrade. Interestingly, I have some scripts that perform searches via the API, and those seem to work fine without causing any hangs.

I’ve reviewed the Jellyfin logs but couldn’t find anything related to the issue. Could you please advise on what further information I could provide to help diagnose and report this problem?

Thank you.

[TheDreadPirate]

You can try reducing the number of threads that Jellyfin uses for library scans and such.

Dashboard > General. Set Parallel library scan tasks limit and Parallel image encoding limit to 1.

Dashboard > Playback > Trickplays. Set FFmpeg Threads to 1.

[10rounds]

Thanks for the suggestions! Just wondering, how would reducing threads help with the search issue on the web interface? My Jellyfin has been working fine for over a year, and searches were perfect before the upgrade.

Also, the library scan time is still the same as before, about 4 minutes, so I don’t think that’s the issue. I checked the logs appears ffmpeg/ffprobe, and they look normal.

Any ideas if this could be related to something else?

Thanks!

[TheDreadPirate]

Can you share your full jellyfin logs via pastebin?

[10rounds]

Here’s the link to my jellyfin logs: https://pastebin.com/tqZnnp26
https://pastebin.com/wkRChWXW
https://pastebin.com/RCxSTiqx
https://pastebin.com/dsjTrgNG
https://pastebin.com/HRRmNCup

grep -v ffprobe: https://pastebin.com/ut1NbSP8


After the hang occurred, I reviewed the logs and made some adjustments on system. I increased sysctl fs.inotify.max_user_instances to 8192 and fs.inotify.max_user_watches to 524288, and I also changed permissions to ensure Jellyfin has write access to all directories. Unfortunately, the hang still happens despite these changes.

Let me know if you need any more information or if you spot anything in the logs.

Thanks.

[10rounds]

I just triggered another hang, but this time it wasn’t during a search. I clicked "Add to Playlist" on the web UI, and the system hung immediately after. The playlist selection box didn't load. Before the upgrade, the playlist selection box was slow, but it would eventually load.

For some background, my library contains ~50,000 m4a files and about 3,800 m3u playlists. Both the database and the library are stored on an SSD.

[TheDreadPirate]

Nothing in the log indicates that Jellyfin is the root of the problem. Check your system logs for any possible issues at the OS level.

Another thing to keep an eye on is system load, which is related to CPU usage but not the same. If you have low CPU usage from an app, but high system load, there is something going on at the OS level. Kernel backlogs, disk I/O bottleneck, memory bottleneck.

[10rounds]

Would it be a good idea to set the Jellyfin log level to debug? I've reviewed the system logs, but there's nothing useful in journalctl.

I'm also considering downgrading back to 10.8. Has the database schema changed with the upgrade? This upgrade has been quite painful.

[TheDreadPirate]

The last RPi user that had this problem didn't have anything in their debug logs that indicated the problem was due to something with Jellyfin, specifically.

You cannot downgrade to 10.8 unless you have a pre-upgrade backup. The database schema changed significantly.

[10rounds]

Thank you. I’m curious, since you mentioned another RPi user had a similar issue, could you share more details or links to their case? It might help me identify if I’m dealing with the same problem.

Appreciate!