compose.yaml

services:
  dcgm_exporter:
    image: nvcr.io/nvidia/k8s/dcgm-exporter:${DCGM_EXPORTER_IMAGE_TAG}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [ gpu ]
    restart: always
    environment:
      - DCGM_EXPORTER_NO_HOSTNAME=1
    cap_add:
      - SYS_ADMIN
    ports:
      - "${DCGM_EXPORTER_HOST_PORT}:9400"
    networks:
      - gpu_metrics
  prometheus:
    image: prom/prometheus:${PROMETHEUS_IMAGE_TAG}
    ports:
      - "${PROMETHEUS_HOST_PORT}:9090"
    command:
      - --config.file=/etc/prometheus/prometheus.yml
      - --storage.tsdb.path=/prometheus
      - --web.console.libraries=/usr/share/prometheus/console_libraries
      - --web.console.templates=/usr/share/prometheus/consoles
      - --storage.tsdb.retention.time=${PROMETHEUS_STORAGE_TSDB_RETENTION_TIME}
    restart: always
    volumes:
      - ./gpu-metrics-prometheus-config.yaml:/etc/prometheus/prometheus.yml:ro
      - prometheus_data:/prometheus
    networks:
      - gpu_metrics
  grafana:
    image: grafana/grafana:${GRAFANA_IMAGE_TAG}
    ports:
      - "${GRAFANA_HOST_PORT}:3000"
    restart: always
    environment:
      - GF_SECURITY_ADMIN_USER=${GRAFANA_ADMIN_USER}
      - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_ADMIN_PASSWORD}
    volumes:
      - grafana_data:/var/lib/grafana
      - ./grafana_provisioning:/etc/grafana/provisioning:ro
      - ./grafana_dashboards:/etc/dashboards:ro
    networks:
      - gpu_metrics

volumes:
  prometheus_data:
    driver: local
  grafana_data:
    driver: local

networks:
  gpu_metrics:
    driver: bridge