Helm 实现高可用 Ingress 部署攻略

Tip：摘抄GitHub ingress-nginx项目说明，具体链接地址。请查看下面的参考文档

• ingress-nginx部署两个副本数，使用pod反亲和。将两个pod调度到不同的节点上。

• ingress-nginx前面挂一个负载均衡器，流量平分两个副本上

1、下载ingress-nginx chart包

$ helm repo add ingress-nginx https://kubernetes.github.io/ingress-nginx
"ingress-nginx" has been added to your repositories

$ helm pull ingress-nginx/ingress-nginx --untar --untardir /etc/kubernetes/addons --version 4.11.2

2、部署 ingress-nginx 配置文件

$ cat<<'EOF' | sudo tee /etc/kubernetes/addons/ingress-nginx-value.yml > /dev/null
controller:
  # 两个副本数
  replicaCount: 2

  # 部署模式
  kind: Deployment
  image:
    registry: 172.139.20.170:5000
    image: library/controller
    tag: "v1.11.2"
    digest: ''

  # pod反亲和
  podAntiAffinity:
    requiredDuringSchedulingIgnoredDuringExecution:
    - labelSelector:
        matchExpressions:
        - key: app.kubernetes.io/component
          operator: In
          values:
          - controller
      topologyKey: kubernetes.io/hostname

  # 标签选择器
  nodeSelector:
    ingress-nginx: controller

  # service配置
  service:
    type: NodePort
    externalTrafficPolicy: Local
    nodePorts:
      http: "30080"
      https: "30443"

  # 配置metrics采集
  metrics:
    enabled: true
    port: 10254

  # 优雅推出
  lifecycle:
    preStop:
      exec:
        command:
          - /wait-shutdown

  admissionWebhooks:
    patch:
      enabled: true
      image:
        registry: 172.139.20.170:5000
        image: library/kube-webhook-certgen
        tag: 'v1.4.3'
        digest: ''
EOF

3、部署ingress-nginx

$ kubectl label node k8s-node01 ingress-nginx=controller 
node/k8s-node01 labeled

$ kubectl label node k8s-node02 ingress-nginx=controller
node/k8s-node02 labeled

$ helm install -n kube-system ingress-nginx -f /etc/kubernetes/addons/ingress-nginx-value.yml /etc/kubernetes/addons/ingress-nginx  

1、负载ingress-nginx流量

listen ingress-http-tcp
  bind *:80
  server ingress01 172.139.20.175:30080 maxconn 32 check
  server ingress02 172.139.20.75:30080  maxconn 32 check
listen ingress-https-tcp
  bind *:443
  server ingress01 172.139.20.175:30443 maxconn 32 check
  server ingress02 172.139.20.75:30443  maxconn 32 check

2、修改docker-compose文件

$ cat /etc/haproxy/docker-compose.yml 
name: haproxy
services:
  haproxy:
    container_name: haproxy
    image: haproxy:2.9-alpine
    restart: always
    volumes:
    - ./haproxy.cfg:/usr/local/etc/haproxy/haproxy.cfg:ro
    ports:
    # 添加以下4行配置
    - 80:80
    - 443:443
    sysctls:
    - net.ipv4.ip_unprivileged_port_start=0

Tip：net.ipv4.ip_unprivileged_port_start=0参数，需要内核4.x版本

3、重启haproxy服务

$ sudo docker-compose -f /etc/haproxy/docker-compose.yml restart

1、验证pod状态

$ kubectl -n kube-system get pod -l app.kubernetes.io/instance=ingress-nginx
NAME                                       READY   STATUS    RESTARTS   AGE
ingress-nginx-controller-579f7d8f4-8jmlw   1/1     Running   0          3h42m
ingress-nginx-controller-579f7d8f4-tf5wc   1/1     Running   0          3h42m

通过 Helm 部署高可用的 Ingress 控制器不仅简化了配置流程，还增强了系统的健壮性。无论是在开发还是生产环境中，这样的部署方式都能帮助我们更好地管理应用的流量，确保服务的连续性和可靠性。

别忘了，关注我们的公众号，获取更多关于容器技术和云原生领域的深度洞察和技术实战，让我们携手在技术的海洋中乘风破浪！