Docker avec HAProxy - Architecture micro-services

Sommaire

Projet pour une infrastructure Docker avec un proxy inversé (HAProxy), des services web (Hugo, PlantUML), des outils de supervision (Grafana, Prometheus, Node Exporter) et un tableau de bord de disponibilité (Uptime Kuma).

Objectifs #

Centraliser l’accès à plusieurs services internes via un seul proxy (HAProxy).
Surveiller l’état des services avec Prometheus + Grafana.
Visualiser la disponibilité grâce à Uptime Kuma.
Déployer un site statique Hugo optimisé derrière Varnish et HAProxy.

Aperçu architectural #

graph TD; subgraph Internet A[Utilisateur] end subgraph ReverseProxy B[HAProxy] V[Varnish] end subgraph Apps H[Hugo Nginx] K[Uptime Kuma] G[Grafana] P[Prometheus] N[Node Exporter] U[UnifiedPush] M[PlantUML] end A --> B B --> V V --> H B --> K B --> G B --> U B --> M G --> P P --> N

Stack Docker Compose #

`docker-compose.yml` #

version: "3.9"

services:

  haproxy:
    image: haproxy:2.9
    container_name: haproxy
    ports:
      - "80:80"
      - "443:443"
      - "8404:8404"
    volumes:
      - ./haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
      - ./certs:/usr/local/etc/haproxy/certs:ro
    networks:
      - reverse_proxy
    restart: unless-stopped

  hugo_nginx:
    image: nginx:alpine
    container_name: hugo_nginx
    volumes:
      - ./site:/usr/share/nginx/html:ro
    networks:
      - reverse_proxy

  plantuml:
    image: plantuml/plantuml-server:latest
    container_name: plantuml
    networks:
      - reverse_proxy

  uptime_kuma:
    image: louislam/uptime-kuma:1
    container_name: uptime_kuma
    volumes:
      - uptime_kuma_data:/app/data
    networks:
      - reverse_proxy

  prometheus:
    image: prom/prometheus:v3.6.0
    container_name: prometheus
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml:ro
    networks:
      - reverse_proxy

  node_exporter:
    image: prom/node-exporter:v1.9.0
    container_name: node_exporter
    command:
      - '--path.rootfs=/host'
    pid: host
    volumes:
      - /:/host:ro,rslave
    networks:
      - reverse_proxy

  grafana:
    image: grafana/grafana:12.2.0
    container_name: grafana
    volumes:
      - grafana_data:/var/lib/grafana
    depends_on:
      - prometheus
    networks:
      - reverse_proxy

networks:
  reverse_proxy:
    driver: bridge

volumes:
  grafana_data:
  uptime_kuma_data:

Extrait de configuration HAProxy #

global
  stats socket /run/haproxy/admin.sock mode 660 level admin
  log stdout format raw local0

defaults
  log global
  mode http
  timeout connect 5s
  timeout client  30s
  timeout server  30s

frontend http_in
  bind *:80
  redirect scheme https

frontend https_in
  bind *:443 ssl crt /usr/local/etc/haproxy/certs/
  acl host_hugo hdr(host) -i hugo.monsite.fr
  acl host_grafana hdr(host) -i grafana.monsite.fr
  acl host_plantuml hdr(host) -i plantuml.monsite.fr
  acl host_kuma hdr(host) -i status.monsite.fr

  use_backend hugo_backend if host_hugo
  use_backend grafana_backend if host_grafana
  use_backend plantuml_backend if host_plantuml
  use_backend kuma_backend if host_kuma

backend hugo_backend
  server hugo hugo_nginx:80 check

backend grafana_backend
  server grafana grafana:3000 check

backend plantuml_backend
  server plantuml plantuml:8080 check

backend kuma_backend
  server kuma uptime_kuma:3001 check

listen stats
  bind *:8404
  stats enable
  stats uri /stats

Lancement #

docker compose up -d

https://hugo.monsite.fr → Portfolio
https://grafana.monsite.fr → Monitoring et dashboards
https://plantuml.monsite.fr → Online PlantUML editor

Supervision #

Prometheus collecte les métriques de node_exporter, haproxy et grafana.
Grafana affiche les tableaux de bord de performance.
Uptime Kuma surveille en continu les URL exposées. Note : Ce projet est inspiré de l’environnement de production réel, simplifié à des fins de démonstration dans ce portfolio.