redis一主一从两台机器高可用

r囧r小猫 2022-05-21 02:20 191阅读 0赞

**目录**

1 机器规划

2 redis的部署安装

2.1配置文件内容

2.2 启动2台机器上的redis服务

2.3 查看2台机器上redis数据信息

3 redis主从之间手动切换

3.1 停掉22上的redis进程，到23从redis上执行上面讲到的SLAVEOF NO ONE参数

3.2 恢复22上的redis数据库

4.keepalived实现主从自动切换

4.1 keepalived执行原理：

4.2 下面我们贴出脚本：

4.3 主从切换

--------------------

## 1 机器规划 ##

## 2 redis的部署安装 ##

### 2.1配置文件内容 ###

这里我们直接贴出简略的配置项，主要测试高可用功能，22机器上的配置内容：

[root@bgsbtsp0006-dqf:scripts]# cat /opt/redis/conf/redis6379.conf 
    bind 172.18.1.22
      
    port 6379
    appendonly yes
    timeout 3600
    daemonize yes
      
    pidfile /opt/redis/pid/redis6379.pid
    logfile "/opt/redis/log/redis6379.log"
    dir "/opt/redis/data/6379"

23上机器配置内容，注意这里要加上slaveof参数配置：

bind 172.18.1.23
      
    port 6380
    appendonly yes 
    timeout 3600
    daemonize yes 
      
    pidfile /opt/redis/pid/redis6380.pid
    logfile "/opt/redis/log/redis6380.log"
    dir "/opt/redis/data/6380"
    slaveof 172.18.1.22 6379

**[SLAVEOF][] 命令用于在 Redis 运行时动态地修改复制(replication)功能的行为。**

**通过执行 `SLAVEOF host port` 命令，可以将当前服务器转变为指定服务器的从属服务器(slave server)。**

**如果当前服务器已经是某个主服务器(master server)的从属服务器，那么执行 `SLAVEOF host port` 将使当前服务器停止对旧主服务器的同步，丢弃旧数据集，转而开始对新主服务器进行同步。**

**另外，对一个从属服务器执行命令 `SLAVEOF NO ONE` 将使得这个从属服务器关闭复制功能，并从从属服务器转变回主服务器，原来同步所得的数据集*不会*被丢弃。**

**利用『 `SLAVEOF NO ONE` 不会丢弃同步所得数据集』这个特性，可以在主服务器失败的时候，将从属服务器用作新的主服务器，从而实现无间断运行。**

### 2.2 启动2台机器上的redis服务 ###

#启动22上redis
    [root@bgsbtsp0006-dqf:opt]# redis-server /opt/redis/conf/redis6379.conf 
    #启动23上redis
    [root@bgsbtsp0007-dqf:opt]# redis-server /opt/redis/conf/redis6380.conf

### 2.3 查看2台机器上redis数据信息 ###

查看22上redis的信息，之前搭建过，所有有一些值。

[root@bgsbtsp0006-dqf:opt]# redis-cli -h 172.18.1.22
    172.18.1.22:6379> get bfd
    "123"
    172.18.1.22:6379> keys *
    1) "wyl"
    2) "zyx"
    3) "bfd"
    4) "lhf"

查看23上redis信息，我们可以看到同步到22上redis数据信息了，但是自己本身是不能够set写操作的

[root@bgsbtsp0007-dqf:opt]# redis-cli -h 172.18.1.23 -p 6380
    172.18.1.23:6380> KEYS *
    1) "lhf"
    2) "bfd"
    3) "zyx"
    4) "wyl"
    172.18.1.23:6380> set mm  '123'
    #同步到22主上的数据了，但是自己本身是无法写入操作的
    (error) READONLY You can't write against a read only slave.
    172.18.1.23:6380>

## 3 redis主从之间手动切换 ##

### 3.1 停掉22上的redis进程，到23从redis上执行上面讲到的**`SLAVEOF NO ONE``参数`** ###

[root@bgsbtsp0006-dqf:opt]# ps -ef |grep redis
    root     2562482       1  0 20:32 ?        00:00:00 redis-server 172.18.1.22:6379
    root     2562701 2537889  0 20:32 pts/2    00:00:00 grep --color=auto redis
    
    [root@bgsbtsp0006-dqf:opt]# kill 2562482

到23从redis上操作，看到ok字样，表示这时候已经不是slave角色了，我们可以进行写操作.

