K8s -redis集群-哨兵sentinels集（外部访问）20260812

master:192.168.91.166

node2:192.168.91.167

node3:192.168.91.168

一、所有节点创建目录

二、创建命名空间

kubectl create namespace redis-cluster

##kubectl get ns

三、创建PV

###创建PV

kubectl apply -f redis-pv.yaml

####查看

kubectl get pv

redis-pv.yaml内容:

apiVersion: v1
kind: PersistentVolume
metadata:
name: redis-pv-0
spec:
capacity:
storage: 1Gi
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Retain
storageClassName: local-path
local:
path: /data/redis/redis0
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- k8s-master
---
apiVersion: v1
kind: PersistentVolume
metadata:
name: redis-pv-1
spec:
capacity:
storage: 1Gi
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Retain
storageClassName: local-path
local:
path: /data/redis/redis1
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- k8s-node2
---
apiVersion: v1
kind: PersistentVolume
metadata:
name: redis-pv-2
spec:
capacity:
storage: 1Gi
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Retain
storageClassName: local-path
local:
path: /data/redis/redis2
nodeAffinity:
required:
nodeSelectorTerms:
- matchExpressions:
- key: kubernetes.io/hostname
operator: In
values:
- k8s-node3

四、创建配置文件ConfigMap

（ConfigMap：redis 主配置 + sentinel 哨兵配置）

#####创建配置文件ConfigMap

kubectl apply -f redis-config.yml

#####基础查看：确认资源存在

kubectl get configmap -n redis-cluster

redis-config.yaml内容:

apiVersion: v1
kind: ConfigMap
metadata:
name: redis-conf
namespace: redis-cluster
data:
redis.conf: |
port 6379
requirepass Redis@123456
masterauth Redis@123456
daemonize no
logfile /data/redis.log
dir /data
appendonly yes
protected-mode no
tcp-keepalive 60

# 允许从节点同步密码

replrequirepass Redis@123456

sentinel.conf: |
port 26379
sentinel monitor redis-master redis-0.redis-service.redis-cluster.svc.cluster.local 6379 2
sentinel auth-pass redis-master Redis@123456
sentinel down-after-milliseconds redis-master 30000
sentinel parallel-syncs redis-master 1
sentinel failover-timeout redis-master 180000

port 26379
# 监控主节点集群名、主节点域名、端口、法定投票数2
sentinel monitor redis-master redis-0.redis-service.redis-cluster.svc.cluster.local 6379 2
sentinel auth-pass redis-master Redis@123456
# 30秒无响应判定主观下线
sentinel down-after-milliseconds redis-master 30000
# 故障转移时同时同步从节点数量
sentinel parallel-syncs redis-master 1
# 故障转移总超时
sentinel failover-timeout redis-master 180000
# 禁止哨兵自动重写配置文件（容器只读环境推荐加上）
sentinel config-epoch redis-master 0

已把哨兵监控域名统一为 redis-service（和你的无头服务名称匹配，之前名称不一致 bug 已修复）

五、创建服务（Headless 无头服务，内部集群通信）

##创建内部无头服务

kubectl apply -f redis-service.yaml

##查看

kubectl get svc -n redis-cluster

redis-service.yaml内容:

apiVersion: v1
kind: Service
metadata:
name: redis-service
namespace: redis-cluster
spec:
selector:
app: redis
ports:
- port: 6379
targetPort: 6379
clusterIP: None

六、创建StatefulSet 一主两从主体，已挂载自定义配置

#部署Redis有状态集群
kubectl apply -f redis-sts.yaml

###查看 StatefulSet 整体状态

kubectl get sts -n redis-cluster

###查看生成的 3 个 Redis Pod（最核心）

kubectl get pods -n redis-cluster

###查看 PVC 绑定情况（核对 PV 是否匹配上）

kubectl get pv,pvc -n redis-cluster

进入容器内部校验配置 & 挂载是否生效

redis-sts.yaml内容:

# Redis StatefulSet 主体
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: redis
namespace: redis-cluster
spec:
replicas: 3
serviceName: redis-service
selector:
matchLabels:
app: redis
template:
metadata:
labels:
app: redis
spec:
# 容忍master控制平面污点，仅本Redis Pod可调度master
tolerations:
- key: node-role.kubernetes.io/control-plane
operator: Exists
effect: NoSchedule
# 容忍磁盘压力污点
- key: node.kubernetes.io/disk-pressure
operator: Exists
effect: NoSchedule
containers:
- name: redis
image: swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/redis:7-alpine
ports:
- containerPort: 6379
name: redis-port
# 数据持久化挂载
volumeMounts:
- name: redis-data
mountPath: /data
# 挂载自定义redis配置文件
- name: redis-config
mountPath: /etc/redis
# 启动加载自定义redis.conf
command: ["redis-server", "/etc/redis/redis.conf"]
# 引入configmap配置卷
volumes:
- name: redis-config
configMap:
name: redis-conf

