master/slave 相同server_id引起的同步失败

昨天在做MySQL SwitchOver遇到一个诡异的想象,切换前后的结构图如下：

当我把一切都切换好之后，应其他需求，重启了04上的mysql,然后show slave status\G发现报错：

Last_IO_Error: Fatal error: The slave I/O thread stops because master and slave have equal MySQL server ids; these ids must be different for replication to work (or the --replicate-same-server-id option must be used on slave but this does not always make sense; please check the manual before using it).

查看了一下03/04的server_id才发现，原来它们的server_id相同，我们知道，master/slave的server_id是不能相同的，但是为什么在我重启之前我show slave status\G的时候没有发现报错？？？ 并且我特别检查了一下slave几个参数：

Master_Log_File: mysql-bin.000001Read_Master_Log_Pos: 38593Relay_Master_Log_File: mysql-bin.000001Exec_Master_Log_Pos: 38593

表示slave已经完全追上master,并且通过SQL_thread线程执行。

问题：在server_id相同的情况下，slave为什么之前没有报错呢，并且还可以继续应用master的binlog呢？？？

废话不多说了，测试如下：

在一个master/slave环境下,刚开始server_id不同，一切都很正常，数据也可以同步过去，我slave的server_id修改成与master的相同：

master:mysql> select @@server_id;+-------------+| @@server_id |+-------------+|      583306 |+-------------+1 row in set (0.00 sec)slave:mysql> select @@server_id;+-------------+| @@server_id |+-------------+|      593306 |+-------------+1 row in set (0.00 sec)mysql> set global server_id=583306;mysql> select @@server_id;+-------------+| @@server_id |+-------------+|      583306 |+-------------+1 row in set (0.00 sec)

然后show slave status\G，还没有报错；

于是我在master上insert了一条数据，观察一下slave有没有同步过去，show slave status\G 发现，position的值变化了，但是当我select这个表的时候，确没有找到刚才插入的那条数据，说明数据没有同步过去。

Master_Log_File: mysql-bin.000001Read_Master_Log_Pos: 38857Relay_Master_Log_File: mysql-bin.000001Exec_Master_Log_Pos: 38857

通过show relaylog events命令也可以看到，insert语句并没有写到relay log中，所以数据没有同步过来，这就说明当我们修改server_id之后，show slave status\G显示的结果并不是这么可靠，

mysql> SHOW RELAYLOG EVENTS in 'sht-sgmhadoopdn-02-relay-bin.000005';+-------------------------------------+------+----------------+-----------+-------------+---------------------------------------------------------------------+| Log_name                            | Pos  | Event_type     | Server_id | End_log_pos | Info                                                                |+-------------------------------------+------+----------------+-----------+-------------+---------------------------------------------------------------------+| sht-sgmhadoopdn-02-relay-bin.000005 |    4 | Format_desc    |    593306 |         123 | Server ver: 5.7.21-log, Binlog ver: 4                               || sht-sgmhadoopdn-02-relay-bin.000005 |  123 | Previous_gtids |    593306 |         194 | 8b94d944-34c8-11e8-9e15-0050568211bd:1-144                          || sht-sgmhadoopdn-02-relay-bin.000005 |  194 | Rotate         |    583306 |           0 | mysql-bin.000001;pos=4                                              || sht-sgmhadoopdn-02-relay-bin.000005 |  241 | Format_desc    |    583306 |         123 | Server ver: 5.7.21-log, Binlog ver: 4                               || sht-sgmhadoopdn-02-relay-bin.000005 |  360 | Rotate         |         0 |         407 | mysql-bin.000001;pos=154                                            || sht-sgmhadoopdn-02-relay-bin.000005 |  407 | Rotate         |         0 |         454 | mysql-bin.000001;pos=37623                                          || sht-sgmhadoopdn-02-relay-bin.000005 |  454 | Gtid           |    583306 |       37688 | SET @@SESSION.GTID_NEXT= '8b94d944-34c8-11e8-9e15-0050568211bd:145' || sht-sgmhadoopdn-02-relay-bin.000005 |  519 | Query          |    583306 |       37807 | use `testdb`; DROP TABLE `t1` /* generated by server */             || sht-sgmhadoopdn-02-relay-bin.000005 |  638 | Gtid           |    583306 |       37872 | SET @@SESSION.GTID_NEXT= '8b94d944-34c8-11e8-9e15-0050568211bd:146' || sht-sgmhadoopdn-02-relay-bin.000005 |  703 | Query          |    583306 |       37964 | use `testdb`; create table t1 like t2                               || sht-sgmhadoopdn-02-relay-bin.000005 |  795 | Gtid           |    583306 |       38029 | SET @@SESSION.GTID_NEXT= '8b94d944-34c8-11e8-9e15-0050568211bd:147' || sht-sgmhadoopdn-02-relay-bin.000005 |  860 | Query          |    583306 |       38148 | use `testdb`; DROP TABLE `t1` /* generated by server */             || sht-sgmhadoopdn-02-relay-bin.000005 |  979 | Gtid           |    583306 |       38213 | SET @@SESSION.GTID_NEXT= '8b94d944-34c8-11e8-9e15-0050568211bd:148' || sht-sgmhadoopdn-02-relay-bin.000005 | 1044 | Query          |    583306 |       38329 | use `testdb`; create table t1(c1 int,c2 varchar(20))                || sht-sgmhadoopdn-02-relay-bin.000005 | 1160 | Gtid           |    583306 |       38394 | SET @@SESSION.GTID_NEXT= '8b94d944-34c8-11e8-9e15-0050568211bd:149' || sht-sgmhadoopdn-02-relay-bin.000005 | 1225 | Query          |    583306 |       38468 | BEGIN                                                               || sht-sgmhadoopdn-02-relay-bin.000005 | 1299 | Table_map      |    583306 |       38518 | table_id: 228 (testdb.t1)                                           || sht-sgmhadoopdn-02-relay-bin.000005 | 1349 | Write_rows     |    583306 |       38562 | table_id: 228 flags: STMT_END_F                                     || sht-sgmhadoopdn-02-relay-bin.000005 | 1393 | Xid            |    583306 |       38593 | COMMIT /* xid=3689 */                                               |+-------------------------------------+------+----------------+-----------+-------------+---------------------------------------------------------------------+

最后当我stop slave; && start slave;之后，就出现了文章一开始报错的server_id相同的错误。

总结：

（1）master/slave要保证server_id不同；

（2）当修改server_id之后，需要执行stop/start slave;最好重启mysql数据库；

（3）不要仅仅通过show slave status\G查看同步是否正确，还在查看数据是否真正的同步过去;