# Monitoring and Observability Services
# Services for monitoring your homelab infrastructure
# Place in /opt/stacks/monitoring/docker-compose.yml

# RESTART POLICY GUIDE:
# - unless-stopped: Core infrastructure services that should always run
# - no: Services with Sablier lazy loading (start on-demand)
# - See individual service comments for specific reasoning

# Service Access URLs:
# - Prometheus: http://${SERVER_IP}:9090 (or configure Traefik)
# - Grafana: http://${SERVER_IP}:3000 (or configure Traefik)
# - Uptime Kuma: https://status.${DOMAIN}
# - Node Exporter: http://${SERVER_IP}:9100/metrics
# - cAdvisor: http://${SERVER_IP}:8082
# - Loki: http://${SERVER_IP}:3100
# NOTE: Prometheus, Grafana, Loki use ports because they need to be accessible to other services
#       Add Traefik labels if you want https://prometheus.${DOMAIN} access

services:
  # Prometheus - Metrics collection and storage
  # Access at: http://${SERVER_IP}:9090
  prometheus:
    image: prom/prometheus:v2.48.1
    deploy:
      resources:
        limits:
          cpus: '0.75'
          memory: 512M
          pids: 1024
        reservations:
          cpus: '0.25'
          memory: 256M
    container_name: prometheus
    restart: unless-stopped
    networks:
      - homelab-network
      - traefik-network
    ports:
      - "9090:9090"
    volumes:
      - ./config/prometheus:/etc/prometheus
      - prometheus-data:/prometheus
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
      - '--storage.tsdb.retention.time=30d'
      - '--web.console.libraries=/etc/prometheus/console_libraries'
      - '--web.console.templates=/etc/prometheus/consoles'
      - '--web.enable-lifecycle'
    labels:
      # TRAEFIK CONFIGURATION
      # ==========================================
      # Service metadata
      - "homelab.category=monitoring"
      - "homelab.description=Metrics collection and time-series database"
      # Traefik reverse proxy (comment/uncomment to disable/enable)
      # If Traefik is on a remote server: these labels are NOT USED; 
      #   configure external yml files in /traefik/dynamic folder instead.
      - "traefik.enable=true"
      - "traefik.http.routers.prometheus.rule=Host(`prometheus.${DOMAIN}`)"
      - "traefik.http.routers.prometheus.entrypoints=websecure"
      - "traefik.http.routers.prometheus.tls=true"
      - "traefik.http.routers.prometheus.tls.certresolver=letsencrypt"
      - "traefik.http.routers.prometheus.middlewares=authelia@docker"
      - "traefik.http.services.prometheus.loadbalancer.server.port=9090"

  # Grafana - Metrics visualization
  # Access at: http://${SERVER_IP}:3000
  # Default credentials: admin / admin (change on first login)
  grafana:
    image: grafana/grafana:10.2.3
    deploy:
      resources:
        limits:
          cpus: '0.50'
          memory: 256M
          pids: 512
        reservations:
          cpus: '0.25'
          memory: 128M
    container_name: grafana
    restart: unless-stopped
    networks:
      - homelab-network
      - traefik-network
    ports:
      - "3000:3000"
    volumes:
      - grafana-data:/var/lib/grafana
      - ./config/grafana/provisioning:/etc/grafana/provisioning
    environment:
      - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_ADMIN_PASSWORD}
      - GF_USERS_ALLOW_SIGN_UP=false
      - GF_SERVER_ROOT_URL=https://grafana.${DOMAIN}
      - GF_INSTALL_PLUGINS=grafana-clock-panel,grafana-simple-json-datasource,grafana-piechart-panel
    user: "${PUID}:${PGID}"
    depends_on:
      - prometheus
    labels:
      # TRAEFIK CONFIGURATION
      # ==========================================
      # Service metadata
      - "homelab.category=monitoring"
      - "homelab.description=Metrics visualization and dashboards"
      # Traefik reverse proxy (comment/uncomment to disable/enable)
      # If Traefik is on a remote server: these labels are NOT USED; 
      #   configure external yml files in /traefik/dynamic folder instead.
      - "traefik.enable=true"
      - "traefik.http.routers.grafana.rule=Host(`grafana.${DOMAIN}`)"
      - "traefik.http.routers.grafana.entrypoints=websecure"
      - "traefik.http.routers.grafana.tls=true"
      - "traefik.http.routers.grafana.tls.certresolver=letsencrypt"
      - "traefik.http.routers.grafana.middlewares=authelia@docker"
      - "traefik.http.services.grafana.loadbalancer.server.port=3000"

  # Node Exporter - Host metrics exporter
  # Metrics at: http://${SERVER_IP}:9100/metrics
  node-exporter:
    image: prom/node-exporter:v1.7.0
    container_name: node-exporter
    restart: unless-stopped
    networks:
      - homelab-network
    ports:
      - "9100:9100"
    volumes:
      - /proc:/host/proc:ro
      - /sys:/host/sys:ro
      - /:/rootfs:ro
    command:
      - '--path.procfs=/host/proc'
      - '--path.rootfs=/rootfs'
      - '--path.sysfs=/host/sys'
      - '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
    labels:
      - "homelab.category=monitoring"
      - "homelab.description=Hardware and OS metrics exporter"

