kubernetes生产实践之redis-cluster

##方案一 自定义yaml文件安装redis cluster

背景

在Kubernetes中部署Redis集群面临挑战，因为每个Redis实例都依赖于一个配置文件，该文件可以跟踪其他集群实例及其角色。为此，需要结合使用Kubernetes StatefulSets和PersistentVolumes实现。
redis cluster架构示意图：

创建StatefulSet yaml文件

---
apiVersion: v1
kind: ConfigMap
metadata:
  name: redis-cluster
data:
  update.sh: |
    #!/bin/sh
    REDIS_NODES="/data/nodes.conf"
    sed -i -e "/myself/ s/[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}\.[0-9]\{1,3\}/${POD_IP}/" ${REDIS_NODES}
    exec "$@"
  redis.conf: |+
    bind 0.0.0.0
    cluster-enabled yes
    cluster-require-full-coverage no
    cluster-node-timeout 30000
    cluster-config-file /data/nodes.conf
    cluster-migration-barrier 1
    appendonly yes
    protected-mode no
---
apiVersion: apps.kruise.io/v1beta1
# apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: redis-cluster
spec:
  serviceName: redis-cluster
  replicas: 6
  selector:
    matchLabels:
      app: redis-cluster
  template:
    metadata:
      labels:
        app: redis-cluster
    spec:
      containers:
      - name: redis
        image: redis:6.2.1-alpine
        ports:
        - containerPort: 6379
          name: client
        - containerPort: 16379
          name: gossip
        command: ["/conf/update.sh", "redis-server", "/conf/redis.conf"]
        env:
        - name: POD_IP
          valueFrom:
            fieldRef:
              fieldPath: status.podIP
        volumeMounts:
        - name: conf
          mountPath: /conf
          readOnly: false
        - name: data
          mountPath: /data
          readOnly: false
      volumes:
      - name: conf
        configMap:
          name: redis-cluster
          defaultMode: 0755
  volumeClaimTemplates:
  - metadata:
      name: data
    spec:
      accessModes: [ "ReadWriteOnce" ]
      resources:
        requests:
          storage: 100Gi
      storageClassName: rbd

注释：

cluster-migration-barrier 那些分配后仍然剩余migration barrier个从节点的主节点才会触发节点分配，而不是分配前有migration barrier个从节点的主节点就会触发节点分配，默认是1，生产环境建议维持默认值
protected-mode no 参数是为了禁止外网访问redis，如果启用了，则只能够通过lookback ip（127.0.0.1）访问Redis，如果外网访问redis，会报出异常。
apiVersion: apps.kruise.io/v1beta1 控制器这里使用kruise提供的Advanced StatefulSet，如果集群没有安装kruise，可以使用 apps/v1

安装redis-cluster

[root@qd01-stop-k8s-master001 redis]# kubectl apply -f install-redis.yaml
configmap/redis-cluster created
statefulset.apps.kruise.io/redis-cluster created

[root@qd01-stop-k8s-master001 redis]# kubectl get po -n op
NAME              READY   STATUS    RESTARTS   AGE
redis-cluster-0   1/1     Running   0          3m26s
redis-cluster-1   1/1     Running   0          3m14s
redis-cluster-2   1/1     Running   0          2m54s
redis-cluster-3   1/1     Running   0          2m23s
redis-cluster-4   1/1     Running   0          2m14s
redis-cluster-5   1/1     Running   0          114s

创建redis-cluster service

---
apiVersion: v1
kind: Service
metadata:
  name: redis-cluster
  namespace: op
spec:
  type: ClusterIP
  ports:
  - port: 6379
    targetPort: 6379
    name: client
  - port: 16379
    targetPort: 16379
    name: gossip
  selector:
    app: redis-cluster

[root@qd01-stop-k8s-master001 redis]# kubectl apply -f redis-svc.yml
service/redis-cluster created
[root