使用Keepalived实现MySQL双主高可用

MySQL双主配置

环境准备：

　　OS： CentOS7
master：192.168.1.10
backup：192.168.1.20
   VIP：192.168.1.30

一、安装MySQL数据库.

在master 和 backup 上安装mysql,安装完后自动启动，mysql root密码为123456

二、修改MySQL配置文件：

1.master端配置文件如下：

.master端配置文件如下：

# vim /etc/my.cnf                                               #添加
server_id =                                                    #backup上设置为2
log-bin = /data/mysql/mysql-bin
log-bin-index=/data/mysql/my-bin.index
binlog-ignore-db = mysql,information_schema   　　　　　　　　　　 #忽略写入binlog日志的库
auto-increment-increment =                         　　　　　　　#字段变化增量值
auto-increment-offset =                                　　　　 #初始字段ID为1
slave-skip-errors = all                                 　　　　 #忽略所有复制产生的错误

# systemctl restart mysqld

2. backup端配置文件如下：

master端和backup端配置只有server_id不一样，别的都一致.

三、创建数据同步用户并查看log bin日志和pos位置：

1.> master上创建 mysql 同步账号并查看log bin日志和pos位置：

# mysql -uroot -p123456

mysql> GRANT  REPLICATION SLAVE ON *.* TO 'repl'@'192.168.1.%' IDENTIFIED  BY 'repl';

mysql> flush  privileges;

mysql> show master status;
+------------------+----------+--------------+--------------------------+-------------------+
| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB         | Executed_Gtid_Set |
+------------------+----------+--------------+--------------------------+-------------------+
| mysql-bin. |       |              | mysql,information_schema |                   |
+------------------+----------+--------------+--------------------------+-------------------+

master配置如下：

# mysql -uroot -p123456

mysql> change master to
    -> master_host='192.168.1.20',                #这里填backup的IP
    -> master_user='repl',
    -> master_password='repl',
    -> master_log_file='mysql-bin.000001',
    -> master_log_pos=;

mysql> start slave;

2.> backup上创建mysql同步账号配置如下：

# mysql -uroot -p123456

mysql> GRANT  REPLICATION SLAVE ON *.* TO 'repl'@'192.168.1.%' IDENTIFIED  BY 'repl';

mysql> flush  privileges;

mysql> change master to
    -> master_host='192.168.1.10',                #这里填master的IP
    -> master_user='repl',
    -> master_password='repl',
    -> master_log_file='mysql-bin.000001',
    -> master_log_pos=;

mysql> start slave;
--------------------- 

分别查看同步状态：

master查看：

mysql> show slave status\G;
*************************** . row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.1.20
                  Master_User: replication
                  Master_Port:
                Connect_Retry:
              Master_Log_File: mysql-bin.
          Read_Master_Log_Pos:
               Relay_Log_File: test2-relay-bin.
                Relay_Log_Pos:
        Relay_Master_Log_File: mysql-bin.
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
--------------------- 

backup查看：

mysql> show slave status\G;
*************************** . row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 192.168.1.10
                  Master_User: replication
                  Master_Port:
                Connect_Retry:
              Master_Log_File: mysql-bin.
          Read_Master_Log_Pos:
               Relay_Log_File: test3-relay-bin.
                Relay_Log_Pos:
        Relay_Master_Log_File: mysql-bin.
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
--------------------- 

Slave_IO和Slave_SQL是YES说明主主同步成功。

四、MySQL主主同步测试

master上插入数据测试：

mysql> create database testdb;

mysql> use testdb;

mysql> create table user (number INT(),name VARCHAR());

mysql> insert into user values(,'testid');

mysql> show tables;
+----------------+
| Tables_in_test |
+----------------+
| user           |
+----------------+
--------------------- 

backup上查看：

mysql> show databases;
+--------------------+
| Database           |
+--------------------+
| information_schema |
| mysql              |
| performance_schema |
| sys                |
| testdb               |
+--------------------+

mysql> use testdb;

mysql> show tables;
+----------------+
| Tables_in_test |
+----------------+
| user           |
+----------------+

mysql> select number,name from user;
+--------+------+
| number | name |
+--------+------+
|       | testid  |
+--------+------+
---------------------

