第 14 章：Docker 与 Kubernetes

本章介绍如何在容器化环境中部署生产级 RabbitMQ，包括单节点、集群和 Kubernetes 方案。

14.1 Docker 单节点部署

基本部署

docker run -d \
  --name rabbitmq \
  -p 5672:5672 \
  -p 15672:15672 \
  -p 15692:15692 \
  -e RABBITMQ_DEFAULT_USER=admin \
  -e RABBITMQ_DEFAULT_PASS=admin123 \
  -v rabbitmq_data:/var/lib/rabbitmq \
  -v rabbitmq_log:/var/log/rabbitmq \
  rabbitmq:4-management

自定义配置文件

# rabbitmq.conf
listeners.tcp.default = 5672
management.tcp.port = 15672
prometheus.tcp.port = 15692

vm_memory_high_watermark.relative = 0.6
disk_free_limit.absolute = 1GB
heartbeat = 60
loopback_users.guest = true

# enabled_plugins
[rabbitmq_management,rabbitmq_prometheus,rabbitmq_delayed_message_exchange].

14.2 Docker Compose 单节点

version: '3.8'

services:
  rabbitmq:
    image: rabbitmq:4-management
    container_name: rabbitmq
    hostname: rabbitmq
    ports:
      - "5672:5672"
      - "15672:15672"
      - "15692:15692"
    environment:
      RABBITMQ_DEFAULT_USER: admin
      RABBITMQ_DEFAULT_PASS: admin123
      RABBITMQ_LOG_LEVEL: info
    volumes:
      - rabbitmq_data:/var/lib/rabbitmq
      - ./rabbitmq.conf:/etc/rabbitmq/rabbitmq.conf:ro
      - ./enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
    ulimits:
      nofile:
        soft: 65536
        hard: 65536
    deploy:
      resources:
        limits:
          memory: 2G
          cpus: '2'
    healthcheck:
      test: ["CMD", "rabbitmq-diagnostics", "-q", "check_running"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s
    restart: unless-stopped

volumes:
  rabbitmq_data:
    driver: local

14.3 Docker Compose 集群

version: '3.8'

services:
  rabbitmq1:
    image: rabbitmq:4-management
    container_name: rabbitmq1
    hostname: rabbitmq1
    ports:
      - "5672:5672"
      - "15672:15672"
    environment:
      RABBITMQ_ERLANG_COOKIE: "SWQOKODSQALRPCLNMEQG"
      RABBITMQ_DEFAULT_USER: admin
      RABBITMQ_DEFAULT_PASS: admin123
    volumes:
      - rabbitmq1_data:/var/lib/rabbitmq
      - ./rabbitmq.conf:/etc/rabbitmq/rabbitmq.conf:ro
      - ./enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
    networks:
      - rabbit_net
    healthcheck:
      test: ["CMD", "rabbitmq-diagnostics", "-q", "check_running"]
      interval: 30s
      timeout: 10s
      retries: 5
      start_period: 60s

  rabbitmq2:
    image: rabbitmq:4-management
    container_name: rabbitmq2
    hostname: rabbitmq2
    ports:
      - "5673:5672"
      - "15673:15672"
    environment:
      RABBITMQ_ERLANG_COOKIE: "SWQOKODSQALRPCLNMEQG"
    volumes:
      - rabbitmq2_data:/var/lib/rabbitmq
      - ./rabbitmq.conf:/etc/rabbitmq/rabbitmq.conf:ro
      - ./enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
    networks:
      - rabbit_net
    depends_on:
      rabbitmq1:
        condition: service_healthy

  rabbitmq3:
    image: rabbitmq:4-management
    container_name: rabbitmq3
    hostname: rabbitmq3
    ports:
      - "5674:5672"
      - "15674:15672"
    environment:
      RABBITMQ_ERLANG_COOKIE: "SWQOKODSQALRPCLNMEQG"
    volumes:
      - rabbitmq3_data:/var/lib/rabbitmq
      - ./rabbitmq.conf:/etc/rabbitmq/rabbitmq.conf:ro
      - ./enabled_plugins:/etc/rabbitmq/enabled_plugins:ro
    networks:
      - rabbit_net
    depends_on:
      rabbitmq1:
        condition: service_healthy

