docker-compose.yml

# Docker Compose for ManimCat
# Production-ready with Redis for task queue and state management

services:
  redis:
    image: redis:7-alpine
    container_name: manim-redis
    ports:
      - "${REDIS_PORT:-6379}:6379"
    volumes:
      - redis-data:/data
    restart: unless-stopped
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 5s
      timeout: 5s
      retries: 10
      start_period: 5s
    command: redis-server --appendonly yes --maxmemory 256mb --maxmemory-policy allkeys-lru
    networks:
      - manimcat-network

  manimcat:
    build:
      context: .
      dockerfile: Dockerfile
      args:
        - NODE_ENV=production
    image: manimcat:latest
    # Compose provides the shared Redis service. Bypass the image's Hugging Face
    # single-container launcher, which owns an embedded localhost Redis instance.
    command: ["node", "dist/server.js"]
    # No fixed container_name: keeping it would prevent `docker compose up --scale manimcat=N`.
    #
    # Host port binding. One replica keeps stable access:
    #   MANIMCAT_PUBLISHED_PORT=3000
    # Multi-replica smoke run on one host lets Docker assign host ports:
    #   MANIMCAT_PUBLISHED_PORT=0 docker compose up --scale manimcat=2
    #   docker compose port manimcat 3000   # prints the assigned host port
    # Production traffic across replicas still requires a reverse proxy, load balancer or
    # orchestrator ingress: this project ships cross-replica event fan-out, not an HTTP LB.
    ports:
      - "${MANIMCAT_PUBLISHED_PORT:-3000}:3000"
    environment:
      - NODE_ENV=production
      - PORT=3000
      - LOG_LEVEL=${LOG_LEVEL:-info}
      - PROD_SUMMARY_LOG_ONLY=${PROD_SUMMARY_LOG_ONLY:-true}
      - HTTP_PROXY=${HTTP_PROXY:-}
      - HTTPS_PROXY=${HTTPS_PROXY:-}
      - NO_PROXY=${NO_PROXY:-}
      - http_proxy=${http_proxy:-}
      - https_proxy=${https_proxy:-}
      - no_proxy=${no_proxy:-}
      - REDIS_HOST=redis
      - REDIS_PORT=6379
      - REDIS_DB=${REDIS_DB:-0}
      # Cross-replica Studio event delivery (Redis Pub/Sub); memory only for single-instance setups
      - STUDIO_EVENT_TRANSPORT=${STUDIO_EVENT_TRANSPORT:-redis}
      # Optional: keep deployments that share one Redis on separate channels
      - STUDIO_EVENT_REDIS_CHANNEL=${STUDIO_EVENT_REDIS_CHANNEL:-}
      # Cross-replica Run coordination: one session lease per Studio session, plus durable
      # cancellation markers. Renewal must stay below half the lease TTL (validated at boot).
      - STUDIO_RUN_COORDINATION=${STUDIO_RUN_COORDINATION:-redis}
      - STUDIO_RUN_LEASE_TTL_MS=${STUDIO_RUN_LEASE_TTL_MS:-60000}
      - STUDIO_RUN_LEASE_RENEW_MS=${STUDIO_RUN_LEASE_RENEW_MS:-15000}
      - STUDIO_RUN_CANCEL_TTL_MS=${STUDIO_RUN_CANCEL_TTL_MS:-3600000}
      # Optional: keep deployments that share one Redis on separate prefixes/channels
      - STUDIO_RUN_REDIS_PREFIX=${STUDIO_RUN_REDIS_PREFIX:-}
      - STUDIO_RUN_CONTROL_CHANNEL=${STUDIO_RUN_CONTROL_CHANNEL:-}
      # Shared Studio persistence is required when requests can reach multiple replicas.
      - ENABLE_STUDIO_DB=${ENABLE_STUDIO_DB:-false}
      - SUPABASE_URL=${SUPABASE_URL:-}
      - SUPABASE_KEY=${SUPABASE_KEY:-}
      # Upstream routing (required for server-side generation unless frontend passes customApiConfig)
      - MANIMCAT_ROUTE_KEYS=${MANIMCAT_ROUTE_KEYS:-}
      - MANIMCAT_ROUTE_API_URLS=${MANIMCAT_ROUTE_API_URLS:-}
      - MANIMCAT_ROUTE_API_KEYS=${MANIMCAT_ROUTE_API_KEYS:-}
      - MANIMCAT_ROUTE_MODELS=${MANIMCAT_ROUTE_MODELS:-}
      - DISPLAY=:99
    volumes:
      # Persist studio workspace sessions, generated code, and studio outputs
      - studio-workspace-data:/app/.studio-workspace
      # Persist generated images and uploaded reference images
      - image-storage:/app/public/images
      # Persist generated videos
      - video-storage:/app/public/videos
      # Persist Manim media cache and intermediate render artifacts
      - manim-media:/app/media
      # Temp directory for rendering
      - manim-tmp:/app/tmp
    depends_on:
      redis:
        condition: service_healthy
    restart: unless-stopped
    healthcheck:
      test: ["CMD-SHELL", "node -e \"require('http').get('http://localhost:3000/health', (r) => process.exit(r.statusCode === 200 ? 0 : 1))\""]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s
    networks:
      - manimcat-network
    # Resource limits (adjust based on your needs)
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 4G
        reservations:
          cpus: '1'
          memory: 2G

networks:
  manimcat-network:
    driver: bridge

volumes:
  studio-workspace-data:
    driver: local
  image-storage:
    driver: local
  manim-tmp:
    driver: local
  manim-media:
    driver: local
  redis-data:
    driver: local
  video-storage:
    driver: local