接下来的replication过程不详细说明了,同步完成后我们经过相当长时间的观察,再也没1062报错了,如下所示:

    mysql> show slave status G;
    *************************** 1. row ***************************
                   Slave_IO_State: Waiting for master to send event
                      Master_Host: 192.168.11.174
                      Master_User: rep1
                      Master_Port: 3306
                    Connect_Retry: 60
                  Master_Log_File: mysql-bin.000008
              Read_Master_Log_Pos: 27880
                   Relay_Log_File: centos3-relay-bin.000002
                    Relay_Log_Pos: 28025
            Relay_Master_Log_File: mysql-bin.000008
                 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: 27880
                  Relay_Log_Space: 28182
                  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
    Master_SSL_Verify_Server_Cert: No
                    Last_IO_Errno: 0
                    Last_IO_Error: 
                   Last_SQL_Errno: 0
                   Last_SQL_Error: 
    1 row in set (0.00 sec)

  以前的项目也比较多的牵涉到InnoDB数据库的备份及replication,较多的一个做法是停库进行replication,虽然也是解决问题的一种思路,但毕竟属于停机维护,在一些特殊应用场景中是不允许的,我们应该多尝试采用mysqldump这种逻辑备份方式来取master主机快照。

  目前在测试ext3和ext4文件系统对数据库的影响,感觉MySQL性能优化不大;反而,固态SSD硬盘对于提升磁盘I/O方面确实影响不少,这方面有研究的朋友也欢迎来信交流。