  # HAProxy 负载均衡
  haproxy:
    image: haproxy:2.8
    container_name: rabbitmq_haproxy
    ports:
      - "5670:5672"   # AMQP
      - "15670:15672" # Management UI
    volumes:
      - ./haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
    networks:
      - rabbit_net
    depends_on:
      - rabbitmq1
      - rabbitmq2
      - rabbitmq3

volumes:
  rabbitmq1_data:
  rabbitmq2_data:
  rabbitmq3_data:

networks:
  rabbit_net:
    driver: bridge

集群初始化脚本

#!/bin/bash
# init_cluster.sh

echo "等待 RabbitMQ 节点启动..."
sleep 10

echo "组建集群..."
# Node 2 加入集群
docker exec rabbitmq2 rabbitmqctl stop_app
docker exec rabbitmq2 rabbitmqctl reset
docker exec rabbitmq2 rabbitmqctl join_cluster rabbit@rabbitmq1
docker exec rabbitmq2 rabbitmqctl start_app

# Node 3 加入集群
docker exec rabbitmq3 rabbitmqctl stop_app
docker exec rabbitmq3 rabbitmqctl reset
docker exec rabbitmq3 rabbitmqctl join_cluster rabbit@rabbitmq1
docker exec rabbitmq3 rabbitmqctl start_app

echo "设置仲裁队列策略..."
# 无需设置（RabbitMQ 4.x 默认使用仲裁队列）

echo "验证集群状态..."
docker exec rabbitmq1 rabbitmqctl cluster_status

echo "集群搭建完成！"

HAProxy 配置

# haproxy.cfg
global
    log stdout format raw local0

defaults
    log global
    mode tcp
    timeout connect 5s
    timeout client 120s
    timeout server 120s

# AMQP 负载均衡
listen rabbitmq_amqp
    bind *:5672
    balance roundrobin
    option tcpka
    server rabbitmq1 rabbitmq1:5672 check inter 5s rise 2 fall 3
    server rabbitmq2 rabbitmq2:5672 check inter 5s rise 2 fall 3
    server rabbitmq3 rabbitmq3:5672 check inter 5s rise 2 fall 3

# Management UI 负载均衡
listen rabbitmq_management
    bind *:15672
    mode http
    balance roundrobin
    server rabbitmq1 rabbitmq1:15672 check
    server rabbitmq2 rabbitmq2:15672 check
    server rabbitmq3 rabbitmq3:15672 check

14.4 持久化配置

数据卷策略

数据	路径	持久化方式
消息数据	`/var/lib/rabbitmq`	Named Volume / PVC
日志	`/var/log/rabbitmq`	可选持久化
配置	`/etc/rabbitmq`	只读挂载

生产级挂载

volumes:
  - type: volume
    source: rabbitmq_data
    target: /var/lib/rabbitmq
    volume:
      nocopy: false
  - type: bind
    source: ./rabbitmq.conf
    target: /etc/rabbitmq/rabbitmq.conf
    read_only: true
  - type: bind
    source: ./advanced.config
    target: /etc/rabbitmq/advanced.config
    read_only: true
  - type: bind
    source: ./enabled_plugins
    target: /etc/rabbitmq/enabled_plugins
    read_only: true

14.5 健康检查

Docker 健康检查

healthcheck:
  test:
    - "CMD"
    - "rabbitmq-diagnostics"
    - "-q"
    - "check_running"
  interval: 30s
  timeout: 10s
  retries: 3
  start_period: 60s

常用健康检查命令

命令	说明	适用场景
`check_running`	检查节点是否运行	基础检查
`check_local_alarms`	检查本地告警	内存/磁盘检查
`check_port_connectivity`	检查端口连通性	网络检查
`check_if_node_is_quorum_critical`	检查仲裁队列状态	集群检查

组合健康检查

healthcheck:
  test: ["CMD-SHELL", "rabbitmq-diagnostics -q check_running && rabbitmq-diagnostics -q check_local_alarms"]
  interval: 30s
  timeout: 10s
  retries: 3
  start_period: 60s

14.6 Kubernetes 部署

使用 RabbitMQ Cluster Operator

# 安装 Operator
kubectl apply -f "https://github.com/rabbitmq/cluster-operator/releases/latest/download/cluster-operator.yml"

