旧主库:
切换前将主库的表上锁,防止切换期间数据写到主库
mysql> flush tables with read lock;
旧备库:
mysql> show processlist ;
Slave has read all relay log; waiting for the slave I/O thread to update it
mysql> show slave status \G
旧主库:
mysql> show processlist ;
Master has sent all binlog to slave; waiting for binlog to be updated
旧从库
mysql> STOP SLAVE IO_THREAD;
mysql> SHOW PROCESSLIST;
确保状态为:has read all relay log
旧主库:
mysql> show master status
mysql> STOP SLAVE;
mysql> RESET MASTER;
mysql> RESET SLAVE;
mysql> show master status \G
关闭旧主库
/etc/init.d/mysql stop
待主库应用切到备库,关闭原主库的表lock状态
unlock tables;
从库变主库
mysql> RESET MASTER;
mysql> RESET SLAVE;
主库重新做备库
新主库:
mysql> grant replication slave on *.* to 'root'@'192.168.1.29' identified by 'test123';
vi /etc/my.cnf
log-bin=mysql-masterbin
service mysql restart
mysql> flush tables with read lock;
mysql> show master status; 记下file和positon
/usr/bin/mysqldump -uroot -ptest123 --log-error=mysqldump47.log --all-databases > mysqldump47.sql
mysql> unlock tables;
scp mysqldump47.sql 192.168.234.29:/root/
新备库:
mysql> stop slave;
mysql> /usr/bin/mysql -uroot -ptest123 < mysqldump47.sql
mysql> change master to master_host='192.168.1.47',master_user='root',master_password='test123',master_log_file='mysql-masterbin.00001' ,master_log_pos=65238;
mysql> start slave;
新主库操作数据,新备库验证一下
附录:
不重启MySQL修改配置参数(万不得已使用)
system gdb -p $(pidof mysqld) -ex "set log_bin = mysql-masterbin47" -batch
数据库只读锁定命令,防止导出数据库的时候有数据写入
flush tables with read lock;
解除锁定
unlock tables;
备份或者准备主备切换前设置数据为只读状态(具有超级管理权限的用户仍然可以DML操作表)
SET GLOBAL READ_ONLY = ON;
SET GLOBAL READ_ONLY = OFF;
SHOW VARIABLES LIKE '%read_only%';
set global read_only=on/off是DBA经常用的一个操作:进行主备切换的时候,一般都会先对主库进行只读操作(on),然后主备同步完成后,再把备库置为可读写(off)。这样可以避免切换的过程中双写引起脏数据。
mysqld.cc中同时定义了2个变量:my_bool read_only= 0, opt_readonly= 0; opt_readonly是当前系统的read_only状态,read_only是要把read_only设置成的值。
来自 “ ITPUB博客 ” ,链接:http://blog.itpub.net/29320885/viewspace-1733654/,如需转载,请注明出处,否则将追究法律责任。
转载于:http://blog.itpub.net/29320885/viewspace-1733654/