基于keepalived搭建MySQL的高可用集群

MySQL的高可用方案一般有如下几种：

keepalived+双主，MHA，MMM，Heartbeat+DRBD，PXC，Galera Cluster

比较常用的是keepalived+双主，MHA和PXC。

对于小公司，一般推荐使用keepalived+双主，简单。

下面来部署一下

配置环境：

角色 主机IP 主机名 操作系统版本 软件版本

VIP 192.168.244.10

master1 192.168.244.145 master1 CentOS7.1 MySQL 5.6.26，Keepalived v1.2.13

master2 192.168.244.146 master2 CentOS7.1 MySQL 5.6.26，Keepalived v1.2.13

一、 配置MySQL双主复制环境

1. 修改配置文件

master1中有关复制的配置如下：

[mysqld]
log-bin=mysql-bin
server-id=1
log_slave_updates=1

master2

[mysqld]
log-bin=mysql-bin
server-id=2
log_slave_updates=1
read_only=1

2. 创建复制用户

master1中创建：

CREATE USER 'repl'@'192.168.244.146' IDENTIFIED BY 'mysql';
GRANT REPLICATION SLAVE ON *.* TO 'repl'@'192.168.244.146';

master2中创建：

CREATE USER 'repl'@'192.168.244.145' IDENTIFIED BY 'mysql';
GRANT REPLICATION SLAVE ON *.* TO 'repl'@'192.168.244.145';

3. 执行CHANGE MASTER TO语句

因是从头搭建MySQL主从复制集群，所以不需要获取全局读锁来得到二进制日志文件的位置，直接根据show master status的输出来确认。

master1上执行：

CHANGE MASTER TO
MASTER_HOST='192.168.244.146',
MASTER_USER='repl',
MASTER_PASSWORD='mysql',
MASTER_LOG_FILE='mysql-bin.000004',
MASTER_LOG_POS=64729;

master2上执行：

CHANGE MASTER TO
MASTER_HOST='192.168.244.145',
MASTER_USER='repl',
MASTER_PASSWORD='mysql',
MASTER_LOG_FILE='mysql-bin.000003',
MASTER_LOG_POS=68479;

4. 分别在两个节点上执行start slave语句并通过show slave status\G查看复制是否搭建成功。

成功标准：

Slave_IO_Running: Yes
Slave_SQL_Running: Yes

二、 配置Keepalived

1. 安装Keepalived

# yum install -y keepalived

当然，也可直接编译官方的源码包。

2. 修改Keepalived的配置文件

master1

[root@master1 ~]# vim /etc/keepalived/keepalived.conf

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 eno16777736
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 1111
}
track_script {
chk_mysql
}

virtual_ipaddress {
192.168.244.10/24
}
}

关于keepalived的参数的详细介绍，可参考：LVS+Keepalived搭建MyCAT高可用负载均衡集群

其中，/etc/keepalived/check_mysql.sh内容如下：

#!/bin/bash

###判断如果上次检查的脚本还没执行完，则退出此次执行
if [ `ps -ef|grep -w "$0"|grep -v "grep"|wc -l` -gt 2 ];then
exit 0
fi
mysql_con='mysql -uroot -p123456'
error_log="/etc/keepalived/logs/check_mysql.err"

###定义一个简单判断mysql是否可用的函数
function excute_query {
${mysql_con} -e "select 1;" 2>> ${error_log}
}

###定义无法执行查询，且mysql服务异常时的处理函数
function service_error {
echo -e "`date "+%F  %H:%M:%S"`    -----mysql service error，now stop keepalived-----" >> ${error_log}
service keepalived stop &>> ${error_log}
echo "DB1 keepalived 已停止"|mail -s "DB1 keepalived 已停止,请及时处理！" slowtech@126.com 2>> ${error_log}
echo -e "\n---------------------------------------------------------\n" >> ${error_log}
}

###定义无法执行查询,但mysql服务正常的处理函数
function query_error {
echo -e "`date "+%F  %H:%M:%S"`    -----query error, but mysql service ok, retry after 30s-----" >> ${error_log}
sleep 30
excute_query
if [ $? -ne 0 ];then
echo -e "`date "+%F  %H:%M:%S"`    -----still can't execute query-----" >> ${error_log}