[root@bgsbtsp0006-dqf:opt]# redis-cli -h 172.18.1.23 -p 6380
    172.18.1.23:6380> SLAVEOF no one
    OK
    172.18.1.23:6380> set wyl '123456'
    OK

此时主从切换已经完成。

### 3.2 恢复22上的redis数据库 ###

此时切换已经完成，但是我们还需要启动之前的主redis，但是这个时候我们已经进行过插入、删除等操作，以前的主redis并没有记录这些，那么就需要将现在的23上redis的dump.rdb替换之前22上redis 的dump.rdb文件。

启动22上的redis，并且指定新的主redis,即23上的redis

[root@bgsbtsp0006-dqf:scripts]# redis-cli -h 172.18.1.22 -p 6379
    172.18.1.22:6379>  slaveof  172.18.1.23 6380
    OK
    172.18.1.22:6379> get wyl
    "123456"
    172.18.1.22:6379> set ymh1 '123'
    (error) READONLY You can't write against a read only slave.
    172.18.1.22:6379>

我们可以看到，22上也查询到了刚刚在23机器上创建的wyl=123456的数据。并且22也变成read only 从角色了。

## 4.keepalived实现主从自动切换 ##

keepalived的基础这里不再描述，之前的文章有些，由于上面我们讲到手动切换，如果这块理解了，keepalived就是将上面的手动过程通过脚本的方式去实现，废话不多说，我们看看keepalived脚本是如何去操作的。

### 4.1 keepalived执行原理： ###

22是redis的主库，我们把keepalived的master节点也放在22上，通过check\_redis\_status.sh判断redis进程是否存在，如果不在了，执行falut脚本，打印日志告诉我们主库已经宕机，并且vip会漂移到23机器上

23机器根据keepalived的状态判断，发现vip到自己这边了，并且状态变成master状态，这时候调用master脚本，我们就在这个脚本里执行了上面手动过程中的slaveof no one这个指令。

### 4.2 下面我们贴出脚本： ###

除了检查redis脚本端口不一样，其余3个脚本都是一样的，同步到23机器上。

#检查redis脚本
    [root@bgsbtsp0006-dqf:scripts]# cat check_redis.sh 
    #!/bin/bash
    
    num=`ps -ef|grep 6379|grep -v grep|wc -l`
    
    if [ $num -ge 1 ];then
       status=0
    else
       status=1
    fi
    exit $status
    
    
    #master脚本
    [root@bgsbtsp0006-dqf:scripts]# cat master.sh 
    #!/bin/bash
    LOGFILE=/etc/keepalived/scripts/keepalived.log
    echo "[Master]" >> $LOGFILE
    date >> $LOGFILE
    
    
    hostname=`hostname`
    if [ $hostname = 'bgsbtsp0006-dqf' ];then
       echo 22 >>$LOGFILE
       echo 'SLAVEOF no one'|/opt/redis/bin/redis-cli -h 172.18.1.22 -p 6379 >> $LOGFILE
       echo -e `date +"%F %T"` "the current redis is master" >> $LOGFILE
    elif [ $'bgsbtsp0007-dqf = 'bfd-yiz-1p23' ];then
      echo 23 >>$LOGFILE
      echo 'SLAVEOF no one'|/opt/redis/bin/redis-cli -h 172.18.1.23 -p 6380 >>$LOGFILE
      echo -e `date +"%F %T"` "the current redis is master" >> $LOGFILE
    fi
    
    
    #backup脚本
    [root@bgsbtsp0006-dqf:scripts]# cat backup.sh 
    #!/bin/bash
    LOGFILE=/etc/keepalived/scripts/keepalived.log
    echo "[Backup]" >> $LOGFILE
    date >> $LOGFILE
    
    echo -e `date +"%F %T"` "the current redis is slave" >> $LOGFILE
    
    # fault脚本
    [root@bgsbtsp0006-dqf:scripts]# cat fault.sh 
    #!/bin/bash
    LOGFILE=/etc/keepalived/scripts/keepalived.log
    echo "[Fault]" >> $LOGFILE
    date >> $LOGFILE
    
    
    echo -e `date +"%F %T"` "the current redis is down" >> $LOGFILE

### 4.3 主从切换 ###

这时候，我们启动22，23上keepalived服务，22查看keepalived的状态，vip绑定在22机器上。

[root@bgsbtsp0006-dqf:scripts]# systemctl status keepalived.service 
    ● keepalived.service - LVS and VRRP High Availability Monitor
       Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
       Active: active (running) since Tue 2019-01-08 20:17:12 CST; 15min ago
      Process: 2558442 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
     Main PID: 2558443 (keepalived)
       CGroup: /system.slice/keepalived.service
               ├─2558443 /usr/sbin/keepalived -D
               ├─2558444 /usr/sbin/keepalived -D
               └─2558445 /usr/sbin/keepalived -D
    
    Jan 08 20:32:23 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:23 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:23 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:23 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Opening script file /etc/keepalived/scripts/ma....sh
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: VRRP_Instance(VI_1) Sending/queueing gratuitou....72
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72
    Jan 08 20:32:28 bgsbtsp0006-dqf Keepalived_vrrp[2558445]: Sending gratuitous ARP on bond0 for 172.18.1.72

23查看keepalived的状态

[root@bgsbtsp0007-dqf:scripts]# systemctl status keepalived.service 
    ● keepalived.service - LVS and VRRP High Availability Monitor
       Loaded: loaded (/usr/lib/systemd/system/keepalived.service; disabled; vendor preset: disabled)
       Active: active (running) since Tue 2019-01-08 20:17:57 CST; 12min ago
      Process: 3863461 ExecStart=/usr/sbin/keepalived $KEEPALIVED_OPTIONS (code=exited, status=0/SUCCESS)
     Main PID: 3863462 (keepalived)
       CGroup: /system.slice/keepalived.service
               ├─3863462 /usr/sbin/keepalived -D
               ├─3863463 /usr/sbin/keepalived -D
               └─3863464 /usr/sbin/keepalived -D
    
    Jan 08 20:22:35 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: VRRP_Instance(VI_1) Now in FAULT state
    Jan 08 20:22:35 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:37 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:39 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:41 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:43 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:45 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: /etc/keepalived/scripts/check_redis.sh exited wit...s 1
    Jan 08 20:22:47 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: VRRP_Script(check_nginx_alived) succeeded
    Jan 08 20:22:47 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: VRRP_Instance(VI_1) Entering BACKUP STATE
    Jan 08 20:22:47 bgsbtsp0007-dqf Keepalived_vrrp[3863464]: Opening script file /etc/keepalived/scripts/backup.s

停掉22机器上主redis服务

查看22上脚本执行的过程,显示调用了master脚本，当停掉redis服务后，执行了fault脚本

[root@bgsbtsp0006-dqf:scripts]# cat keepalived.log 
    [Master]
    Tue Jan  8 20:30:09 CST 2019
    
    [Fault]
    Tue Jan  8 20:33:07 CST 2019
    2019-01-08 20:33:07 the current redis is down

查看23上脚本执行的过程,显示调用了backup脚本，当停掉22上redis服务后，23上执行了master脚本

[root@bfd-yiz-1p23 /etc/keepalived/scripts]# cat keepalived.log 
    [Backup]
    Tue Jan  8 20:31:09 CST 2019
    2019-01-08 20:33:09 the current database is master
    
    [Master]
    Tue Jan  8 20:33:09 CST 2019
    2019-01-08 20:33:09 the current database is master

我们发现已经自动切换了，其实我们也可以用客户端去连接一下，去验证，这比较准确，我已经验证过了，这里就不再过多赘述。

同理，当我们恢复22上主机后，要手动执行slaveof  172.18.1.23 6380，这样数据就同步到23了，后面你们也可以试一下当23上redis挂了，22是否会变成主。

[SLAVEOF]: http://redisdoc.com/server/slaveof.html#slaveof