Docker swarm 配置 redis 高可用 哨兵模式

sloppy
882 阅读4分钟

Docker swarm 配置 redis 高可用 哨兵模式

一，环境准备条件

docker版本 ：Docker version 20.10.1, build 831ebea

mongodb版本：4.2.7

服务器版本：CentOS Linux release 7.7.1908 (Core)

1，准备三台服务器互相可以访问

例如：

10.11.32.25

10.11.32.26

10.11.32.27

2，三台服务器都要安装docker 环境

配置 daemon.conf

三台服务器都要创建 docker.service 文件 ,具体内容见附录

vi /lib/systemd/system/docker.service

重启docker

systemctl daemon-reload
systemctl restart docker

3，redis 集群架构

image-20210906153659268.png

二，创建swarm 集群及集群网络

1，在一台服务器上创建swarm 集群

docker swarm init --advertise-addr  10.11.32.25（本机ip）

在其它两台服务器执行加入swarm集群

docker swarm join --token SWMTKN-1-42w2dmkv2u18k9d6vi65rhqggjcqcc31iijnecd4bbi43bmom2-b3o89g28thkcken0v3qxtczuj 172.16.10.85:2377

忘记token 执行以下命令

docker swarm join-token manager  查看加入manager的命令

查看集群关系

docker node ls  

管理节点：赋值其他工作节点为管理节点

docker node promote 工作节点主机名1
docker node promote 工作节点主机名2

2，manager节点创建集群网络**

docker network create -d overlay --attachable GIE-IOT-mongodbs

--attachable 允许其他容器加入此网络

查看集群网络

docker network ls

image-20210904224257693.png

三，配置mongodb高可用步骤

1，初始化目录创建yml文件

所有服务器创建相创建相关目录

mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/redis1/{conf,data,logs}
mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/redis2/{conf,data,logs}
mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/redis3/{conf,data,logs}
mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/sentinel1/{conf,data,logs}
mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/sentinel2/{conf,data,logs}
mkdir -pv /data/docker/composes/prod/middleware/redis_cluster/sentinel3/{conf,data,logs}