# 自动创建PVC模板
volumeClaimTemplates:
- metadata:
name: redis-data
spec:
accessModes: [ "ReadWriteOnce" ]
storageClassName: local-path
resources:
requests:
storage: 1Gi

七、创建服务（外网服务）

redis-service-out.yaml（NodePort 外网访问入口）

####开启外网访问端口

kubectl apply -f redis-service-out.yaml

####查看所有 service 列表

kubectl get svc -n redis-cluster

redis-service-out.yaml内容:

apiVersion: v1
kind: Service
metadata:
name: redis-nodeport
namespace: redis-cluster
spec:
selector:
app: redis
type: NodePort
ports:
- port: 6379
targetPort: 6379
nodePort: 30084

八、集群外部访问（任选集群k8节点 IP）

九、创建哨兵sentinel集群:Deployment

####创建哨兵群

kubectl apply -f redis-sentinel.yaml

####观察状态

两步验证哨兵集群有效性

任意哨兵客户端查看集群状态

kubectl exec -it redis-sentinel-5d5cd9fdf-22m2j -n redis-cluster -- redis-cli -p 26379

哨兵内部执行查询命令

# 查看主节点完整信息

SENTINEL master redis-master

# 查看另外2个哨兵节点

SENTINEL sentinels redis-master

# 查看从节点列表

SENTINEL slaves redis-master

核心异常问题

SENTINEL slaves redis-master 返回 empty array（空数组）

哨兵没有发现任何从库，也就是：

redis-1、redis-2 并未配置为 redis-0 的从节点，主从复制没搭建起来

1. 给两台从库配置主从复制
分别进入 redis-1、redis-2 执行：

# 进入redis-1
kubectl exec -it redis-1 -n redis-cluster -- redis-cli

AUTH Redis@123456
# 设置为从节点，填写主节点IP和端口
SLAVEOF 10.244.1.173 6379

# 进入redis-2
kubectl exec -it redis-2 -n redis-cluster -- redis-cli

AUTH Redis@123456
SLAVEOF 10.244.1.173 6379

出现问题

哨兵 pod 执行日志查看：

kubectl logs redis-sentinel-86ddbf8ffc-d2pj2 -n redis-cluster

kubectl describe pod redis-sentinel-86ddbf8ffc-d2pj2 -n redis-cluster

解决问题

方案 2：直接使用 redis-0 Pod 固定 IP（彻底规避 DNS 解析问题）

####先获取 redis-0 的集群 IP

kubectl get pod redis-0 -n redis-cluster -o jsonpath='{.status.podIP}'

redis-config.yaml把 sentinel 配置改成 IP 形式

sentinel monitor redis-master 这里粘贴查到的IP 6379 2

####更新配置文件

kubectl apply -f redis-config.yaml

####删除异常哨兵

kubectl delete -f redis-sentinel.yaml

####重新拉起哨兵

kubectl apply -f redis-sentinel.yaml

redis-redis-sentinel.yaml内容:

apiVersion: apps/v1
kind: Deployment
metadata:
name: redis-sentinel
namespace: redis-cluster
spec:
replicas: 3
selector:
matchLabels:
app: redis-sentinel
template:
metadata:
labels:
app: redis-sentinel
spec:
containers:
- name: redis-sentinel
image: swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/redis:7-alpine
ports:
- containerPort: 26379
volumeMounts:
# 只读初始配置
- name: config-init
mountPath: /etc/redis-init
# 可读写运行配置目录
- name: conf-write
mountPath: /etc/redis
# 数据目录
- name: sentinel-data
mountPath: /data
# 启动脚本：复制配置 + 启动哨兵
command:
- sh
- -c
- |
cp /etc/redis-init/sentinel.conf /etc/redis/
exec redis-sentinel /etc/redis/sentinel.conf
volumes:
# 原始配置来自configmap（只读）
- name: config-init
configMap:
name: redis-conf
defaultMode: 0644
# 可读写临时目录
- name: conf-write
emptyDir: {}
- name: sentinel-data
emptyDir: {}