docker-compose.yaml

version: '3.8'

# Production Docker Compose Configuration
# Includes: Gateway, MCPByDojoGenesis, OTEL Collector, Langfuse observability stack

services:
  # ─── AgenticGateway (Main Service) ─────────────────────────────────────
  gateway:
    build:
      context: .
      dockerfile: Dockerfile
    ports:
      - "7340:7340"
    environment:
      # Server configuration
      PORT: "7340"
      ENVIRONMENT: "production"

      # MCP configuration
      MCP_CONFIG_PATH: "/etc/gateway/gateway-config.yaml"
      MCP_BY_DOJO_BINARY: "/opt/mcp-servers/mcp-by-dojo-genesis"

      # OTEL configuration
      OTEL_ENABLED: "true"
      OTEL_EXPORTER_OTLP_ENDPOINT: "http://otel-collector:4318"
      OTEL_SERVICE_NAME: "agentic-gateway"
      OTEL_SAMPLING_RATE: "1.0"

      # Memory configuration
      MEMORY_DB_PATH: "/app/data/dojo_memory.db"

    volumes:
      # Mount gateway MCP configuration
      - ./gateway-config.yaml:/etc/gateway/gateway-config.yaml:ro

      # Shared volume for MCP server binaries
      - mcp-bin:/opt/mcp-servers:ro

      # Persistent storage for memory
      - gateway-data:/app/data

    depends_on:
      mcp-by-dojo:
        condition: service_completed_successfully
      otel-collector:
        condition: service_started
      postgres:
        condition: service_healthy

    networks:
      - agentic-stack

    healthcheck:
      test: ["CMD", "/agentic-gateway", "--health-check"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 40s

  # ─── MCPByDojoGenesis (MCP Server Binary Provider) ────────────────────
  # This service copies the MCP server binary to a shared volume and exits
  # The gateway starts the binary as a subprocess via stdio transport
  mcp-by-dojo:
    image: ghcr.io/trespies-source/mcpbydojogenesis:latest
    volumes:
      - mcp-bin:/opt/mcp-servers
    # Copy binary to shared volume and exit
    entrypoint: ["sh", "-c", "cp /usr/local/bin/mcp-by-dojo-genesis /opt/mcp-servers/ && chmod +x /opt/mcp-servers/mcp-by-dojo-genesis && ls -la /opt/mcp-servers/"]
    networks:
      - agentic-stack

  # ─── OTEL Collector (Observability Pipeline) ──────────────────────────
  otel-collector:
    image: otel/opentelemetry-collector-contrib:0.115.0
    ports:
      - "4317:4317"  # OTLP gRPC receiver
      - "4318:4318"  # OTLP HTTP receiver
      - "8888:8888"  # Prometheus metrics (collector internal)
      - "8889:8889"  # Prometheus exporter metrics
    volumes:
      - ./otel-config.yaml:/etc/otel/config.yaml:ro
    command:
      - "--config=/etc/otel/config.yaml"
    networks:
      - agentic-stack
    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:13133/"]
      interval: 30s
      timeout: 10s
      retries: 3

  # ─── Langfuse (Observability UI & Trace Storage) ──────────────────────
  langfuse:
    image: langfuse/langfuse:2
    ports:
      - "3000:3000"
    environment:
      # Database connection
      DATABASE_URL: "postgresql://langfuse:${POSTGRES_PASSWORD:-langfuse_password}@postgres:5432/langfuse"

      # NextAuth configuration (required for Langfuse)
      NEXTAUTH_URL: "http://localhost:3000"
      NEXTAUTH_SECRET: "${LANGFUSE_NEXTAUTH_SECRET:-changeme-random-secret-key-minimum-32-characters-required}"

      # Langfuse salt (for encryption)
      SALT: "${LANGFUSE_SALT:-changeme-random-salt-minimum-32-characters-required}"

      # Optional: Langfuse API keys for programmatic access
      # LANGFUSE_SECRET_KEY: "${LANGFUSE_SECRET_KEY}"
      # LANGFUSE_PUBLIC_KEY: "${LANGFUSE_PUBLIC_KEY}"

    depends_on:
      postgres:
        condition: service_healthy

    networks:
      - agentic-stack

    healthcheck:
      test: ["CMD", "wget", "--spider", "-q", "http://localhost:3000/api/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s

  # ─── PostgreSQL (Langfuse Database) ────────────────────────────────────
  postgres:
    image: postgres:15
    ports:
      - "5432:5432"
    environment:
      POSTGRES_USER: langfuse
      POSTGRES_PASSWORD: "${POSTGRES_PASSWORD:-langfuse_password}"
      POSTGRES_DB: langfuse
    volumes:
      - postgres-data:/var/lib/postgresql/data
    networks:
      - agentic-stack
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U langfuse"]
      interval: 10s
      timeout: 5s
      retries: 5
      start_period: 30s

# ─── Networks ────────────────────────────────────────────────────────────
networks:
  agentic-stack:
    driver: bridge

# ─── Volumes ─────────────────────────────────────────────────────────────
volumes:
  # Shared volume for MCP server binaries
  mcp-bin:
    driver: local

  # Gateway persistent data (memory database)
  gateway-data:
    driver: local

  # Langfuse/PostgreSQL data
  postgres-data:
    driver: local