可以看到已经成功同步过去，同样在backup插入到user表数据，一样同步过去，双主配置没有问题。

五、配置keepalived实现双机热备

1.master安装keepalived并配置：

# yum install -y keepalived

# vim /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {
   notification_email {
     admin@test.com
   }
   notification_email_from admin@test.com
   smtp_server 127.0.0.1
   smtp_connect_timeout
   router_id MYSQL_HA
}

vrrp_instance VI_1 {
    state BACKUP
    interface eth0             #根据实际网络接口进行更改
    virtual_router_id
    priority                #优先级，master设置为100
    advert_int
    nopreempt                  #不主动抢占资源，只在master上设置
    authentication {
        auth_type PASS
        auth_pass
    }
    virtual_ipaddress {
        192.168.1.30
    }
}

virtual_server 192.168.1.30  {
    delay_loop
    #lb_algo rr
    #lb_kind NAT
    persistence_timeout
    protocol TCP

    real_server 192.168.1.10  {               #检测本地mysql
        weight
        notify_down /tmp/mysql.sh               #当mysql服务down时，执行此脚本，杀死keepalived实现切换
        TCP_CHECK {
            connect_timeout
            nb_get_retry
            delay_before_retry
        }
    }
}

backup安装keepalived并配置：

# yum install -y keepalived

# vim /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {
   notification_email {
     admin@test.com
   }
   notification_email_from admin@test.com
   smtp_server 127.0.0.1
   smtp_connect_timeout
   router_id MYSQL_HA
}

vrrp_instance VI_1 {
    state BACKUP
    interface eth0             #根据实际网络接口进行更改
    virtual_router_id
    priority                 #优先级，backup设置为90
    advert_int
    #nopreempt                 #主动抢占资源
    authentication {
        auth_type PASS
        auth_pass
    }
    virtual_ipaddress {
        192.168.1.30
    }
}   

virtual_server 192.168.1.30  {
    delay_loop
    #lb_algo rr
    #lb_kind NAT
    persistence_timeout
    protocol TCP

    real_server 192.168.1.20  {               #检测本地mysql
        weight
        notify_down /tmp/mysql.sh                 #当mysql服务down时，执行此脚本，杀死keepalived实现切换
        TCP_CHECK {
            connect_timeout
            nb_get_retry
            delay_before_retry
        }
    }
}

master 和 backup上编辑mysql.sh

# vim /tmp/mysql.sh

#!/bin/bash
pkill keepalived

# chmod +x !$
# systemctl start keepalived

两台mysql服务器授权允许root远程登录：

# mysql -uroot -p123456789

mysql> grant all on *.* to 'root'@'192.168.1.%' identified by '';

mysql> flush privileges;

测试高可用
通过mysql客户端通过VIP连接，看是否连接成功。
这里我用同网段的另一台机器，连接测试：

# mysql -h192.168.1. -uroot -p123456

Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.

mysql> select * from test.user;
+--------+------+
| number | name |
+--------+------+
|       | testid  |
+--------+------+
 row in set (0.01 sec)
--------------------- 

可以看到，连接成功，且查询数据没有问题，停止master上mysql服务，是否能正常切换到backup上，可以使用 ip addr命令来查看VIP在哪台服务器上。

master上查看是否有VIP,可以看到VIP在master上

# ip addr
: lo: <LOOPBACK,UP,LOWER_UP> mtu  qdisc noqueue state UNKNOWN group default qlen
    link/loopback ::::: brd :::::
    inet 127.0.0.1/ scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::/ scope host
       valid_lft forever preferred_lft forever
: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu  qdisc pfifo_fast state UP group default qlen
    link/ether :0c::cf:ab:c4 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.10/ brd 192.168.1.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet 192.168.1.30/ scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::fe8e:3c2f:4d32:e9fd/ scope link noprefixroute
       valid_lft forever preferred_lft forever
--------------------- 

停掉master上mysql服务：

# systemctl stop mysqld

# ps axu |grep keepalived
root        0.0  0.0     pts/    S+   :   : grep --color=autokeepalived

# ip addr
: lo: <LOOPBACK,UP,LOWER_UP> mtu  qdisc noqueue state UNKNOWN group default qlen
    link/loopback ::::: brd :::::
    inet 127.0.0.1/ scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::/ scope host
       valid_lft forever preferred_lft forever
: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu  qdisc pfifo_fast state UP group default qlen
    link/ether :0c::cf:ab:c4 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.10/ brd 192.168.1.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::fe8e:3c2f:4d32:e9fd/ scope link noprefixroute
       valid_lft forever preferred_lft forever
--------------------- 

可以看到，keepalived在mysql服务停掉之后也被停掉，VIP不在master上。

backup上查看是否有VIP，可以看到VIP在backup上。

# ip addr
: lo: <LOOPBACK,UP,LOWER_UP> mtu  qdisc noqueue state UNKNOWN group default qlen
    link/loopback ::::: brd :::::
    inet 127.0.0.1/ scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::/ scope host
       valid_lft forever preferred_lft forever
: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu  qdisc pfifo_fast state UP group default qlen
    link/ether :0c::::d5 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.20/ brd 192.168.1.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet 192.168.1.30/ scope global eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::4b20:2e16:a957:f9a1/ scope link noprefixroute
       valid_lft forever preferred_lft forever
--------------------- 

查看/var/log/messages日志，可以看到主备切换过程：

Apr   :: hosts systemd: Stopping MySQL Server...
Apr   :: hosts Keepalived_healthcheckers[]: TCP connection to [192.168.1.10]: failed.
Apr   :: hosts Keepalived_healthcheckers[]: TCP connection to [192.168.1.10]: failed.
Apr   :: hosts Keepalived_healthcheckers[]: Check on service [192.168.1.10]: failed after  retry.
Apr   :: hosts Keepalived_healthcheckers[]: Removing service [192.168.1.10]: from VS [192.168.1.30]:
Apr   :: hosts Keepalived_healthcheckers[]: IPVS (cmd , errno ): No such destination
Apr   :: hosts Keepalived_healthcheckers[]: Executing [/tmp/mysql.sh] for service [192.168.1.10]: in VS [192.168.1.30]:
Apr   :: hosts Keepalived_healthcheckers[]: Lost quorum -= >  for VS [192.168.1.30]:
Apr   :: hosts Keepalived_healthcheckers[]: Remote SMTP server [127.0.0.1]: connected.
Apr   :: hosts Keepalived_vrrp[]: VRRP_Instance(VI_1) sent  priority
Apr   :: hosts Keepalived_vrrp[]: VRRP_Instance(VI_1) removing protocol VIPs.
Apr   :: hosts Keepalived[]: Stopping
Apr   :: hosts Keepalived_healthcheckers[]: IPVS (cmd , errno ): No such file or directory
Apr   :: hosts Keepalived_healthcheckers[]: Stopped
Apr   :: hosts Keepalived_vrrp[]: Stopped
Apr   :: hosts Keepalived[]: Stopped Keepalived v1.3.5 (/,), git commit v1.3.5--g6fa32f2
Apr   :: hosts systemd: Stopped MySQL Server.
--------------------- 

恢复master服务器故障，看是否主动抢占资源，成为活动服务器。

master上启动mysql服务和keepalived服务：

# systemctl start mysqld

# systemctl start keepalived

# ip addr
: lo: <LOOPBACK,UP,LOWER_UP> mtu  qdisc noqueue state UNKNOWN group default qlen
    link/loopback ::::: brd :::::
    inet 127.0.0.1/ scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::/ scope host
       valid_lft forever preferred_lft forever
: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu  qdisc pfifo_fast state UP group default qlen
    link/ether :0c::cf:ab:c4 brd ff:ff:ff:ff:ff:ff
    inet 192.168.1.10/ brd 192.168.1.255 scope global noprefixroute eth0
       valid_lft forever preferred_lft forever
    inet6 fe80::fe8e:3c2f:4d32:e9fd/ scope link noprefixroute
       valid_lft forever preferred_lft forever
--------------------- 

可以看到，即使master故障恢复，也没有抢占资源，VIP仍然在backup上，这是因为之前已经配置了master为非抢占模式（nopreempt）。

不过需要注意的是：

nopreempt这个参数只能用于state为BACKUP的情况，所以在配置的时候要把master和backup的state都设置成BACKUP，这样才会实现keepalived的非抢占模式！

也就是说：

* 当state状态一个为MASTER，一个为BACKUP的时候，加不加nopreempt这个参数都是一样的效果。即都是根据priority优先级来决定谁抢占vip资源的，是抢占模式！

* 当state状态都设置成BACKUP，如果不配置nopreempt参数，那么也是看priority优先级决定谁抢占vip资源，即也是抢占模式。

* 当state状态都设置成BACKUP，如果配置nopreempt参数，那么就不会去考虑priority优先级了，是非抢占模式！即只有vip当前所在机器发生故障，另一台机器才能接管vip。
即使优先级高的那一台机器恢复正常后也不会主动抢回vip，只能等到对方发生故障，才会将vip切回来。

关闭主从复制：

登录到从库服务器进行配置
关闭复制

mysql> STOP SLAVE;

重置，清除复制信息，这样再启动时就不会进行复制了。

mysql> RESET SLAVE ALL;

参考文档：https://blog.csdn.net/miss1181248983/article/details/89139951