# Deploy on the Sky1 host (Radxa Orion O6). This file runs the image you
# cross-built on Windows and pushed/loaded onto the board.
#
# Find your VPU nodes first:   ls -l /dev/video*   and   ls -l /dev/dma_heap
# The Linlon/Sky1 VPU typically exposes several /dev/videoN nodes; pass the
# ones that are the codec (m2m) devices. When unsure, pass all of them.

services:
  jellyfin:
    image: jellyfin-orion-o6:latest
    container_name: jellyfin-orion-o6
    restart: unless-stopped
    ports:
      - "8096:8096"          # web UI / API
      # - "8920:8920"        # optional HTTPS
      # - "7359:7359/udp"    # client auto-discovery
      # - "1900:1900/udp"    # DLNA

    # --- Sky1 VPU passthrough -------------------------------------------
    group_add:
      - "video"              # or the numeric GID of the 'video' group on the host
    devices:
      - "/dev/video0:/dev/video0"
      - "/dev/video1:/dev/video1"
      # add every codec m2m node the host exposes, e.g. /dev/video2 ...
    # DMABUF allocator the V4L2 M2M path uses for zero-copy buffers.
    volumes:
      - "/dev/dma_heap:/dev/dma_heap"
      - "./config:/config"
      - "./cache:/cache"
      - "/srv/media:/media:ro"   # <-- point at your library

    # If device passthrough is fiddly while you get it working, you can
    # temporarily uncomment this instead of the devices/group_add block:
    # privileged: true