siftlode/docker-compose.selfhost.yml

# Self-hosting Siftlode — pulls the prebuilt image, no source build needed.
#
#   1. Run ./install.sh  (or install.ps1 on Windows) once — it generates a .env with secrets.
#   2. docker compose -f docker-compose.selfhost.yml up -d
#   3. Open the setup wizard at the URL printed in the logs:
#        docker compose -f docker-compose.selfhost.yml logs api | grep SETUP
#
# The .env only needs four values (the install script generates all of them):
#   POSTGRES_PASSWORD, SECRET_KEY, TOKEN_ENCRYPTION_KEY, OAUTH_REDIRECT_URL
# Everything else — the admin account, Google sign-in, SMTP — is set in the web wizard on
# first run. See docs/self-hosting.md.

# Bounded container logs. Docker's json-file driver keeps every line forever by default, and the API
# now writes one structured line per request (E22 S2) — 3×50 MB per service is days of history on a
# busy instance and a hard ceiling on the disk it can take.
x-logging: &logging
  driver: json-file
  options:
    max-size: "50m"
    max-file: "3"

services:
  db:
    image: postgres:16-alpine
    environment:
      POSTGRES_USER: ${POSTGRES_USER:-siftlode}
      POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:?POSTGRES_PASSWORD must be set (run install.sh)}
      POSTGRES_DB: ${POSTGRES_DB:-siftlode}
    volumes:
      - siftlode_pgdata:/var/lib/postgresql/data
    networks: [internal]
    security_opt:
      - no-new-privileges:true
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U ${POSTGRES_USER:-siftlode} -d ${POSTGRES_DB:-siftlode}"]
      interval: 10s
      timeout: 5s
      retries: 12
    logging: *logging
    restart: unless-stopped

  api:
    image: git.b1fr0st.eu/peter/siftlode:${IMAGE_TAG:-latest}
    env_file:
      - .env
    environment:
      DATABASE_URL: postgresql+psycopg://${POSTGRES_USER:-siftlode}:${POSTGRES_PASSWORD}@db:5432/${POSTGRES_DB:-siftlode}
      # This instance owns the background scheduler (single writer).
      SCHEDULER_ENABLED: "true"
      # Download center: the API serves finished files + builds filmstrips; the worker (below) runs
      # the yt-dlp job loop. Read-only root is fine — it only writes to the /downloads mount + tmpfs.
      DOWNLOAD_ROOT: /downloads
      WORKER_ENABLED: "false"
    depends_on:
      db:
        condition: service_healthy
    networks: [internal]
    ports:
      # Reachable on the host's LAN at http://<host>:8080. Put a reverse proxy (Caddy/Nginx) in
      # front for HTTPS / public exposure — and set OAUTH_REDIRECT_URL to the https URL.
      - "${HTTP_PORT:-8080}:8000"
    volumes:
      # Downloaded media. Defaults to a Docker-managed named volume; set DOWNLOAD_HOST_PATH in .env
      # to a host directory (e.g. one your Plex server reads) to keep the Plex-style tree there.
      - ${DOWNLOAD_HOST_PATH:-siftlode_downloads}:/downloads
      # Plex library, read-only. The Plex module streams files itself, so it needs to SEE them: set
      # PLEX_MEDIA_HOST_PATH in .env to the host directory holding the library (the local side of the
      # admin "path map", whose default is /data=/plex-media). Left unset it falls back to an empty
      # named volume — harmless, and the Plex module is off by default.
      - ${PLEX_MEDIA_HOST_PATH:-siftlode_plex_media}:/plex-media:ro
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    read_only: true
    tmpfs:
      - /tmp
    healthcheck:
      test: ["CMD", "python", "-c", "import urllib.request,sys; sys.exit(0 if urllib.request.urlopen('http://localhost:8000/healthz').status==200 else 1)"]
      interval: 15s
      timeout: 5s
      retries: 5
      start_period: 30s
    logging: *logging
    restart: unless-stopped

  # Download worker: same image, runs the yt-dlp / ffmpeg job loop instead of the API. Shares the
  # DB (job queue) and the downloads mount with the API. Not read-only (yt-dlp/deno/ffmpeg write to
  # $HOME caches + the staging dir).
  worker:
    image: git.b1fr0st.eu/peter/siftlode:${IMAGE_TAG:-latest}
    command: ["python", "-m", "app.worker"]
    # tini as PID 1: reaps orphaned ffmpeg/yt-dlp grandchildren when the watchdog hard-kills a
    # download child's process group (the worker reaps its own direct children via subprocess.wait).
    init: true
    env_file:
      - .env
    environment:
      DATABASE_URL: postgresql+psycopg://${POSTGRES_USER:-siftlode}:${POSTGRES_PASSWORD}@db:5432/${POSTGRES_DB:-siftlode}
      SCHEDULER_ENABLED: "false"
      DOWNLOAD_ROOT: /downloads
      WORKER_ENABLED: "true"
    depends_on:
      db:
        condition: service_healthy
      # Start only after the API is healthy (it applies the DB migrations on startup), so the
      # download tables exist before the worker touches them.
      api:
        condition: service_healthy
      bgutil-pot:
        condition: service_started
    networks: [internal]
    volumes:
      - ${DOWNLOAD_HOST_PATH:-siftlode_downloads}:/downloads
      # Same read-only Plex library as the API — the worker resolves library paths too.
      - ${PLEX_MEDIA_HOST_PATH:-siftlode_plex_media}:/plex-media:ro
    security_opt:
      - no-new-privileges:true
    cap_drop:
      - ALL
    tmpfs:
      - /tmp
    logging: *logging
    restart: unless-stopped

  # PO-token provider sidecar: mints YouTube Proof-of-Origin tokens so the worker beats bot-detection
  # and unlocks high-quality formats (reached at http://bgutil-pot:4416, the config default).
  bgutil-pot:
    image: brainicism/bgutil-ytdlp-pot-provider:1.3.1
    init: true
    networks: [internal]
    security_opt:
      - no-new-privileges:true
    logging: *logging
    restart: unless-stopped

volumes:
  siftlode_pgdata:
  siftlode_downloads:
  # Empty fallback for the Plex media mount when PLEX_MEDIA_HOST_PATH is unset (Plex module off).
  siftlode_plex_media:

networks:
  internal: