MYSQL高可用方案主要分为两大类，一类是前置管理，一类是后置管理。

前置管理的思想是利用各种前置管理工具，动态切换或者分发请求到不同的实例并切换master，如keepalived、MHA、packmaker+Corosync、MySQLProxy等，一般通过暴露VIP将整个MYQL集群隐藏起来，其中Proxy主要用以读写分离。后置管理则是利用共享存储的思想，如NDB方案。由于我们的用户尚不足百万，所以一开始的计划是选择最简单的前置管理模式-双主+keepalived方案，后来因为阿里云停用了VIP服务，于是购买了其提供的负载均衡SLB服务(替代keepalived使用)，最终完成了双主+VIP的搭建。不管怎样，keepalived的思想还是比较经典的，所以也做了简单记录。

Part1 mysql bin 方式安装和配置

#查询已经存在的mysql并删除

rpm -qa |grep mysql

rpm -qa |grep mariadb

#删除已经存在的mysql

rpm -e --nodeps mariadbxxxx

rpm -e --nodeps mysqlxxxx

#下载解压并copy到/usr/local/mysql

#创建mysql数据日志目录

mkdir -p /data/mysql/data

mkdir -p /data/mysql/log

mkdir -p /data/mysql/bak

#初始化my.cnf

[mysql]

default-character-set=utf8

[mysqld]

character_set_server=utf8

default-storage-engine=INNODB

datadir=/data/mysql/data

#socket=/data/mysql/data/mysql.sock

port=3306

symbolic-links=0

log-error=/data/mysql/log/mysqld.log

pid-file=/data/mysql/mysqld.pid

###将support-files的mysql.server copy 到etc/init.d/以开机启动

cd /usr/local/mysql/support-files/

cp mysql.server /etc/init.d/mysql

#修改mysql默认配置

vim /etc/init.d/mysql

basedir=/usr/local/mysql

datadir=/data/mysql/data

#创建mysql系统用户并授权到mysql启动

groupadd mysql

useradd -r -g mysql mysql

passwd mysql

mysqlpwd

chown -R mysql:mysql /usr/local/mysql/

chown -R mysql:mysql /data/mysql/

#将mysqlbin目录添加到环境变量

echo 'export PATH=/usr/local/mysql/bin:$PATH' >> /etc/profile

source /etc/profile

#初始化mysql数据库

yum install libaio-devel.x86_64

mysqld --initialize --user=mysql --basedir=/usr/local/mysql --datadir=/data/mysql/data

#获取初始密码

tail -f /data/mysql/log/mysql.log

[Note] A temporary password is generated for root@localhost: im>wqfae&7tY

#给数据库加密

mysql_ssl_rsa_setup --datadir=/data/mysql/data

chown -R mysql:mysql /data/mysql/

##设置开机自动重启

chkconfig --add mysql

chkconfig mysql on

##查看、停止、重启mysql

systemclt status mysql.service

systemctl start mysql.service

systemctl stop mysql.service

systemctl restart mysql.service

#启动mysql服务

systemctl start mysql.service

#使用初始密码登录

mysql -uroot -p --socket=/data/mysql/data/mysql.sock

#设置新密码,授权root用户访问

set password=password('*******');

grant all privileges on *.* to root@'%' identified by '('*******');';

flush privileges;

Part2 mysql高可用配置 - keepalived

1. 配置master1和master2 binlog

#master1配置

[mysqld]

log-bin=mysql-bin

server-id=70

gtid_mode = on

enforce_gtid_consistency = 1

log_slave_updates=1

#binlog-ignore=mysql

#binlog-ignore=information_schema

#replicate-do-db=platform88

#master2配置

[mysqld]

log-bin=mysql-bin

server-id=71

gtid_mode = on

enforce_gtid_consistency = 1

log_slave_updates=1

#binlog-ignore=mysql

#binlog-ignore=information_schema

#replicate-do-db=platform88

#read_only=1

###备注：手动切换主库

mysql> stop slave;

#千万不要执行 reset master，否则会从最先的GTID上开始执行。

mysql> change master to master_host='127.0.0.1',master_user='rep',master_password='rep',master_port=3307,master_auto_position=1; #指定到另一个比较接近主的从上。

mysql> start slave;