  # cAdvisor - Container metrics exporter
  # Access at: http://${SERVER_IP}:8082
  cadvisor:
    image: gcr.io/cadvisor/cadvisor:v0.47.2
    container_name: cadvisor
    restart: unless-stopped
    networks:
      - homelab-network
      - traefik-network
    ports:
      - "8082:8080"
    volumes:
      - /:/rootfs:ro
      - /var/run:/var/run:ro
      - /sys:/sys:ro
      - /var/lib/docker:/var/lib/docker:ro
      - /dev/disk:/dev/disk:ro
    privileged: true
    devices:
      - /dev/kmsg
    labels:
      # TRAEFIK CONFIGURATION
      # ==========================================
      # Service metadata
      - "homelab.category=monitoring"
      - "homelab.description=Container metrics and performance monitoring"
      # Traefik reverse proxy (comment/uncomment to disable/enable)
      # If Traefik is on a remote server: these labels are NOT USED; 
      #   configure external yml files in /traefik/dynamic folder instead.
      - "traefik.enable=true"
      - "traefik.http.routers.cadvisor.rule=Host(`cadvisor.${DOMAIN}`)"
      - "traefik.http.routers.cadvisor.entrypoints=websecure"
      - "traefik.http.routers.cadvisor.tls=true"
      - "traefik.http.routers.cadvisor.tls.certresolver=letsencrypt"
      - "traefik.http.routers.cadvisor.middlewares=authelia@docker"
      - "traefik.http.services.cadvisor.loadbalancer.server.port=8080"

  # Uptime Kuma - Uptime monitoring
  # Access at: https://uptime-kuma.${DOMAIN}
  uptime-kuma:
    image: louislam/uptime-kuma:1
    deploy:
      resources:
        limits:
          cpus: '0.50'
          memory: 256M
          pids: 512
        reservations:
          cpus: '0.25'
          memory: 128M
    container_name: uptime-kuma
    restart: unless-stopped
    networks:
      - homelab-network
      - traefik-network
    ports:
      - "3001:3001"
    volumes:
      - uptime-kuma-data:/app/data
      - /var/run/docker.sock:/var/run/docker.sock:ro
    labels:
      # TRAEFIK CONFIGURATION
      # ==========================================
      # Service metadata
      - "homelab.category=monitoring"
      - "homelab.description=Service uptime monitoring and alerts"
      # Traefik reverse proxy (comment/uncomment to disable/enable)
      # If Traefik is on a remote server: these labels are NOT USED; 
      #   configure external yml files in /traefik/dynamic folder instead.
      - "traefik.enable=true"
      - "traefik.http.routers.uptime-kuma.rule=Host(`uptime-kuma.${DOMAIN}`)"
      - "traefik.http.routers.uptime-kuma.entrypoints=websecure"
      - "traefik.http.routers.uptime-kuma.tls=true"
      - "traefik.http.routers.uptime-kuma.tls.certresolver=letsencrypt"
      - "traefik.http.routers.uptime-kuma.middlewares=authelia@docker"
      - "traefik.http.services.uptime-kuma.loadbalancer.server.port=3001"

  # Loki - Log aggregation
  # Access at: http://${SERVER_IP}:3100
  loki:
    image: grafana/loki:2.9.3
    deploy:
      resources:
        limits:
          cpus: '0.75'
          memory: 512M
          pids: 1024
        reservations:
          cpus: '0.25'
          memory: 256M
    container_name: loki
    restart: unless-stopped
    networks:
      - homelab-network
      - traefik-network
    ports:
      - "3100:3100"
    volumes:
      - ./config/loki:/etc/loki
      - loki-data:/loki
    command: -config.file=/etc/loki/loki-config.yml
    labels:
      # TRAEFIK CONFIGURATION
      # ==========================================
      # Service metadata
      - "homelab.category=monitoring"
      - "homelab.description=Log aggregation system"
      # Traefik reverse proxy (comment/uncomment to disable/enable)
      # If Traefik is on a remote server: these labels are NOT USED; 
      #   configure external yml files in /traefik/dynamic folder instead.
      - "traefik.enable=true"
      - "traefik.http.routers.loki.rule=Host(`loki.${DOMAIN}`)"
      - "traefik.http.routers.loki.entrypoints=websecure"
      - "traefik.http.routers.loki.tls=true"
      - "traefik.http.routers.loki.tls.certresolver=letsencrypt"
      - "traefik.http.routers.loki.middlewares=authelia@docker"
      - "traefik.http.services.loki.loadbalancer.server.port=3100"

  # Promtail - Log shipper for Loki
  # Ships Docker container logs to Loki
  promtail:
    image: grafana/promtail:2.9.3
    container_name: promtail
    restart: unless-stopped
    networks:
      - homelab-network
    volumes:
      - ./config/promtail:/etc/promtail
      - /var/log:/var/log:ro
      - /var/lib/docker/containers:/var/lib/docker/containers:ro
    command: -config.file=/etc/promtail/promtail-config.yml
    depends_on:
      - loki
    labels:
      - "homelab.category=monitoring"
      - "homelab.description=Log collector for Loki"

volumes:
  prometheus-data:
    driver: local
  grafana-data:
    driver: local
  uptime-kuma-data:
    driver: local
  loki-data:
    driver: local

networks:
  homelab-network:
    external: true
  traefik-network:
    external: true

x-dockge:
  urls:
    # Proxied URLs (through Traefik)
    - http://${SERVER_IP}:9090
    - http://${SERVER_IP}:3000
    - https://uptime-kuma.${DOMAIN}
    - http://${SERVER_IP}:9100/metrics
    - http://${SERVER_IP}:8082
    - http://${SERVER_IP}:3100