# 验证安装
kubectl get pods -n rabbitmq-system

RabbitmqCluster CRD

apiVersion: rabbitmq.com/v1beta1
kind: RabbitmqCluster
metadata:
  name: production-rabbitmq
  namespace: messaging
spec:
  replicas: 3
  
  image: rabbitmq:4-management
  
  persistence:
    storageClassName: fast-ssd
    storage: 50Gi
  
  resources:
    requests:
      cpu: "500m"
      memory: "1Gi"
    limits:
      cpu: "2"
      memory: "4Gi"
  
  rabbitmq:
    additionalConfig: |
      vm_memory_high_watermark.relative = 0.6
      disk_free_limit.absolute = 2GB
      heartbeat = 60
      queue_leader_locator = balanced
    additionalPlugins:
      - rabbitmq_prometheus
      - rabbitmq_delayed_message_exchange
  
  override:
    statefulSet:
      spec:
        template:
          spec:
            containers:
              - name: rabbitmq
                env:
                  - name: RABBITMQ_DEFAULT_USER
                    value: admin
                  - name: RABBITMQ_DEFAULT_PASS
                    valueFrom:
                      secretKeyRef:
                        name: rabbitmq-secret
                        key: password
  
  tls:
    secretName: rabbitmq-tls
    caSecretName: rabbitmq-ca
  
  affinity:
    podAntiAffinity:
      preferredDuringSchedulingIgnoredDuringExecution:
        - weight: 100
          podAffinityTerm:
            labelSelector:
              matchLabels:
                app.kubernetes.io/name: production-rabbitmq
            topologyKey: kubernetes.io/hostname

部署和管理

# 部署集群
kubectl apply -f rabbitmq-cluster.yaml

# 查看状态
kubectl get rabbitmqclusters -n messaging
kubectl get pods -l app.kubernetes.io/name=production-rabbitmq -n messaging

# 获取访问密码
kubectl get secret production-rabbitmq-default-user \
  -o jsonpath='{.data.password}' -n messaging | base64 --decode

# 端口转发（调试）
kubectl port-forward svc/production-rabbitmq 15672:15672 -n messaging

# 查看集群状态
kubectl exec -it production-rabbitmq-server-0 -n messaging -- \
  rabbitmqctl cluster_status

Service 配置

apiVersion: v1
kind: Service
metadata:
  name: rabbitmq-service
  namespace: messaging
spec:
  type: ClusterIP
  ports:
    - name: amqp
      port: 5672
      targetPort: 5672
    - name: management
      port: 15672
      targetPort: 15672
    - name: prometheus
      port: 15692
      targetPort: 15692
  selector:
    app.kubernetes.io/name: production-rabbitmq

14.7 K8s 资源配置

ConfigMap

apiVersion: v1
kind: ConfigMap
metadata:
  name: rabbitmq-config
  namespace: messaging
data:
  rabbitmq.conf: |
    listeners.tcp.default = 5672
    management.tcp.port = 15672
    vm_memory_high_watermark.relative = 0.6
    disk_free_limit.absolute = 2GB
    heartbeat = 60
  enabled_plugins: |
    [rabbitmq_management,rabbitmq_prometheus].

Secret

apiVersion: v1
kind: Secret
metadata:
  name: rabbitmq-secret
  namespace: messaging
type: Opaque
stringData:
  password: "secure_password_here"
  erlang_cookie: "SWQOKODSQALRPCLNMEQG"

14.8 注意事项

⚠️ 文件描述符限制

Docker 容器默认的 ulimit 可能不够。必须设置 --ulimit nofile=65536:65536。

⚠️ Erlang Cookie 一致性

集群中所有节点的 Erlang Cookie 必须完全一致。

⚠️ 时钟同步

集群节点之间需要时钟同步（NTP），否则可能出现消息顺序问题。

⚠️ Pod 反亲和性

在 K8s 中使用反亲和性确保 RabbitMQ Pod 分布在不同节点上。

⚠️ 存储性能

RabbitMQ 对磁盘 I/O 较敏感，建议使用 SSD 存储。

🔥 最佳实践: 生产环境使用 RabbitMQ Cluster Operator + 3 副本 + SSD 持久化 + Pod 反亲和性 + Prometheus 监控。

14.9 扩展阅读

下一章: 第 15 章：故障排查 — 快速定位和解决常见问题。

强曰为道

第 14 章：Docker 与 Kubernetes

第 14 章：Docker 与 Kubernetes

14.1 Docker 单节点部署

基本部署

推荐的完整配置

自定义配置文件

14.2 Docker Compose 单节点

14.3 Docker Compose 集群

集群初始化脚本

HAProxy 配置

14.4 持久化配置

数据卷策略

生产级挂载

14.5 健康检查

Docker 健康检查

常用健康检查命令

组合健康检查

14.6 Kubernetes 部署

使用 RabbitMQ Cluster Operator

RabbitmqCluster CRD

部署和管理

Service 配置

14.7 K8s 资源配置

ConfigMap

Secret

14.8 注意事项

14.9 扩展阅读