dex/docker/docker-compose.cuda.yml

version: '3.8'

services:
  dex-cuda:
    build:
      context: .
      dockerfile: Dockerfile.cuda
    image: lux-dex:cuda
    container_name: lux-dex-cuda
    
    # Enable GPU support
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    
    environment:
      - CGO_ENABLED=1
      - LD_LIBRARY_PATH=/usr/local/cuda/lib64:/app
      - ENGINE_MODE=cuda
      - LOG_LEVEL=debug
      - METRICS_ENABLED=true
    
    ports:
      - "8080:8080"  # REST API
      - "8081:8081"  # WebSocket
      - "9090:9090"  # Metrics
    
    volumes:
      - ./data:/app/data
      - ./logs:/app/logs
    
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 30s
      timeout: 3s
      retries: 3
      start_period: 10s
    
    restart: unless-stopped

  dex-cuda-test:
    build:
      context: .
      dockerfile: Dockerfile.cuda
    image: lux-dex:cuda-test
    container_name: lux-dex-cuda-test
    
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: 1
              capabilities: [gpu]
    
    environment:
      - CGO_ENABLED=1
      - LD_LIBRARY_PATH=/usr/local/cuda/lib64:/app
    
    command: |
      bash -c "
        cd /build
        echo '=== CUDA Detection ==='
        nvidia-smi
        echo ''
        echo '=== Running Tests ==='
        go test -v ./pkg/mlx/...
        echo ''
        echo '=== Running Benchmarks ==='
        go test -bench=. -benchtime=30s ./pkg/mlx/...
      "
    
    volumes:
      - ./benchmark-results:/results
    
    profiles:
      - test

  monitoring:
    image: prom/prometheus:latest
    container_name: lux-dex-prometheus
    
    volumes:
      - ./monitoring/prometheus.yml:/etc/prometheus/prometheus.yml
      - prometheus-data:/prometheus
    
    ports:
      - "9091:9090"
    
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
    
    restart: unless-stopped

  grafana:
    image: grafana/grafana:latest
    container_name: lux-dex-grafana
    
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=admin
      - GF_USERS_ALLOW_SIGN_UP=false
    
    volumes:
      - ./monitoring/grafana/dashboards:/etc/grafana/provisioning/dashboards
      - grafana-data:/var/lib/grafana
    
    ports:
      - "3000:3000"
    
    depends_on:
      - monitoring
    
    restart: unless-stopped

volumes:
  prometheus-data:
  grafana-data: