Kubernetes环境部署Redis集群过程

kind: ConfigMap
apiVersion: v1
metadata:
  name: redis-cluster-cm
data:
  init-redis-cluster.sh: |2-
        #!/bin/bash
        set -e

        # 直接在脚本中声明变量
        REDIS_PASSWORD="ClusterRedis#20251112"
        SERVICE_NAME="redis-cluster"

        NODES=(
          "redis-cluster-0.$SERVICE_NAME:6379"
          "redis-cluster-1.$SERVICE_NAME:6379"
          "redis-cluster-2.$SERVICE_NAME:6379"
          "redis-cluster-3.$SERVICE_NAME:6379"
          "redis-cluster-4.$SERVICE_NAME:6379"
          "redis-cluster-5.$SERVICE_NAME:6379"
        )

        echo "检查所有 Redis 节点是否就绪..."
        all_ready=1
        for node in "${NODES[@]}"; do
          host=$(echo "$node" | cut -d: -f1)
          port=$(echo "$node" | cut -d: -f2)
          echo "检查节点: $host:$port"
          if [ "$(redis-cli -h $host -p $port -a $REDIS_PASSWORD ping 2>/dev/null)" != "PONG" ]; then
            echo "节点 $node 未就绪"
            all_ready=0
            break
          else
            echo "节点 $node 已就绪"
          fi
        done

        if [ $all_ready -ne 1 ]; then
          echo "有节点未就绪，退出"
          exit 0
        fi

        echo "检查集群状态..."
        cluster_state=$(redis-cli -h ${NODES[0]%:*} -p ${NODES[0]#*:} -a $REDIS_PASSWORD cluster info 2>/dev/null | grep "cluster_state" | cut -d: -f2 | tr -d '[:space:]')
        if [ "$cluster_state" = "ok" ]; then
          echo "集群已初始化，状态正常"
          exit 0
        fi

        echo "开始初始化 Redis 集群..."
        redis-cli -a $REDIS_PASSWORD \
          --cluster create \
          --cluster-replicas 1 \
          ${NODES[@]} 

        
        echo "集群初始化完成"
        
  redis-cluster.conf: |
    bind 0.0.0.0
    port 6379
    daemonize no
    protected-mode no
    dir /data
    cluster-announce-bus-port 16379
    cluster-enabled yes
    cluster-node-timeout 15000
    cluster-config-file /data/nodes.conf
    requirepass ClusterRedis#20251112
    masterauth ClusterRedis#20251112

    # backup settings
    # aof backup mode
    save 3600 1
    stop-writes-on-bgsave-error no
    appendonly yes
    appendfilename "appendonly.aof"
    appendfsync everysec
    no-appendfsync-on-rewrite no
    auto-aof-rewrite-percentage 100
    auto-aof-rewrite-min-size 64mb
    aof-load-truncated yes
    maxmemory 16G
    maxmemory-policy allkeys-lru

#!/bin/bash
# 1. 指定脚本解释器为bash，确保在K8s容器（通常为Linux环境）中可正常执行
set -e
# 2. 容错配置：脚本中任一命令返回非0状态码（执行失败）时，立即退出脚本
#    避免错误延续（如某节点未就绪仍强行创建集群，导致集群初始化失败）

# -------------------------- 变量定义部分 --------------------------
# 3. Redis密码变量：统一管理客户端连接与主从同步密码，避免硬编码分散
#    密码含大小写、数字、特殊字符（ClusterRedis#20251112），满足基础安全复杂度，防止未授权访问
REDIS_PASSWORD="ClusterRedis#20251112"
# 4. K8s Service名变量：对应Redis集群的Service名称（redis-cluster）
#    后续节点地址拼接时使用Service域名（如redis-cluster-0.redis-cluster），适配K8s Pod动态IP特性
SERVICE_NAME="redis-cluster"
# 5. 集群节点列表：定义6个节点（3主3从架构，Redis Cluster最小高可用架构）
#    地址格式为“Pod名.Service名:端口”，K8s中通过该域名可稳定访问各Pod（无需依赖动态IP）
NODES=(
  "redis-cluster-0.$SERVICE_NAME:6379"
  "redis-cluster-1.$SERVICE_NAME:6379"
  "redis-cluster-2.$SERVICE_NAME:6379"
  "redis-cluster-3.$SERVICE_NAME:6379"
  "redis-cluster-4.$SERVICE_NAME:6379"
  "redis-cluster-5.$SERVICE_NAME:6379"
)

# -------------------------- 节点就绪检查部分 --------------------------
echo "检查所有 Redis 节点是否就绪..."
all_ready=1
# 6. 循环遍历所有节点：确保集群初始化前，所有节点均已启动并能正常响应
for node in "${NODES[@]}"; do
  # 7. 拆分节点地址为“主机名”和“端口”：从节点字符串（如xxx:6379）中提取关键信息，用于后续ping检查
  host=$(echo "$node" | cut -d: -f1)
  port=$(echo "$node" | cut -d: -f2)
  echo "检查节点: $host:$port"
  # 8. Redis节点可用性检查：通过redis-cli ping命令判断节点是否存活
  #    -h $host：指定节点主机名（K8s Service域名）
  #    -p $port：指定Redis客户端端口（6379，与配置文件一致）
  #    -a $REDIS_PASSWORD：携带密码认证（避免未授权访问拦截）
  #    2>/dev/null：屏蔽错误输出（如连接超时的报错），仅通过“PONG”判断状态，避免日志冗余
  if [ "$(redis-cli -h $host -p $port -a $REDIS_PASSWORD ping 2>/dev/null)" != "PONG" ]; then
    echo "节点 $node 未就绪"
    all_ready=0
    break
  else
    echo "节点 $node 已就绪"
  fi
done

# 9. 未就绪节点处理：若存在未就绪节点，脚本以0状态码退出（非错误退出）
#    原因：K8s会将非0退出视为“脚本执行失败”并重启Pod，此处用exit 0避免无效重启
#    后续Pod重启时，脚本会重新执行检查，直到所有节点就绪
if [ $all_ready -ne 1 ]; then
  echo "有节点未就绪，退出"
  exit 0
fi

# -------------------------- 集群状态判断部分 --------------------------
echo "检查集群状态..."
# 10. 获取集群状态：通过第一个节点的cluster info命令获取集群状态（cluster_state）
#    原因：Redis Cluster中所有节点的集群状态全局一致，无需遍历所有节点，减少检查开销
#    ${NODES[0]%:*}：提取第一个节点的主机名（删除字符串最后一个“:”及后面内容）
#    ${NODES[0]#*:}：提取第一个节点的端口（删除字符串第一个“:”及前面内容）
#    tr -d '[:space:]'：去除结果中的空格，避免空格导致状态判断失效
cluster_state=$(redis-cli -h ${NODES[0]%:*} -p ${NODES[0]#*:} -a $REDIS_PASSWORD cluster info 2>/dev/null | grep "cluster_state" | cut -d: -f2 | tr -d '[:space:]')

# 11. 已初始化集群处理：若集群状态为“ok”（已完成初始化），脚本退出
#    原因：K8s Pod可能因重启（如配置更新、节点故障恢复）重新执行脚本
#    重复执行--cluster create会导致集群元数据混乱（如节点角色冲突），需避免
if [ "$cluster_state" = "ok" ]; then
  echo "集群已初始化，状态正常"
  exit 0
fi

# -------------------------- 集群初始化部分 --------------------------
echo "开始初始化 Redis 集群..."
# 12. 集群创建命令：核心指令，完成3主3从集群的自动初始化
#    -a $REDIS_PASSWORD：携带密码，确保初始化过程中认证通过
#    --cluster create：触发Redis Cluster创建流程
#    --cluster-replicas 1：指定“1主1从”架构，6个节点自动分配为3主3从（按节点顺序：0/1/2为主，3/4/5为从）
#    优势：无需手动指定主从关系，适配K8s自动化部署场景，减少运维操作
redis-cli -a $REDIS_PASSWORD \
  --cluster create \
  --cluster-replicas 1 \
  ${NODES[@]} 

echo "集群初始化完成"

kind: Service
apiVersion: v1
metadata:
  name: redis-cluster
spec:
  ports:
    - name: redis
      protocol: TCP
      port: 6379
      targetPort: 6379
    - name: election
      protocol: TCP
      port: 16379
      targetPort: 16379
  selector:
    app: redis-cluster
  clusterIP: None
  type: ClusterIP

kind: StatefulSet
apiVersion: apps/v1
metadata:
  name: redis-cluster
  labels:
    app: redis-cluster
spec:
  replicas: 6
  selector:
    matchLabels:
      app: redis-cluster
  template:
    metadata:
      labels:
        app: redis-cluster
    spec:
      volumes:
        - name: redis-conf
          configMap:
            name: redis-cluster-cm
            items:
              - key: redis-cluster.conf
                path: redis-cluster.conf
            defaultMode: 420
        - name: redis-init
          configMap:
            name: redis-cluster-cm
            items:
              - key: init-redis-cluster.sh
                path: init-redis-cluster.sh
            defaultMode: 420
      containers:
        - name: redis-cluster
          image: 'redis:7.2.6-amd64'
          args:
            - /etc/redis/redis-cluster.conf
            - '--cluster-announce-ip "$(POD_NAME).$(POD_SERVICE_NAME)"'
          ports:
            - name: redis
              containerPort: 6379
              protocol: TCP
            - name: election
              containerPort: 16379
              protocol: TCP
          env:
            - name: TZ
              value: Asia/Shanghai
            - name: POD_NAME
              valueFrom:
                fieldRef:
                  apiVersion: v1
                  fieldPath: metadata.name
            - name: POD_SERVICE_NAME
              value: redis-cluster
          resources:
            limits:
              cpu: '1'
              memory: 2Gi
            requests:
              cpu: 500m
              memory: 1Gi
          volumeMounts:
            - name: redis-conf
              mountPath: /etc/redis
            - name: redis-init
              mountPath: /redis-init
            - name: redis-cluster-data
              mountPath: /data
          livenessProbe:
            tcpSocket:
              port: redis
            initialDelaySeconds: 16
            timeoutSeconds: 1
            periodSeconds: 3
            successThreshold: 1
            failureThreshold: 2
          readinessProbe:
            tcpSocket:
              port: redis
            initialDelaySeconds: 16
            timeoutSeconds: 1
            periodSeconds: 3
            successThreshold: 1
            failureThreshold: 2
          imagePullPolicy: IfNotPresent
      restartPolicy: Always
      affinity:
        podAntiAffinity:
          preferredDuringSchedulingIgnoredDuringExecution:
            - weight: 100
              podAffinityTerm:
                labelSelector:
                  matchExpressions:
                    - key: app
                      operator: In
                      values:
                        - redis-cluster
                topologyKey: kubernetes.io/hostname
      schedulerName: default-scheduler
  volumeClaimTemplates:
    - kind: PersistentVolumeClaim
      apiVersion: v1
      metadata:
        name: redis-cluster-data
      spec:
        accessModes:
          - ReadWriteMany
        resources:
          requests:
            storage: 10Gi
        storageClassName: nfs
        volumeMode: Filesystem
  serviceName: redis-cluster

affinity:
  podAntiAffinity:
    preferredDuringSchedulingIgnoredDuringExecution:  # 优先调度策略（非强制）
      - weight: 100  # 权重100（最高优先级）
        podAffinityTerm:
          labelSelector:  # 匹配标签为 app=redis-cluster 的 Pod
            matchExpressions:
              - key: app
                operator: In
                values: [redis-cluster]
          topologyKey: kubernetes.io/hostname  # 按主机（Node）划分拓扑域

kubectl apply -f redis-cluster/*.yaml

# exec 进入容器
kubectl exec -it redis-cluster-0 -- /bin/bash

# 初始化集群
cd /redis-init
bash init-redis-cluster.sh

查看集群状态、测试数据读写，进入任意一个pod执行:

```bash
# kubectl -n test exec -it redis-cluster-sts-0 -- bash
root@redis-cluster-sts-0:/data# redis-cli -a ClusterRedis#20251112 -c  
  
Warning: Using a password with '-a' or '-u' option on the command line interface may not be safe.
​
# CLUSTER NODES 查看集群节点，正常
127.0.0.1:6379> CLUSTER NODES

​
# CLUSTER INFO 查看集群信息：cluster_state:ok表示集群正常
127.0.0.1:6379> CLUSTER INFO

# 写入，读取key， 正常
127.0.0.1:6379> set boo foo
127.0.0.1:6379> get boo
"foo"