AWS EC2 MySQL迁移到RDS案例

Amazon Relational Database Service (Amazon RDS) 是一种Web 服务，可让用户更轻松地在云中设置、操作和扩展关系数据库。它可以为行业标准关系数据库提供经济高效且可以调节大小的容量，并管理常见

数据库管理任务。

现有的环境如下,计划将MySQL主库billingdb迁移到RDS上,减少运维成本.

计划迁移之后的环境如下

现在开始迁移过程

1. 在当前的RDS slave上记录需要过滤的表(RDS有自己的系统表,当RDS slave变成master后,这些表不需要被同步到EC2上的slave上)

mysql> show slave status\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 10.188.102.92
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: mysql-bin.000861
          Read_Master_Log_Pos: 18626649
               Relay_Log_File: relaylog.005026
                Relay_Log_Pos: 685
        Relay_Master_Log_File: mysql-bin.000861
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB:
          Replicate_Ignore_DB:
           Replicate_Do_Table:
       Replicate_Ignore_Table: mysql.rds_sysinfo,innodb_memcache.cache_policies,mysql.rds_replication_status,innodb_memcache.config_options,mysql.rds_history,mysql.plugin,mysql.rds_monitor

2. 在RDS slave上创建复制及相关账户(创建RDS作为备库的时候,并没有导入系统表,所以不会有主库上的数据库用户)

mysql> grant replication slave on *.* to 'repl'@'%'  identified by 'repl';
mysql> flush privileges;

3. 修改RDS slave上二进制日志保留时间(默认情况下,RDS并不会保留二进制日志,这样RDS切为master后就没有二进制日志,也就谈不上将EC2转换为slave连上RDS了)

mysql> call mysql.rds_show_configuration;
mysql> call mysql.rds_set_configuration('binlog retention hours', 72);

4.  记录当前master的session连接信息

/usr/local/mysql/bin/mysql -u root -p -e "show processlist"|awk '{print $3}'|grep ":"|awk '{print substr($0,0,index($0,":")-1) }'|sort|uniq -c
       10.188.100.36
       10.188.104.196
       ec2-backoffice-.mypna.com
      ec2-billingcommerce-.mypna.com
      ec2-billingcommerce-.mypna.com
      ec2-billingen-.mypna.com
      ec2-billingen-.mypna.com
      ec2-billingengines-.mypna.com
      ec2-billingentest-.mypna.com
      ec2-billingentest-.mypna.com
      ec2-billingiden-.mypna.com
      ec2-billingiden-.mypna.com
      ec2-billingoffer-.mypna.com
      ec2-billingoffer-.mypna.com
      ec2-billingoffer-.mypna.com
      ec2-billingui-.mypna.com
      ec2-billingui-.mypna.com
       ec2-cliwebreport-.mypna.com
       ec2-dbmon-.mypna.com

5. 准备第4步主机名到IP地址的解析(RDS并不能解析出客户端主机名,迁移到RDS后,检查session连接的时候会方便很多)

IP address	Hostname
10.188.100.36	ec2-billingdb-04
10.188.104.196	ec2-csr-tracking-01
10.188.3.22	ec2-backoffice-01.mypna.com
10.188.1.102	ec2-billingcommerce-01.mypna.com
10.188.1.105	ec2-billingcommerce-02.mypna.com
10.188.1.252	ec2-billingen-01.mypna.com
10.188.1.59	ec2-billingen-02.mypna.com
10.188.1.189	ec2-billingengines-01.mypna.com
10.188.1.246	ec2-billingentest-01.mypna.com
10.188.1.232	ec2-billingentest-02.mypna.com
10.188.1.15	ec2-billingiden-01.mypna.com
10.188.1.16	ec2-billingiden-02.mypna.com
10.188.101.54	ec2-billingoffer-01.mypna.com
10.188.101.128	ec2-billingoffer-02.mypna.com
10.188.101.63	ec2-billingoffer-03.mypna.com
10.188.1.235	ec2-billingui-01.mypna.com
10.188.1.236	ec2-billingui-02.mypna.com
10.188.102.92	ec2-cliwebreport-01.mypna.com
10.188.100.65	ec2-dbmon-01.mypna.com
10.188.102.92	ec2-cliwebreport-01.mypna.com

6. 将主库master重启到只读状态,并记录当前的master日志位置信息(这样做是为了让所有的slave都达到一致性状态)

/etc/init.d/mysql restart --read-only

mysql> show master status\G
*************************** 1. row ***************************
             File: mysql-bin.000865
         Position: 120
     Binlog_Do_DB:
 Binlog_Ignore_DB:
Executed_Gtid_Set:
1 row in set (0.00 sec)

7. 稍等片刻,检查所有EC2 slave,确保所有的EC2的slave已经赶上master的二进制日志位置

mysql> show slave status\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 10.188.100.73
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: mysql-bin.000865          #看该行
          Read_Master_Log_Pos: 120                       #看该行
               Relay_Log_File: ec2-billingdb-04-relay-bin.000047
                Relay_Log_Pos: 279
        Relay_Master_Log_File: mysql-bin.000865
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB:
          Replicate_Ignore_DB:
           Replicate_Do_Table:
       Replicate_Ignore_Table:
      Replicate_Wild_Do_Table:
  Replicate_Wild_Ignore_Table:
                   Last_Errno: 0
                   Last_Error:
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 120                             #看该行
              Relay_Log_Space: 618
              Until_Condition: None
               Until_Log_File:
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File:
           Master_SSL_CA_Path:
              Master_SSL_Cert:
            Master_SSL_Cipher:
               Master_SSL_Key:
        Seconds_Behind_Master: 0                     #看该行

8. EC2的机器ec2-billingdb-06是RDS的master,所以RDS的二进制日志要看是否追上ec2-billingdb-06

• 在ec2-billingdb-06查看

mysql> show master status\G
*************************** 1. row ***************************
             File: mysql-bin.000861
         Position: 18626649
     Binlog_Do_DB:
 Binlog_Ignore_DB:
Executed_Gtid_Set:
1 row in set (0.00 sec)

• 在RDS上查看

mysql> show slave status\G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: 10.188.102.92
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: mysql-bin.000861     #这里和上一步骤06机器的二进制日志一致
          Read_Master_Log_Pos: 18626649             #这里和上一步骤06的机器的二进制日志一致
               Relay_Log_File: relaylog.005026
                Relay_Log_Pos: 685
        Relay_Master_Log_File: mysql-bin.000861
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
              Replicate_Do_DB:
          Replicate_Ignore_DB:
           Replicate_Do_Table:
       Replicate_Ignore_Table: mysql.rds_sysinfo,innodb_memcache.cache_policies,mysql.rds_replication_status,innodb_memcache.config_options,mysql.rds_history,mysql.plugin,mysql.rds_monitor
      Replicate_Wild_Do_Table:
  Replicate_Wild_Ignore_Table:
                   Last_Errno: 0
                   Last_Error:
                 Skip_Counter: 0
          Exec_Master_Log_Pos: 18626649            #这里和上一步骤06机器的二进制日志一致
              Relay_Log_Space: 79732
              Until_Condition: None
               Until_Log_File:
                Until_Log_Pos: 0
           Master_SSL_Allowed: No
           Master_SSL_CA_File:
           Master_SSL_CA_Path:
              Master_SSL_Cert:
            Master_SSL_Cipher:
               Master_SSL_Key:
        Seconds_Behind_Master: 0

8. 将master主库数据库关机

/etc/init.d/mysql stop 

9. 记录RDS slave的二进制日志信息

mysql> show master status;
*************************** 1. row ***************************
             File: mysql-bin-changelog.007492
         Position: 120
     Binlog_Do_DB:
 Binlog_Ignore_DB:
Executed_Gtid_Set:
1 row in set (0.00 sec)

 10. 清除RDS的slave信息,使其作为独立的主库运行

mysql>CALL mysql.rds_stop_replication;
mysql>CALL mysql.rds_reset_external_master;
mysql>show slave status\G;
mysql>show master status\G;
File: mysql-bin-changelog.007492
         Position: 1025

11. 将步骤1中需要过滤的表添加到原master的配置文件/etc/my.cnf里

replicate-ignore-table=mysql.rds_sysinfo
replicate-ignore-table=innodb_memcache.cache_policies
replicate-ignore-table=mysql.rds_replication_status
replicate-ignore-table=innodb_memcache.config_options
replicate-ignore-table=mysql.rds_history
replicate-ignore-table=mysql.plugin
replicate-ignore-table=mysql.rds_monitor
replicate-ignore-table=mysql.rds_heartbeat2

12. 将原master以只读的方式启动(确保应用不会连接进来写脏数据)

/etc/init.d/mysql start --read-only 

13. 将原master作为slave连接到RDS

mysql> change master to
     master_host='RDS endpoint name',
     master_user='repl',
     master_password='repl',
     master_port=3306,
     master_log_file='mysql-bin-changelog.xxxxxx',
     master_log_pos=xxx;
mysql> start slave;
mysql> show slave status\G;

14, 将应用解析到RDS上,并检查RDS的session信息,确保应用可以正常接入