monitoring/compose.yml

---
version: "3.8"

services:
  node_exporter:
    image: prom/node-exporter:v1.0.1
    user: root
    environment:
      - NODE_ID={{.Node.ID}}
    volumes:
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /:/rootfs:ro
      - /etc/hostname:/etc/nodename
    command:
      - "--path.sysfs=/host/sys"
      - "--path.procfs=/host/proc"
      - "--path.rootfs=/rootfs"
      - "--collector.textfile.directory=/etc/node-exporter/"
      - "--collector.filesystem.ignored-mount-points=^/(sys|proc|dev|host|etc)($$|/)"
      - "--no-collector.ipvs"
    configs:
      - source: node_exporter_entrypoint_sh
        target: /entrypoint.sh
    entrypoint: ["/bin/sh", "-e", "/entrypoint.sh"]
    networks:
      - exporters
    deploy:
      mode: global
      endpoint_mode: dnsrr

  cadvisor:
    image: google/cadvisor:latest
    command: -logtostderr -docker_only
    volumes:
      - /var/lib/docker/:/var/lib/docker
      - /dev/disk/:/dev/disk
      - /sys:/sys
      - /var/run:/var/run
      - /:/rootfs
    networks:
      - exporters
    deploy:
      mode: global
      endpoint_mode: dnsrr

  loki:
    image: grafana/loki:2.0.0
    command: -config.file=/etc/loki/local-config.yaml
    networks:
      - loki
      - api
    deploy:
      endpoint_mode: dnsrr
    ports:
      - target: 3100
        published: 3100
        protocol: tcp
        mode: host
    configs:
      - source: loki_yml
        target: /etc/loki/local-config.yaml
    volumes:
      - loki-data:/loki

  promtail:
    image: grafana/promtail:2.0.0
    volumes:
      - /var/log:/var/log
    command: -config.file=/etc/promtail/config.yml
    networks:
      - loki
    deploy:
      mode: global
      endpoint_mode: dnsrr

  app:
    image: prom/prometheus:latest
    volumes:
      - prometheus-data:/prometheus:rw
    configs:
      - source: prometheus_yml
        target: /etc/prometheus/prometheus.yml
    networks:
      api:
        aliases:
          - prometheus_api
      exporters: ~
      proxy: ~
    deploy:
      restart_policy:
        condition: on-failure
      labels:
        - "traefik.enable=true"
        - "traefik.http.services.${STACK_NAME}_prometheus.loadbalancer.server.port=9090"
        - "traefik.http.routers.${STACK_NAME}-prometheus.rule=Host(`${PROMETHEUS_DOMAIN}`)"
        - "traefik.http.routers.${STACK_NAME}-prometheus.entrypoints=web-secure"
        - "traefik.http.routers.${STACK_NAME}-prometheus.tls=true"
        - "traefik.http.routers.${STACK_NAME}-prometheus.tls.certresolver=${LETS_ENCRYPT_ENV}"
    healthcheck:
      test: "wget -q http://localhost:9090/graph -O/dev/null"
      interval: 5s
      timeout: 10s
      retries: 3
      start_period: 10s

  grafana:
    image: grafana/grafana
    volumes:
      - grafana-data:/var/lib/grafana:rw
    configs:
      - source: grafana_datasources_yml
        target: /etc/grafana/provisioning/datasources/datasources.yml
      - source: grafana_dashboards_yml
        target: /etc/grafana/provisioning/dashboards/dashboards.yml
      - source: grafana_swarm_dashboard_json
        target: /var/lib/grafana/dashboards/docker-swarm-nodes.json
      - source: grafana_stacks_dashboard_json
        target: /var/lib/grafana/dashboards/docker-swarm-stacks.json
      - source: grafana_traefik_dashboard_json
        target: /var/lib/grafana/dashboards/traefik.json
    networks:
      - api
      - proxy
    environment:
      - GF_SMTP_HOST
      - GF_SMTP_ENABLED
      - GF_SMTP_FROM_ADDRESS
      - GF_SMTP_SKIP_VERIFY
      - GF_SECURITY_ALLOW_EMBEDDING
      - GF_INSTALL_PLUGINS=grafana-piechart-panel
      - GF_SERVER_ROOT_URL=https://${GRAFANA_DOMAIN}
    deploy:
      labels:
        - "traefik.enable=true"
        - "traefik.http.services.${STACK_NAME}-grafana.loadbalancer.server.port=3000"
        - "traefik.http.routers.${STACK_NAME}-grafana.rule=Host(`${GRAFANA_DOMAIN}`)"
        - "traefik.http.routers.${STACK_NAME}-grafana.entrypoints=web-secure"
        - "traefik.http.routers.${STACK_NAME}-grafana.tls=true"
        - "traefik.http.routers.${STACK_NAME}-grafana.tls.certresolver=${LETS_ENCRYPT_ENV}"
    healthcheck:
      test: "wget -q http://localhost:3000/ -O/dev/null"
      interval: 5s
      timeout: 10s
      retries: 3
      start_period: 10s

configs:
  prometheus_yml:
    template_driver: golang
    name: ${STACK_NAME}_prometheus_yml_${PROMETHEUS_YML_VERSION}
    file: prometheus.yml.tmpl
  promtail_yml:
    name: ${STACK_NAME}_promtail_yml_${PROMTAIL_YML_VERSION}
    file: promtail.yml
  loki_yml:
    template_driver: golang
    name: ${STACK_NAME}_loki_yml_${LOKI_YML_VERSION}
    file: loki.yml.tmpl
  grafana_datasources_yml:
    name: ${STACK_NAME}_grafana_datasources_yml_${GRAFANA_DATASOURCES_YML_VERSION}
    file: grafana-datasources.yml
  grafana_dashboards_yml:
    name: ${STACK_NAME}_grafana_dashboards_yml_${GRAFANA_DASHBOARDS_YML_VERSION}
    file: grafana-dashboards.yml
  grafana_swarm_dashboard_json:
    name: ${STACK_NAME}_grafana_swarm_dashboard_json_${GRAFANA_SWARM_DASHBOARD_JSON_VERSION}
    file: grafana-swarm-dashboard.json
  grafana_stacks_dashboard_json:
    name: ${STACK_NAME}_grafana_stacks_dashboard_json_${GRAFANA_STACKS_DASHBOARD_JSON_VERSION}
    file: grafana-stacks-dashboard.json
  grafana_traefik_dashboard_json:
    name: ${STACK_NAME}_grafana_traefik_dashboard_json_${GRAFANA_TRAEFIK_DASHBOARD_JSON_VERSION}
    file: grafana-traefik-dashboard.json
  node_exporter_entrypoint_sh:
    name: ${STACK_NAME}_node_exporter_entrypoint_${NODE_EXPORTER_ENTRYPOINT_VERSION}
    file: node-exporter-entrypoint.sh

volumes:
  prometheus-data:
  grafana-data:
  loki-data:

networks:
  api:
    driver: overlay
  exporters:
    driver: overlay
    attachable: true
  loki:
    driver: overlay
    attachable: true
  proxy:
    external: true