SOLVED: Transcoding fails in docker image due to missing render group

[esmorun]

Pulled the latest docker image, version 10.11.0: https://hub.docker.com/layers/jellyfin/j...4485bb73ed

Created a container like this;

Code:

sudo podman run \
  --detach \
  --name jellyfin \
  --publish 8096:8096/tcp \
  --publish 7359:7359/udp \
  --user 128:136 \ # jellyfin user and group
  --group-add=44 \ # Video group
  --group-add=110 \ # Render group
  --device /dev/dri/:/dev/dri/ \ # VAAPI devices
  --network=host \
  --volume jellyfin-cache:/cache \
  --volume jellyfin-config:/config \
  --mount type=bind,source=/mnt/Movies,destination=/mnt/Movies,ro=true \
  --restart=unless-stopped \
  --label "io.containers.autoupdate=registry" \
  docker.io/jellyfin/jellyfin:latest

After this, hardware transcoding was not working inside the container. After some troubleshooting i realized the issue was that the Render group did not exist inside the container (notice there's no name next to id 110):

Code:

jellyfin@MyServer:/$ id
uid=128(jellyfin) gid=136(136) groups=136(136),44(video),110

I was able to get it working by manually creating the group inside the container:

Code:

sudo podman exec --user 0:0 jellyfin groupadd -g 110 render

Surely the group should have already been there? Or am I missing something obvious?

[esmorun]

I found a permanent workaround by creating a systemd file that launches the container and automatically runs the command to fix the render group problem at startup:

/etc/systemd/system/jellyfin-pod.service

Code:

[Unit]
Description=Jellyfin container (persistent) via Podman
After=network-online.target
Wants=network-online.target

[Service]
Restart=always
RestartSec=5s
TimeoutStartSec=300
ExecStart=/usr/bin/podman start -a jellyfin
ExecStartPost=/usr/bin/podman exec -u 0 jellyfin groupadd -f -g 110 render
ExecStop=/usr/bin/podman stop -t 30 jellyfin
KillMode=process
Type=simple

[Install]
WantedBy=multi-user.target

sudo systemctl daemon-reload
sudo systemctl enable --now jellyfin-pod.service