leader服务器创建 docker-compose-redis-cluster.yml文件

version: '3.8'
services:
  redis1:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 6379
        published: 6379
        protocol: tcp
        mode: host
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/redis1/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/redis1/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node03
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/redis.conf ];then
      echo 'port 6379' > /conf/redis.conf ;
      echo 'bind 0.0.0.0' >> /conf/redis.conf ;
      echo 'slave-announce-ip 10.11.32.25' >> /conf/redis.conf ;
      echo 'slave-announce-port 6379' >> /conf/redis.conf ;
      echo 'masterauth 123456' >> /conf/redis.conf;
      echo 'requirepass 123456' >> /conf/redis.conf ;
      echo 'appendonly yes' >> /conf/redis.conf ; fi &&
      redis-server /conf/redis.conf"
      # 行1：如果redis.conf不存在
      # 行2-8：新建redis.conf文件，并写入配置信息
      # 行9：启动redis实例
  redis2:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 6379
        published: 6379
        protocol: tcp
        mode: host
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/redis2/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/redis2/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    depends_on:
      - redis1
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node04
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/redis.conf ];then
      echo 'port 6379' > /conf/redis.conf ;
      echo 'bind 0.0.0.0' >> /conf/redis.conf ;
      echo 'slave-announce-ip 10.11.32.26' >> /conf/redis.conf ;
      echo 'slave-announce-port 6379' >> /conf/redis.conf ;
      echo 'masterauth 123456' >> /conf/redis.conf;
      echo 'requirepass 123456' >> /conf/redis.conf ;
      echo 'replicaof 10.11.32.25 6379' >> /conf/redis.conf ;
      echo 'appendonly yes' >> /conf/redis.conf ; fi &&
      redis-server /conf/redis.conf"
  redis3:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 6379
        published: 6379
        protocol: tcp
        mode: host
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/redis3/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/redis3/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    depends_on:
      - redis1
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node05
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/redis.conf ];then
      echo 'port 6379' > /conf/redis.conf ;
      echo 'bind 0.0.0.0' >> /conf/redis.conf ;
      echo 'slave-announce-ip 10.11.32.27' >> /conf/redis.conf ;
      echo 'slave-announce-port 6379' >> /conf/redis.conf ;
      echo 'masterauth 123456' >> /conf/redis.conf;
      echo 'requirepass 123456' >> /conf/redis.conf ;
      echo 'replicaof 10.11.32.25 6379' >> /conf/redis.conf ;
      echo 'appendonly yes' >> /conf/redis.conf ; fi &&
      redis-server /conf/redis.conf"
  sentinel1:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 26379
        published: 26379
        protocol: tcp
        mode: host
    depends_on:
      - redis1
      - redis2
      - redis3
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel1/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel1/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node03
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/sentinel.conf ];then
      echo 'port 26379' > /conf/sentinel.conf ;
      echo 'sentinel announce-ip 10.11.32.25' >> /conf/sentinel.conf ;
      echo 'sentinel monitor mymaster 10.11.32.25 6379 2' >> /conf/sentinel.conf ;
      echo 'sentinel auth-pass mymaster 123456' >> /conf/sentinel.conf ; fi &&
      redis-server /conf/sentinel.conf --sentinel"
      # 行1：如果sentinel.conf不存在
      # 行2-5：新建sentinel.conf文件，并写入配置信息
      # 行6：启动sentinel实例
  sentinel2:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 26379
        published: 26379
        protocol: tcp
        mode: host
    depends_on:
      - redis1
      - redis2
      - redis3
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel2/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel2/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node04
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/sentinel.conf ];then
      echo 'port 26379' > /conf/sentinel.conf ;
      echo 'sentinel announce-ip 10.11.32.26' >> /conf/sentinel.conf ;
      echo 'sentinel monitor mymaster 10.11.32.25 6379 2' >> /conf/sentinel.conf ;
      echo 'sentinel auth-pass mymaster 123456' >> /conf/sentinel.conf ; fi &&
      redis-server /conf/sentinel.conf --sentinel"
  sentinel3:
    image: 10.11.32.23:5000/redis:v5.0.10
    environment:
      TZ: "Asia/Shanghai"
    ports:
      - target: 26379
        published: 26379
        protocol: tcp
        mode: host
    depends_on:
      - redis1
      - redis2
      - redis3
    volumes:
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel3/data:/data
      - /data/docker/composes/prod/middleware/redis_cluster/sentinel3/conf:/conf
    networks:
      - GIE-IOT-mongodbs
    deploy:
      replicas: 1
      placement:
        constraints:
          - node.hostname == GIE-EC-Node05
    sysctls:
      net.core.somaxconn: '511'
    command: >
      bash -c "if [ ! -f /conf/sentinel.conf ];then
      echo 'port 26379' > /conf/sentinel.conf ;
      echo 'sentinel announce-ip 10.11.32.27' >> /conf/sentinel.conf ;
      echo 'sentinel monitor mymaster 10.11.32.25 6379 2' >> /conf/sentinel.conf ;
      echo 'sentinel auth-pass mymaster 123456' >> /conf/sentinel.conf ; fi &&
      redis-server /conf/sentinel.conf --sentinel"
networks:
  GIE-IOT-mongodbs:
    external: true

启动服务，在 manager上执行

docker stack deploy -c docker-compose-mongo-cluster.yml mongo

在 manager查看服务的启动情况

docker service ls

image-20210906154120515.png

2， 查看是否配置成功

登录任意一台服务器通过客户端访问redis

例如： 10.11.32.26

docker exec -it redis_cluster_redis2  redis-cli

image-20210906154621152.png

执行 输入密码

127.0.0.1:6379> auth 123456

执行 info 可以查看主从信息

127.0.0.1:6379> info

image-20210906154833176.png

其它服务器用同样的方法查看发现是 slave

image-20210906155143397.png

3， 测试数据是否同步，主节点挂掉从节点是否顶替主机节点

删除redis主节点

docker service rm  [主节点serviceId]

访问从节点发现有一个从节点已经成为 master

127.0.0.1:6379> info  查看

已经测试过，这里就赘述了。

附录

参考文档

blog.csdn.net/liushengit/…

blog.csdn.net/weixin_4217…

www.cnblogs.com/aspirant/p/… yml文件说明

avatar
文章
阅读
粉丝