mysql双主双从技术

一、准备环境

[root@localhost ~]# vim /etc/hosts

192.168.40.154 master1     192.168.40.129 master2                                                                                         
192.168.40.138 slave1       192.168.40.128 slave2
注：四台虚拟机配置都相同，且防火墙关闭

二、双主设置

master1（其数据库中提前准备点库）

1、[root@localhost ~]# vim /etc/my.cnf

log_bin
server-id=1
gtid_mode=ON
enforce_gtid_consistency=1

2、[root@localhost ~]# systemctl restart mysqld

3、授权：mysql>grant replication slave,replication client on *.* to 'rep'@'192.168.40.%' identified by 'Zjz@5740';

4、mysql>  flush privileges;（刷新）

5、备份：mysqldump -p'Zjz@5740' --all-databases --single-transaction --master-data=2 --flush-logs > `date +%F`-mysql-all.sql

6、#scp -r 2017-1-1-mysql-all.sql master2:/tmp

7、观察二进制日志分割点

CHANGE MASTER TO MASTER_LOG_FILE='localhost-bin.000002', MASTER_LOG_POS=154;

master2

1、[root@localhost ~]#vim /etc/my.cnf（ 启动二进制日志，服务器ID，GTID）

log_bin
server-id=2
gtid_mode=ON
enforce_gtid_consistency=1

2、[root@localhost ~]# systemctl restart mysqld

还恢复手动同步数据

mysql>set sql_log_bin=0;

mysql>source /tmp/2017-1-1-mysql-full.sql

mysql>select * from master1db.master1tab;

3、设置主服务器

mysql> change master to
master_host='master1',
master_user='rep',
master_password='Zjz@5740',
master_auto_position=1;

start slave;

show slave status\G;

master1认master2为主

master2授权：mysql>grant replication slave,replication client on *.* to 'rep'@'192.168.40.%' identified by 'Zjz@5740';

flush privileges;（刷新）

master1:设置主服务器

mysql> change master to
master_host='master2',
master_user='rep',
master_password='Zjz@5740',
master_auto_position=1;

start slave;

show slave status\G;

mysqldump -p'Zjz@5740' --all-databases --single-transaction --master-data=2  --flush-logs > `date +%F`-mysql-all.sql

scp -r 2017-8-9-mysql-all.sql slave1:/tmp

scp -r 2017-8-9-mysql-all.sql slave2:/tmp

三、双从设置

slave1    # mysql -p'Zjz@5740' < /tmp/2017-8-9-mysql-all.sql

slave2     #mysql -p'Zjz@5740' < /tmp/2017-8-9-mysql-all.sql

启动从服务器ID，gtid

slave1                                                              slave2

#vim /etc/my.cnf                                              # vim /etc/my.cnf
server-id=3                                                       server-id=4
gtid_mode=ON                                                 gtid_mode=ON
enforce_gtid_consistency=1                            enforce_gtid_consistency=1
master-info-repository=TABLE                        master-info-repository=TABLE
relay-log-info-repository=TABLE                     relay-log-info-repository=TABLE

#systemctl restart mysqld                                  #systemctl restart mysqld

设置主服务器

slave1(认两个主服务器)

mysql> change master to                                             mysql> change master to
master_host='master1',                                                master_host='master2',
master_user='rep',                                                        master_user='rep',
master_password='Zjz@5740',                                    master_password='Zjz@5740',
master_auto_position=1 for channel 'master1';            master_auto_position=1 for channel 'master1';

mysql>start slave;

mysql>show slave status\G;（双yes成功）

slave2同上操作

四、总结

1、出现connettion，重启电脑试试

2、有些步骤先停止slave再操作

3、[root@slave1 ~]#  mysql -p'Zjz@5740' < /tmp/2019-08-29-mysql-all.sqlmysql:

[Warning] Using a password on the command line interface can be insecure.

ERROR 1840 (HY000) at line 24: @@GLOBAL.GTID_PURGED can only be set when @@GLOBAL.GTID_EXECUTED is empty.

解决办法：mysql> reset master;（进入slave1）

Query OK, 0 rows affected (0.10 sec)

mysql> \q

Bye

4、Slave_IO_Running: No

Slave_SQL_Running: Yes

报错：Last_IO_Error: Got fatal error 1236 from master when reading data from binary log: 'The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1, but the master has purged binary logs containing GTIDs that the slave requires.'

解决：这个应该是由于你在主库上执行过purge binary logs，然后当从库change master的时候，却要执行那些事务。
你可以在主库上先查找哪些gtid被purge了。
show global variables like 'gtid_purged';
然后拿着这个value，去从库上依次
stop slave;
set global gtid_purged = 'xxx'; # xxx是你主库上查到的value。
start slave;
这样能跳过执行被主库已经purge的事务了。