生产环境中mysql+keepalive双主模式，keepalive守护进程实现双主切换提供数据库服务

mysql+keepalive实现浮动地址自动切换，由于keepalive无自带健康检查功能，所以必须自动编写健康检查守护进程（监控DB1和DB2数据库的监控状态，来保证浮动地址双机自动切换。）

一，部署说明及拓扑架构：

　　1、mysql安装在非root用户下（Mysql 版本5.7.18）

　　2、keepalive安装在root用户下

　　3、两台服务器安装mysql+keepalive，DB1真实ip地址为：10.112.188.70；DB2真实IP地址为：10.112.188.71；两台数据库除了建立同步账号rep1外，需建立用于守护进程账号check权限只有selecte。

　　4、DB1与DB2采用双主模式进行部署，但正常情况下只有DB1采用keepalive虚拟浮动IP：10.112.188.69提供服务；一旦DB2异常则由keepalive虚拟浮动IP地址：10.112.188.69自动切换至DB2主机上提供服务。

　　5、keepalive虚拟浮动地址切换思路（编写守护进程需要）；

　　　　1）正常情况下只有DB1采用keepalive虚拟浮动IP：10.112.188.69提供服务；DB2中keepalive为stop状态。DB1采用守护进程检查自身mysql状态是否异常：mysql -ucheck-p'check1' -S /iddbs/mysql.sock -e "select version();"一旦DB1自身数据库异常则自动stop keepalive 否则保证 keepalive是启动状态；且还需监控浮动虚拟地址：10.112.188.69下的mysql是否能连接成功，若能连接不成功启动keepalive（备注：监控浮动虚拟地址主要用途是：当DB1异常发生且恢复后（此时DB2已经自动启动keepalive采用虚拟IP地址接管提供服务），监控DB2是否已经释放浮动IP，保证IP地址不冲突。）

　　　　2）正常情况下只有DB1采用keepalive虚拟浮动IP：10.112.188.69提供服务；DB2中keepalive为stop状态。DB2采用守护进程监控DB1（真实IP：10.112.188.70）Mysql数据状态（是否异常）采用mysql -ucheck -p'check1' -h 10.112.188.70 -e "select version();" 一旦DB1中MYSQL数据库出现异常（或者宕机）DB2则启动keepalive并浮动VIP为：10.112.188.69

　　6、keepalive浮动地址切换逻辑图：

　　生产环境中mysql+keepalive双主模式，keepalive守护进程实现双主切换提供数据库服务

二、mysql双主部署及权限账号建立：

　　1、同步账号建立：

　　　　在10.112.188.70上建立：

　　　　mysql> grant replication slave on *.* to 'rep1'@'10.112.188.71' identified by 'rep1';

　　　　在10.112.188.71

　　　　mysql> grant replication slave on *.* to 'rep1'@'10.112.188.70' identified by 'rep1';

　　2、守护进程账号建立：

　　　　在DB1和DB2上均建立守护进程状态查询权限，只授予select：

　　　　mysql> GRANT select ON *.* TO 'check'@'10.112.188.70' identified by 'check';

　　　　mysql> GRANT select ON *.* TO 'check'@'10.112.188.71' identified by 'check';

　　3、同步操作以下在DB2上操作：

　　　　1)在DB1和DB2库配置文件中my.cnf开启log-bin

　　　　　　server-id = 1

　　　　　　# Uncomment the following if you want to log updates

　　　　　　log-bin=/iddbsdata/mysql-bin

　　　　2)mysql> show variables like 'log%'; #查看主库的binlog开关是否生效（ON状态）

　　　　　　+---------------------------------+---------------------------------------------+

　　　　　　| Variable_name | Value |

　　　　　　+---------------------------------+---------------------------------------------+

　　　　　　| log | OFF |

　　　　　　| log_bin | ON |

　　　　3)查看binlog节点位置。

　　　　　　mysql> show master status;

　　　　　　+------------------+----------+--------------+------------------+

　　　　　　+------------------+----------+--------------+------------------+

　　　　　　| mysql-bin.000002 | 1160 | | |

　　　　　　+------------------+----------+--------------+------------------+

　　　　　　1 row in set (0.00 sec)

　　　　4）同步（以10.112.188.71同步10.112.188.70为例，及DB2同步DB1）：

　　　　　　mysql>CHANGE MASTER TO

　　　　　　>MASTER_HOST='10.112.188.71',

　　　　　　>MASTER_PORT=3307,

　　　　　　>MASTER_USER='rep1',

　　　　　　>MASTER_PASSWORD='rep1',

　　　　　　>MASTER_LOG_FILE='mysql-bin.000002',

　　　　　　>MASTER_LOG_POS=1160;

　　4、在DB1上重复进行3操作。

　　5、注意，由于DB1和DB2属于是新库（新安装无数据），在同步之前省去了mysqldump，如果是有数据的数据库必须进行mysqldump操作才能进行数据同步。

三、keepalive部署（需在root用户下）

　　1、安装部署：

　　　　yum install -y openssl openssl-devel

　　　　gunzip keepalived-1.2.12.tar.gz

　　　　tar -xvf keepalived-1.2.12.tar

　　　　cd keepalived-1.2.12

　　　　./configure --prefix=/usr/local/keepalived

　　　　make

　　　　make install

　　　　ls -lart /usr/local/keepalived/

　　　　cp /usr/local/keepalived/sbin/keepalived /usr/sbin

　　　　cp /usr/local/keepalived/etc/sysconfig/keepalived /etc/sysconfig

　　　　cp /usr/local/keepalived/etc/rc.d/init.d/keepalived /etc/init.d

　　　　mkdir /etc/keepalived

　　　　cp /usr/local/keepalived/etc/keepalived/keepalived.conf /etc/keepalived

　　　　cd /etc

　　　　cp keepalived.conf keepalived.conf.bak

　　2、配置文件

[root@xqcldb001 /etc/keepalived]#cat /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {

   notification_email {

     acassen@firewall.loc

     failover@firewall.loc

     sysadmin@firewall.loc

   }

#   notification_email_from Alexandre.Cassen@firewall.loc

#   smtp_server 192.168.200.1

#   smtp_connect_timeout

   router_id mysql

}

vrrp_instance VI_1 {

    state BACKUP

    interface bond0

    virtual_router_id

    priority

    nopreempt

    advert_int

    authentication {

        auth_type PASS

        auth_pass

    }

    virtual_ipaddress {

#10.208.218.99  dev eth0 label eth0:

10.112.188.69/

}

}

virtual_server 10.112.188.69  {

    delay_loop

    lb_algo rr

    lb_kind NAT

    net_mask 255.255.255.0

    persistence_timeout

    protocol TCP

    real_server 10.112.188.70  {

        weight

    # notify_down /iddbs/app/check_mysql.sh

     TCP_CHECK {

         connect_timeout

         nb_get_retry

         connect_port

         }

   }

}

DB1

[root@xqcldb002 /etc/keepalived]#cat /etc/keepalived/keepalived.conf

! Configuration File for keepalived

global_defs {

   notification_email {

     acassen@firewall.loc

     failover@firewall.loc

     sysadmin@firewall.loc

   }

#   notification_email_from Alexandre.Cassen@firewall.loc

#   smtp_server 192.168.200.1

#   smtp_connect_timeout

   router_id mysql

}

vrrp_instance VI_1 {

    state BACKUP

    interface bond0

    virtual_router_id

    priority

    nopreempt

    advert_int

    authentication {

        auth_type PASS

        auth_pass

    }

    virtual_ipaddress {

#10.208.218.99  dev eth0 label eth0:

10.112.188.69/

}

}

virtual_server 10.112.188.69  {

    delay_loop

    lb_algo rr

    lb_kind NAT

    net_mask 255.255.255.0

    persistence_timeout

    protocol TCP

    real_server 10.112.188.71  {

        weight

    # notify_down /iddbs/app/check_mysql.sh

     TCP_CHECK {

         connect_timeout

         nb_get_retry

         connect_port

         }

   }

}

DB2

　　备注经测试VRRP state BACKUP两台必须均为BACKUP，否则一旦重启keepalive会导致网卡IP错误。

四、keepalive守护进程

#!/bin/sh

while true

do

/iddbs/bin/mysql -uroot -p'Siina@678' -S /iddbsdata/mysql.sock -e "select version();">/dev/null >&

if [ $? -eq   ]

        then

                /etc/init.d/keepalived start >/dev/null >&

        else

                /etc/init.d/keepalived stop >/dev/null >&

fi

sleep 

/iddbs/bin/mysql -h10.112.188. -ucheck -p'check1'  -P  -e "select version();">/dev/null >&

if [ $? -eq  ]

        then

                /etc/init.d/keepalived stop >/dev/null >&

        else

                /etc/init.d/keepalived start >/dev/null >&

fi

sleep

done

DB1守护进程

#!/bin/sh

while true

do

/iddbs/bin/mysql -h10.112.188. -ucheck -p'check1'  -P  -e "select version();">/dev/null >&

if [ $? -eq  ]

        then

                /etc/init.d/keepalived stop >/dev/null >&

                sleep

        else

                /etc/init.d/keepalived start >/dev/null >&

fi

sleep

done

DB2守护进程

　　守护进程启动：[root@xqcldb001 ~]#sh check_mysql.sh &

五、双主同步及模拟数据库异常keepalive自动切换浮动IP

1、正常情况下：

正常情况在备用服务器上bond0网卡上只有一个地址。

xqcldb002:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:a1:0b:48 brd ff:ff:ff:ff:ff:ff

inet 10.112.188.71/26 brd 10.112.188.127 scope global bond0

inet6 fe80::7279:90ff:fea1:b48/64 scope link

valid_lft forever preferred_lft forever

正常情况在主服务器上bond0网卡上有一个地址和一个虚拟IP地址

xqcldb001:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:7d:bf:3e brd ff:ff:ff:ff:ff:ff

inet 10.112.188.70/26 brd 10.112.188.127 scope global bond0

inet 10.112.188.69/26 scope global secondary bond0

inet6 fe80::7279:90ff:fe7d:bf3e/64 scope link

valid_lft forever preferred_lft forever

2、正常情况下：10.112.188.70数据库异常（模拟宕机）：

　　此种情况下70（DB1）bond0网卡上有一个真实地址，已经无浮动IP地址10.112.188.69

xqcldb001:/iddbs>sh mysql_start stop

stop Mysql. [ OK ]

xqcldb001:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:7d:bf:3e brd ff:ff:ff:ff:ff:ff

inet 10.112.188.70/26 brd 10.112.188.127 scope global bond0

inet6 fe80::7279:90ff:fe7d:bf3e/64 scope lin

　　71（DB2）bond0网卡上有一个真实地址，浮动IP地址10.112.188.69

xqcldb002:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:a1:0b:48 brd ff:ff:ff:ff:ff:ff

inet 10.112.188.71/26 brd 10.112.188.127 scope global bond0

inet 10.112.188.69/26 scope global secondary bond0

inet6 fe80::7279:90ff:fea1:b48/64 scope link

valid_lft forever preferred_lft forever

3、数据库恢复：

xqcldb001:/iddbs>sh mysql_start start

starting Mysql.... [ OK ]

xqcldb001:/iddbs>

xqcldb001:/iddbs>ps -ef | grep mysql

root 2182 24200 0 11:30 pts/2 00:00:00 sh check_mysql.sh #守护进程

iddbs 4485 1 0 11:35 pts/2 00:00:00 /bin/sh /iddbs/bin/mysqld_safe --defaults-file=/iddbs/my.cnf

iddbs 4890 4485 93 11:35 pts/2 00:00:06 /iddbs/bin/mysqld --defaults-file=/iddbs/my.cnf --basedir=/iddbs --datadir=/iddbsdata --plugin-dir=/iddbs/lib/plugin --log-error=/iddbsdata/mysql-error.log --pid-file=/iddbsdata/mysql.pid --socket=/iddbsdata/mysql.sock --port=3307

iddbs 4949 3619 0 11:36 pts/2 00:00:00 grep mysql

地址已经自动切换至DB1：

xqcldb001:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:7d:bf:3e brd ff:ff:ff:ff:ff:ff

inet 10.112.188.70/26 brd 10.112.188.127 scope global bond0

inet 10.112.188.69/26 scope global secondary bond0

inet6 fe80::7279:90ff:fe7d:bf3e/64 scope link

valid_lft forever preferred_lft forever

DB2只有真实地址而无虚拟浮动地址：

xqcldb002:/iddbs>ip a

10: bond0: <BROADCAST,MULTICAST,MASTER,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP

link/ether 70:79:90:a1:0b:48 brd ff:ff:ff:ff:ff:ff

inet 10.112.188.71/26 brd 10.112.188.127 scope global bond0

inet6 fe80::7279:90ff:fea1:b48/64 scope link

valid_lft forever preferred_lft forever