###对DB1设置read_only属性
echo -e "`date "+%F  %H:%M:%S"`    -----set read_only = 1 on DB1-----" >> ${error_log}
mysql_con -e "set global read_only = 1;" 2>> ${error_log}

###kill掉当前客户端连接
echo -e "`date "+%F  %H:%M:%S"`    -----kill current client thread-----" >> ${error_log}
rm -f /tmp/kill.sql &>/dev/null
###这里其实是一个批量kill线程的小技巧
mysql_con -e 'select concat("kill ",id,";") from  information_schema.PROCESSLIST where command="Query" or command="Execute" into outfile "/tmp/kill.sql";'
mysql_con -e "source /tmp/kill.sql"
sleep 2    ###给kill一个执行和缓冲时间
###关闭本机keepalived
echo -e "`date "+%F  %H:%M:%S"`    -----stop keepalived-----" >> ${error_log}
service keepalived stop &>> ${error_log}
echo "DB1 keepalived 已停止"|mail -s "DB1 keepalived 已停止,请及时处理！" slowtech@126.com 2>> ${error_log}
echo -e "\n---------------------------------------------------------\n" >> ${error_log}
else
echo -e "`date "+%F  %H:%M:%S"`    -----query ok after 30s-----" >> ${error_log}
echo -e "\n---------------------------------------------------------\n" >> ${error_log}
fi
}

###检查开始: 执行查询
excute_query
if [ $? -ne 0 ];then
service mysqld status &>/dev/null
if [ $? -ne 0 ];then
service_error
else
query_error
fi
fi

通过具体的查询语句来判断数据库服务的可用性，如果查询失败，则判断mysqld进程本身的状态，如果不正常，则直接停止当前节点的keepalived，将VIP转移到另外一个节点，如果正常，则等待30s，再次执行查询语句，还是失败，则将当前的master节点设置为read_only，并kill掉当前的客户端连接，然后停止当前的keepalived。

master2

[root@master2 ~]# vim /etc/keepalived/keepalived.conf

! Configuration File for keepalived

vrrp_instance VI_1 {
state BACKUP
interface eno16777736
virtual_router_id 51
priority 90
advert_int 1
authentication {
auth_type PASS
auth_pass 1111
}
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 {
192.168.244.10/24
}
}

其中，/etc/keepalived/notify_master_mysql.sh的内容如下：

#!/bin/bash
###当keepalived监测到本机转为MASTER状态时，执行该脚本

change_log=/etc/keepalived/logs/state_change.log
mysql_con='mysql -uroot -p123456'
echo -e "`date "+%F  %H:%M:%S"`   -----keepalived change to MASTER-----" >> $change_log

slave_info() {
###统一定义一个函数取得slave的position、running、和log_file等信息
###根据函数后面所跟参数来决定取得哪些数据
if [ $1 = slave_status ];then
slave_stat=`${mysql_con} -e "show slave status\G;"|egrep -w "Slave_IO_Running|Slave_SQL_Running"`
Slave_IO_Running=`echo $slave_stat|awk '{print $2}'`
Slave_SQL_Running=`echo $slave_stat|awk '{print $4}'`
elif [ $1 = log_file -a $2 = pos ];then
log_file_pos=`${mysql_con} -e "show slave status\G;"|egrep -w "Master_Log_File|Read_Master_Log_Pos|Relay_Master_Log_File|Exec_Master_Log_Pos"`
Master_Log_File=`echo $log_file_pos|awk '{print $2}'`
Read_Master_Log_Pos=`echo $log_file_pos|awk '{print $4}'`
Relay_Master_Log_File=`echo $log_file_pos|awk '{print $6}'`
Exec_Master_Log_Pos=`echo $log_file_pos|awk '{print $8}'`
fi
}

action() {
###经判断'应该&可以'切换时执行的动作
echo -e "`date "+%F  %H:%M:%S"`    -----set read_only = 0 on DB2-----" >> $change_log

