docker-compose.yml

services:
  # FIML MCP Server
  fiml-server:
    build:
      context: .
      dockerfile: Dockerfile
    container_name: fiml-server
    command: uvicorn fiml.server:app --host 0.0.0.0 --port 8000 --workers 1
    ports:
      - "8000:8000"
      - "9090:9090"
    environment:
      - FIML_ENV=development
      - REDIS_HOST=redis
      - POSTGRES_HOST=postgres
      - KAFKA_BOOTSTRAP_SERVERS=kafka:9092
      - RAY_ADDRESS=ray-head:6379
    env_file:
      - .env
    depends_on:
      - redis
      - postgres
      - kafka
      - ray-head
    volumes:
      - ./fiml:/app/fiml
      - ./logs:/app/logs
    networks:
      - fiml-network
    restart: unless-stopped

  # Redis (L1 Cache)
  redis:
    image: redis:7-alpine
    container_name: fiml-redis
    ports:
      - "6380:6379"
    command: redis-server --appendonly yes
    volumes:
      - redis-data:/data
    networks:
      - fiml-network
    restart: unless-stopped

  # PostgreSQL + TimescaleDB (L2 Cache)
  postgres:
    image: timescale/timescaledb:latest-pg16
    container_name: fiml-postgres
    ports:
      - "5432:5432"
    environment:
      - POSTGRES_DB=fiml
      - POSTGRES_USER=fiml
      - POSTGRES_PASSWORD=fiml_password
    volumes:
      - postgres-data:/var/lib/postgresql/data
      - ./scripts/init-db.sql:/docker-entrypoint-initdb.d/init.sql
    networks:
      - fiml-network
    restart: unless-stopped

  # Kafka
  zookeeper:
    image: confluentinc/cp-zookeeper:7.5.0
    container_name: fiml-zookeeper
    environment:
      ZOOKEEPER_CLIENT_PORT: 2181
      ZOOKEEPER_TICK_TIME: 2000
    networks:
      - fiml-network
    restart: unless-stopped

  kafka:
    image: confluentinc/cp-kafka:7.5.0
    container_name: fiml-kafka
    depends_on:
      - zookeeper
    ports:
      - "9092:9092"
    environment:
      KAFKA_BROKER_ID: 1
      KAFKA_ZOOKEEPER_CONNECT: zookeeper:2181
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://kafka:9092
      KAFKA_OFFSETS_TOPIC_REPLICATION_FACTOR: 1
      KAFKA_AUTO_CREATE_TOPICS_ENABLE: "true"
    networks:
      - fiml-network
    restart: unless-stopped

  # Ray Head Node
  ray-head:
    image: rayproject/ray:2.52.0-py311
    container_name: fiml-ray-head
    command: ray start --head --port=6379 --dashboard-host=0.0.0.0 --metrics-export-port=8080 --block
    shm_size: '5gb'
    ports:
      - "6379:6379"
      - "8265:8265" # Ray dashboard
      - "8080:8080" # Ray metrics
    volumes:
      - ./fiml:/app/fiml
    networks:
      - fiml-network
    restart: unless-stopped

  # Ray Worker Nodes
  ray-worker:
    image: rayproject/ray:2.52.0-py311
    command: ray start --address=ray-head:6379 --block
    shm_size: '5gb'
    depends_on:
      - ray-head
    volumes:
      - ./fiml:/app/fiml
    networks:
      - fiml-network
    deploy:
      replicas: 2
    restart: unless-stopped

  # Celery Worker
  celery-worker:
    build:
      context: .
      dockerfile: Dockerfile
    command: celery -A fiml.tasks.celery worker --loglevel=info --concurrency=4
    depends_on:
      - redis
      - postgres
    env_file:
      - .env
    volumes:
      - ./fiml:/app/fiml
    networks:
      - fiml-network
    deploy:
      replicas: 2
    restart: unless-stopped
    healthcheck:
      test: [ "CMD-SHELL", "celery -A fiml.tasks.celery inspect ping -d celery@$$HOSTNAME || exit 1" ]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  # Celery Beat (Scheduler)
  celery-beat:
    build:
      context: .
      dockerfile: Dockerfile
    command: celery -A fiml.tasks.celery beat --loglevel=info
    depends_on:
      - redis
      - celery-worker
    env_file:
      - .env
    volumes:
      - ./fiml:/app/fiml
    networks:
      - fiml-network
    restart: unless-stopped
    healthcheck:
      test: [ "CMD-SHELL", "python -c 'import sys; from celery import Celery; app = Celery(\"fiml\"); app.config_from_object(\"fiml.core.config:settings\"); sys.exit(0)' || exit 1" ]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  # Exporters
  postgres-exporter:
    image: prometheuscommunity/postgres-exporter
    container_name: fiml-postgres-exporter
    environment:
      DATA_SOURCE_NAME: "postgresql://fiml:fiml_password@postgres:5432/fiml?sslmode=disable"
    ports:
      - "9187:9187"
    networks:
      - fiml-network
    restart: unless-stopped

  redis-exporter:
    image: oliver006/redis_exporter
    container_name: fiml-redis-exporter
    environment:
      REDIS_ADDR: "redis:6379"
    ports:
      - "9121:9121"
    networks:
      - fiml-network
    restart: unless-stopped

  node-exporter:
    image: prom/node-exporter:latest
    container_name: fiml-node-exporter
    ports:
      - "9100:9100"
    volumes:
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /:/rootfs:ro
    command:
      - '--path.procfs=/host/proc'
      - '--path.sysfs=/host/sys'
      - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
    networks:
      - fiml-network
    restart: unless-stopped

  # Prometheus
  prometheus:
    image: prom/prometheus:latest
    container_name: fiml-prometheus
    ports:
      - "9091:9090"
    volumes:
      - ./config/prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus-data:/prometheus
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
    networks:
      - fiml-network
    restart: unless-stopped

  # Grafana
  grafana:
    image: grafana/grafana:latest
    container_name: fiml-grafana
    ports:
      - "3000:3000"
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=admin
      - GF_USERS_ALLOW_SIGN_UP=false
    volumes:
      - grafana-data:/var/lib/grafana
      - ./config/grafana/dashboards:/etc/grafana/provisioning/dashboards
      - ./config/grafana/datasources:/etc/grafana/provisioning/datasources
    networks:
      - fiml-network
    restart: unless-stopped

networks:
  fiml-network:
    driver: bridge

volumes:
  redis-data:
  postgres-data:
  prometheus-data:
  grafana-data: