4 hours ago
(6 hours ago)grzywa Wrote: I have similar situation.
I have Jellyfin in Docker on TrueNAS (24.10). I have set Jellyfin Transcode Storage as "tmpfs (Temporary directory created on the RAM)". I have also configured Jellyfin Cache Storage to it's own dataset on SSD.
What I noticed is when playing large files (example 70 GB remux of my BlueRay) on my android TV player that if fully capable of decoding all so the file is sent without changes - the RAM consumed by Jellyfin doecker grows and grows and when it reaches limit set in Docker config (initially 16GB) then it restarts container. On client side there could be freeze for a moment (since jellyfin restarts very fast) or even sometimes it says it failed to play and it restarts the playback..
Now the restarts could have been caused by Linux rights (I made some changes now and I will see again how it behaves). So this needs to be confirmed as an issue.
But, my question is - it it normal for Jellyfin server to load basically full file in memory ?
I mean - I have increased RAM size for Jellyfin Docker to 64G and observed how it behaves (move was in 1/3 of it's length when restarted watching) and it consumed 42G of RAM. No more restart issues.
Is there a setting to limit that - or it's normal behavior and I just need to solve the reboot issue ?
Container crashing due to lack of free ram is normal, you need to throttle transcodes and delete downloaded segments to free up ram. Attached is my current settings, note however that I have Jellyfin running on Truenas Scale server with 256GB ram, so you might want to adjust accordingly. Also worth noting for Truenas cases is that the container will also crash if allocated ram is <tmpfs size