###解除read_only属性
${mysql_con} -e "set global read_only = 0;" 2>> $change_log

echo "DB2 keepalived转为MASTER状态，线上数据库切换至DB2"|mail -s "DB2 keepalived change to MASTER"\
slowtech@126.com 2>> $change_log

echo -e "---------------------------------------------------------\n" >> $change_log
}

slave_info slave_status
if [ $Slave_SQL_Running = Yes ];then
i=0    #一个计数器
slave_info log_file pos
###判断从master接收到的binlog是否全部在本地执行(这样仍无法完全确定从库已追上主库，因为无法完全保证io_thread没有延时(由网络传输问题导致的从库落后的概率很小)
until [ $Master_Log_File = $Relay_Master_Log_File -a $Read_Master_Log_Pos = $Exec_Master_Log_Pos ]
do
if [ $i -lt 10 ];then    #将等待exec_pos追上read_pos的时间限制为10s
echo -e "`date "+%F  %H:%M:%S"`    -----Relay_Master_Log_File=$Relay_Master_Log_File,Exec_Master_Log_Pos=$Exec_Master_Log_Pos is behind Master_Log_File=$Master_Log_File,Read_Master_Log_Pos=$Read_Master_Log_Pos, wait......" >> $change_log    #输出消息到日志，等待exec_pos=read_pos
i=$(($i+1))
sleep 1
slave_info log_file pos
else
echo -e "The waits time is more than 10s,now force change. Master_Log_File=$Master_Log_File Read_Master_Log_Pos=$Read_Master_Log_Pos Relay_Master_Log_File=$Relay_Master_Log_File Exec_Master_Log_Pos=$Exec_Master_Log_Pos" >> $change_log
action
exit 0
fi
done
action

else
slave_info log_file pos
echo -e "DB2's slave status is wrong,now force change. Master_Log_File=$Master_Log_File Read_Master_Log_Pos=$Read_Master_Log_Pos Relay_Master_Log_File=$Relay_Master_Log_File Exec_Master_Log_Pos=$Exec_Master_Log_Pos" >> $change_log
action
fi

整个脚本的逻辑是让从的Exec_Master_Log_Pos尽可能的追上Read_Master_Log_Pos，它给了10s的限制，如果还是没有追上，则直接将master2设置为主（通过解除read_only属性），其实这里面还是有待商榷的，譬如10s的限制是否合理，还是一定需要Exec_Master_Log_Pos=Read_Master_Log_Pos才切换。

当原主恢复正常后，如何将VIP从master2切回到master1中呢？

#!/bin/bash
###手动执行将主库切换回DB1的操作

mysql_con='mysql -uroot -p123456'

echo -e "`date "+%F  %H:%M:%S"`    -----change to BACKUP manually-----" >> /etc/keepalived/logs/state_change.log
echo -e "`date "+%F  %H:%M:%S"`    -----set read_only = 1 on DB2-----" >> /etc/keepalived/logs/state_change.log
$mysql_con -e "set global read_only = 1;" 2>> /etc/keepalived/logs/state_change.log

###kill掉当前客户端连接
echo -e "`date "+%F  %H:%M:%S"`    -----kill current client thread-----" >> /etc/keepalived/logs/state_change.log
rm -f /tmp/kill.sql &>/dev/null
###这里其实是一个批量kill线程的小技巧
$mysql_con -e 'select concat("kill ",id,";") from  information_schema.PROCESSLIST where command="Query" or command="Execute" into outfile "/tmp/kill.sql";'
$mysql_con -e "source /tmp/kill.sql" 2>> /etc/keepalived/logs/state_change.log
sleep 2    ###给kill一个执行和缓冲时间

###确保DB1已经追上了,下面的repl为复制所用的账户，-h后跟DB1的内网IP
log_file_pos=`mysql -urepl -pmysql -h192.168.244.145 -e "show slave status\G;"|egrep -w "Master_Log_File|Read_Master_Log_Pos|Relay_Master_Log_File|Exec_Master_Log_Pos"`
Master_Log_File=`echo $log_file_pos|awk '{print $2}'`
Read_Master_Log_Pos=`echo $log_file_pos|awk '{print $4}'`
Relay_Master_Log_File=`echo $log_file_pos|awk '{print $6}'`
Exec_Master_Log_Pos=`echo $log_file_pos|awk '{print $8}'`
until [ $Read_Master_Log_Pos = $Exec_Master_Log_Pos -a $Master_Log_File = $Relay_Master_Log_File ]
do
echo -e "`date "+%F  %H:%M:%S"`    -----DB1 Exec_Master_Log_Pos($exec_pos) is behind Read_Master_Log_Pos($read_pos), wait......" >> /etc/keepalived/logs/state_change.log
sleep 1
done

###然后解除DB1的read_only属性
echo -e "`date "+%F  %H:%M:%S"`    -----set read_only = 0 on DB1-----" >> /etc/keepalived/logs/state_change.log
ssh 192.168.244.145 'mysql -uroot -p123456 -e "set global read_only = 0;" && /etc/init.d/keepalived start' 2>> /etc/keepalived/logs/state_change.log

###重启DB2的keepalived使VIP漂移到DB1
echo -e "`date "+%F  %H:%M:%S"`    -----make VIP move to DB1-----" >> /etc/keepalived/logs/state_change.log
/sbin/service keepalived restart &>> /etc/keepalived/logs/state_change.log

echo "DB2 keepalived转为BACKUP状态，线上数据库切换至DB1"|mail -s "DB2 keepalived change to BACKUP" slowtech@126.com 2>> /etc/keepalived/logs/state_change.log

echo -e "--------------------------------------------------\n" >> /etc/keepalived/logs/state_change.log

总结：

1. /etc/keepalived/check_mysql.sh和/etc/keepalived/notify_master_mysql.sh必须加可执行权限。

如果前者没有加可执行权限，则master1上将不会绑定VIP，日志直接提示如下信息：

May 25 14:37:09 master1 Keepalived_vrrp[3165]: VRRP_Instance(VI_1) Entering BACKUP STATE
May 25 14:37:09 master1 Keepalived_vrrp[3165]: VRRP sockpool: [ifindex(2), proto(112), unicast(0), fd(10,11)]
May 25 14:37:50 master1 Keepalived_vrrp[3165]: VRRP_Instance(VI_1) Now in FAULT state

2. 在Keepalived中有两种模式，分别是master->backup模式和backup->backup模式，这两种模式有什么区别呢？

在master->backup模式下，一旦主库宕掉，虚拟IP会自动漂移到从库，当主库修复后，keepalived启动后，还会把虚拟IP抢过来，即使你设置nopreempt（不抢占）的方式抢占IP的动作也会发生。在backup->backup模式下，当主库宕掉后虚拟IP会自动漂移到从库上，当原主恢复之后重启keepalived服务，并不会抢占新主的虚拟IP，即使是优先级高于从库的优先级别，也不会抢占IP。为了减少IP的漂移次数，生产中我们通常是把修复好的主库当做新主库的备库。

3. 本文是在MySQL主库高可用 -- 双主单活故障自动切换方案 基础上，结合自己对于MySQL的理解整理的。原文的脚本直接执行有点问题，思路有有点瑕疵，于是结合自己的实际环境，重新修改了一把。

4. 在测试的过程中，有以下几点需要注意：

1> master1检测脚本的逻辑是如果MySQL的服务不可用，则通过service keepalived stop命令来关闭keepalived，但在实际测试的过程中，却出现了即便执行了service keepalived stop命令，keepalived进程依然没有停止，导致MySQL的服务虽然不可用了，但VIP并不没有漂移到master2上。

优化方案：在执行service keepalived stop后，等待5s，再次检测keepalived的状态，如果keepalived没有关闭，则直接kill掉。

2> keepalived的日志默认是输出到/var/log/messages中，这样不便于查看。如何自定义keepalived的日志输出文件呢？

如果是用service启动的，修改/etc/sysconfig/keepalived文件

KEEPALIVED_OPTIONS="-D -d -S 0"

如果不是，则启动的时候指定以上参数，如：

/usr/local/keepalived/sbin/keepalived -d -D -S 0

修改/etc/syslog.conf

# keepalived -S 0
local0.*                                                /var/log/keepalived.log

重启syslog

RHEL 5&6：service syslog restart

RHEL 7：service rsyslog restart