原创

利用MHA实现mysql主从切换之在线手动切换

MHA是一位日本MySQL大牛用Perl写的一套MySQL故障切换方案,来保证数据库系统的高可用.在宕机的时间内(通常10—30秒内),完成故障切换,部署MHA,可避免主从一致性问题,节约购买新服务器的费用,不影响服务器性能,易安装,不改变现有部署。


一、准备工作

1、分别在Master和Slave执行如下,方便mha检查复制:

grant all privileges on *.* to 'root'@'10.1.1.231' identified by 'rootpass';
grant all privileges on *.* to 'root'@'10.1.1.234' identified by 'rootpass';
grant replication slave on *.* to 'jpsync'@'10.1.1.231' identified by 'jppasswd';
grant replication slave on *.* to 'jpsync'@'10.1.1.234' identified by 'jppasswd';
flush privileges;


2、将master设置为只读

mysql> set global read_only=1;
Query OK, 0 rows affected (0.00 sec)
mysql> show variables like 'read_only';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| read_only     | ON    |
+---------------+-------+
1 row in set (0.00 sec)


交互模式:

#masterha_master_switch --master_state=alive 
--conf=/etc/masterha/app1.cnf    --new_master_host=10.1.1.231  
--new_master_port=63306


或非交互模式:

#masterha_master_switch --master_state=alive 
--conf=/etc/masterha/app1.cnf    --new_master_host=10.1.1.231  
--new_master_port=63306 —interactive=0

二、切换完以后,如何让10.1.1.231为主,10.1.1.234为从,操作步骤:

1、主上执行:

mysql> show master status;
+-------------------------+----------+--------------+--------------------------------------+-------------------+
| File                    | Position | Binlog_Do_DB | Binlog_Ignore_DB                     | Executed_Gtid_Set |
+-------------------------+----------+--------------+--------------------------------------+-------------------+
| mysql-master-bin.000013 |      120 | denovo_ng    | mysql,denovo,test,information_schema |                   |
+-------------------------+----------+--------------+--------------------------------------+-------------------+
1 row in set (0.00 sec)


2、在10.1.1.234上执行如下sql命令;

change master to master_host='10.1.1.231',master_port=63306,master_user='jpsync',
master_password='jppasswd', master_log_file='mysql-master-bin.000013',master_log_pos=120;
mysql> show slave status\G;
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 10.1.1.231
                  Master_User: jpsync
                  Master_Port: 63306
                Connect_Retry: 60
              Master_Log_File: mysql-master-bin.000013
          Read_Master_Log_Pos: 120
               Relay_Log_File: compute-0-52-relay-bin.000002
                Relay_Log_Pos: 290
        Relay_Master_Log_File: mysql-master-bin.000013
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes


3、查看master状态,并测试

mysql> show slave hosts;
+-----------+------+-------+-----------+--------------------------------------+
| Server_id | Host | Port  | Master_id | Slave_UUID                           |
+-----------+------+-------+-----------+--------------------------------------+
|      1052 |      | 63306 |      1025 | e25a3e4a-39c0-11e4-80cb-00259086c4b6 |
+-----------+------+-------+-----------+--------------------------------------+
1 row in set (0.00 sec)


主库10.1.1.231上插入记录

mysql> insert into  test_slave_002 values(555551111,1,55555,99999,44.11,2222,91919);
Query OK, 1 row affected (0.00 sec)


从库查询记录已经存在

mysql> select * from test_slave_002 where id=555551111;
+-----------+-----+-----------+--------------+----------+----------------+--------------+
| id        | tag | ticket_id | candidate_id | duration | source_file_id | source_start |
+-----------+-----+-----------+--------------+----------+----------------+--------------+
| 555551111 |   1 |     55555 |        99999 |    44.11 |           2222 |        91919 |
+-----------+-----+-----------+--------------+----------+----------------+--------------+
1 row in set (0.00 sec)


4、更新配置文件:

更新主库my.cnf配置添加

skip_slave_start


注意:防止重启数据库,启动slave进程,导致数据不一致。

更新从库my.cnf配置添加,设置slave库为只读:

read_only=1
relay_log_purge=0


然后重启主库和从库,观察库的信息:

主库信息:

mysql> show processlist;
+----+--------+------------------+------+-------------+------+-----------------------------------------------------------------------+------------------+
|
 Id | User   | Host             | db   | Command     | Time | 
State                                                                 | 
Info             |
+----+--------+------------------+------+-------------+------+-----------------------------------------------------------------------+------------------+
| 
 1 | jpsync | 10.1.1.234:49085 | NULL | Binlog Dump |   17 | Master has 
sent all binlog to slave; waiting for binlog to be updated | 
NULL             |
|  2 | root   | localhost        | NULL | 
Query       |    0 | 
init                                                                  | 
show processlist |
+----+--------+------------------+------+-------------+------+-----------------------------------------------------------------------+------------------+
2 rows in set (0.00 sec)
mysql> show master status;
+-------------------------+----------+--------------+--------------------------------------+-------------------+
| File                    | Position | Binlog_Do_DB | Binlog_Ignore_DB                     | Executed_Gtid_Set |
+-------------------------+----------+--------------+--------------------------------------+-------------------+
| mysql-master-bin.000014 |      120 | denovo_ng    | mysql,denovo,test,information_schema |                   |
+-------------------------+----------+--------------+--------------------------------------+-------------------+
1 row in set (0.00 sec)


从库信息:

mysql> show slave status\G;
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 10.1.1.231
                  Master_User: jpsync
                  Master_Port: 63306
                Connect_Retry: 60
              Master_Log_File: mysql-master-bin.000014
          Read_Master_Log_Pos: 120
               Relay_Log_File: compute-0-52-relay-bin.000005
                Relay_Log_Pos: 290
        Relay_Master_Log_File: mysql-master-bin.000014
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
mysql> show processlist;
+----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+
|
 Id | User        | Host      | db   | Command | Time | 
State                                                                      
 | Info             |
+----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+
| 
 1 | system user |           | NULL | Connect |   58 | Waiting for 
master to send event                                            | 
NULL             |
|  2 | system user |           | NULL | Connect 
|   58 | Slave has read all relay log; waiting for the slave I/O thread 
to update it | NULL             |
|  3 | root        | localhost | 
NULL | Query   |    0 | 
init                                                                       
 | show processlist |
+----+-------------+-----------+------+---------+------+-----------------------------------------------------------------------------+------------------+
3 rows in set (0.00 sec)


关注下方微信公众号“Java精选”(w_z90110),回复关键字领取资料:如HadoopDubboCAS源码等等,免费领取资料视频和项目。 

涵盖:程序人生、搞笑视频、算法与数据结构、黑客技术与网络安全、前端开发、Java、Python、Redis缓存、Spring源码、各大主流框架、Web开发、大数据技术、Storm、Hadoop、MapReduce、Spark、elasticsearch、单点登录统一认证、分布式框架、集群、安卓开发、iOS开发、C/C++、.NET、Linux、Mysql、Oracle、NoSQL非关系型数据库、运维等。

评论

分享:

支付宝

微信