Many users reported high memory/RAM usage, some 8GB+.
In my case gone from 1.5GB+ to 400MB or less on Raspberry Pi 4.
Adding MALLOC_TRIM_THRESHOLD_=100000can make a big difference.
With Docker:
Add to your docker-compose.yml and docker compose down && docker compose up -d
...
environment:
- MALLOC_TRIM_THRESHOLD_=100000
...
With systemd:
Edit /etc/default/jellyfin change the value of MALLOC_TRIM_THRESHOLD_ and restart the service
# Disable glibc dynamic heap adjustment
MALLOC_TRIM_THRESHOLD_=100000
Source: https://github.com/jellyfin/jellyfin/issues/6306#issuecomment-1774093928
Official docker,Debian,Fedora packages already contain MALLOC_TRIM_THRESHOLD_.
Not present on some docker images like linuxserver/jellyfin
Check is container (already) have the variable
docker exec -it jellyfin printenv | grep MALLOC_TRIM_THRESHO LD_
How well does jellyfin work on a rpi? How many simultaneous 1080p users with AVC\ACC media can it do? Active cooling?
Didn’t try but depends on which RPI. 4? 5?
Also depends on the storage medium (SD? SSD?), assuming there’s no transcoding.