2. 创建可互相访问的用户

#master1配置

CREATE USER 'ms_repl_usr'@'172.26.27.71' IDENTIFIED BY 'ms_repl_pwd';

GRANT REPLICATION SLAVE ON *.* TO 'ms_repl_usr'@'172.26.27.71';

FLUSH PRIVILEGES;

#master2配置

CREATE USER 'ms_repl_usr'@'172.26.27.70' IDENTIFIED BY 'ms_repl_pwd';

GRANT REPLICATION SLAVE ON *.* TO 'ms_repl_usr'@'172.26.27.70';

FLUSH PRIVILEGES;

查看binlog及位移

show master status \G

#master1配置

CHANGE MASTER TO

MASTER_HOST='172.26.27.71',

MASTER_PORT=30468,

MASTER_USER='ms_repl_usr',

MASTER_PASSWORD='ms_repl_pwd',

MASTER_LOG_FILE='mysql-bin.000002',

MASTER_LOG_POS=1459;

或者：

change master to master_host='172.26.27.71',master_user='ms_repl_usr',master_password='ms_repl_pwd', master_auto_position=1;

#master2配置

CHANGE MASTER TO

MASTER_HOST='172.26.27.70',

MASTER_PORT=30468,

MASTER_USER='ms_repl_usr',

MASTER_PASSWORD='ms_repl_pwd',

MASTER_LOG_FILE='mysql-bin.000002',

MASTER_LOG_POS=2086;

或者：

change master to master_host='172.26.27.70'',master_user='ms_repl_usr',master_password='ms_repl_pwd', master_auto_position=1;

#分别重启master1和master2

查看 salve status

show slave status \G

Slave_IO_Running: Yes

Slave_SQL_Running: Yes

3. keepalived 配置

keepalived主要用检测对方服务的状态，并通过关闭或者重启自身服务达到切换VIP优先级的效果。keepalived可以监控4层或者7层网络，4层通过配置虚拟服务实现，本例子中使用7层监控。

#master1 配置

vrrp_script chk_mysql {

script "/etc/keepalived/check_mysql.sh"

interval 30 #设置检查间隔时长，可根据自己的需求自行设定

}

vrrp_instance VI_1 {

state BACKUP #通过下面的priority来区分MASTER和BACKUP，也只有如此，底下的nopreempt才有效

interface eth0

virtual_router_id 51

priority 100

advert_int 1

nopreempt #防止切换到从库后，主keepalived恢复后自动切换回主库

authentication {

auth_type PASS

auth_pass PWD

}

track_script {

chk_mysql

}

virtual_ipaddress {

192.168.72.109

}

}

#master2配置

! Configuration File for keepalived

vrrp_instance VI_1 {

state BACKUP

interface eth0

virtual_router_id 51

priority 90

advert_int 1

authentication {

auth_type PASS

auth_pass PWD

}

notify_master /etc/keepalived/notify_master_mysql.sh #此条指令告诉keepalived发现自己转为MASTER后执行的脚本

virtual_ipaddress {

192.168.72.109

}

}

#check_mysql.sh主要用来监听或者链接mysql实例，如果发现mysql无法连接，则直接重启keepalived，或者停止keepalived，让虚拟IP漂移到另一台服务器上。这里以监听端口为例：

#!/bin/bash

#This scripts is check for Mysql Slave status

echo 'testing...'

counter=`netstat -luntp|grep 3306|wc -l`

echo 'after counting...'

if [ "$counter" -eq 0 ];

then

echo "3306 is down..."

systemctl stop keepalived

else

echo '3306 is working..'

fi

check脚本中经常遇到的一个问题是，命令需要用全路径，如mysql登录，

/usr/local/mysql/bin/mysql -uroot -pxxxx --port 3306

修改keepalived的日志输出，默认输出在/var/log/message。从keepalived的日志中可以看到，master切换的过程。

vim /etc/rsyslog.conf

keepalived -S 0

local0.* /var/log/keepalived.log

阿里云SLB配置实现keepalived效果

#通过负载均衡管理界面创建主备服务器组

#添加双主实例

这样就完成了阿里云双主的搭建，从这个结构延伸开来，可以分别为两个主节点添加slave节点，行程4或者6节点结构。如果需要读写分离，提高资源的利用率，可以通过前置mysqlproxy来将读写分开。