本次實驗將使用MySQL 5.6.x作為多“主”。

〇 測試環境：
OS：CentOS 6.5
master_1: 192.168.1.185(MySQL 5.6.30)
master_2: 192.168.1.186(MySQL 5.6.30)
slave: 192.168.1.1.187(MySQL 5.7.15)

〇 配置：
master_1相關配置：

1. [mysqld]
2. server_id = 185
3. log-bin = master_1
4. log-bin-index = master_1.index

master_2相關配置：

1. [mysqld]
2. server_id = 186
3. log-bin = master_2
4. log-bin-index = master_2.index

slave相關配置：

1. [mysqld]
2. server_id = 187
3. relay-log = slave
4. relay-log-index = slave.index
6. # 多源復制結構中的slave，官方要求master-info和relay-log-info存放處必須為TABLE.
7. # 如果為FILE，則在添加多個master時，會失敗：ER_SLAVE_NEW_CHANNEL_WRONG_REPOSITORY.
8. master-info-repository = TABLE
9. relay-log-info-repository = TABLE

〇 為master_1 & master_2上建立復制用戶：

1. GRANT REPLICATION SLAVE ON *.* to repl@‘192.168.1.187‘ IDENTIFIED BY ‘repl‘;
2. FLUSH PRIVILEGES;

〇 測試數據準備：
master_1測試數據：

1. master_1> FLUSH LOGS;
2. Query OK, 0 rows affected (0.00 sec)
4. master_1> SHOW BINARY LOGS; -- 記住當前binlog的name和position
5. +-----------------+-----------+
6. | Log_name | File_size |
7. +-----------------+-----------+
8. | master_1.000001 | 166 |
9. | master_1.000002 | 455 |
10. | master_1.000003 | 120 |
11. +-----------------+-----------+
12. 3 rows in set (0.00 sec)
14. master_1> CREATE DATABASE master_1;
15. Query OK, 1 row affected (0.03 sec)

master_2測試數據：

1. master_2> FLUSH LOGS;
2. Query OK, 0 rows affected (0.00 sec)
4. master_2> SHOW BINARY LOGS; -- 記住當前binlog的name和position
5. +-----------------+-----------+
6. | Log_name | File_size |
7. +-----------------+-----------+
8. | master_2.000001 | 166 |
9. | master_2.000002 | 455 |
10. | master_2.000003 | 120 |
11. +-----------------+-----------+
12. 3 rows in set (0.00 sec)
14. master_2> CREATE DATABASE master_2;
15. Query OK, 1 row affected (0.02 sec)

〇 在slave上執行：

1. salve> CHANGE MASTER TO
2. -> MASTER_HOST=‘192.168.1.185‘,
3. -> MASTER_USER=‘repl‘,
4. -> MASTER_PORT=3306,
5. -> MASTER_PASSWORD=‘repl‘,
6. -> MASTER_LOG_FILE=‘master_1.000003‘,
7. -> MASTER_LOG_POS=120
8. -> FOR CHANNEL ‘master_1‘;
9. Query OK, 0 rows affected, 2 warnings (0.02 sec) -- 此處產生的warnings是一些安全建議和警告，本實驗無視。
11. salve> CHANGE MASTER TO
12. -> MASTER_HOST=‘192.168.1.186‘,
13. -> MASTER_USER=‘repl‘,
14. -> MASTER_PORT=3306,
15. -> MASTER_PASSWORD=‘repl‘,
16. -> MASTER_LOG_FILE=‘master_2.000003‘,
17. -> MASTER_LOG_POS=120
18. -> FOR CHANNEL ‘master_2‘;
19. Query OK, 0 rows affected, 2 warnings (0.02 sec)
21. slave> START SLAVE;
22. Query OK, 0 rows affected (0.01 sec)
24. salve> SHOW DATABASES; -- 此時在master_1和master_2上的binlog events已經被正常的apply了
25. +--------------------+
26. | Database |
27. +--------------------+
28. | information_schema |
29. | master_1 |
30. | master_2 |
31. | mysql |
32. | performance_schema |
33. | sys |
34. +--------------------+
35. 6 rows in set (0.00 sec)

最後通過start slave status即可查到復制狀態

1. slave> SHOW SLAVE STATUS\G
2. *************************** 1. row ***************************
3. Slave_IO_State: Waiting for master to send event
4. Master_Host: 192.168.1.185
5. Master_User: repl
6. Master_Port: 3306
7. ……………………………………………………
8. Slave_IO_Running: Yes
9. Slave_SQL_Running: Yes
10. ……………………………………………………
11. Master_Server_Id: 185
12. Master_UUID: ee1f8704-58c4-11e6-95b5-000c297f23b7
13. Master_Info_File: mysql.slave_master_info
14. SQL_Delay: 0
15. SQL_Remaining_Delay: NULL
16. Slave_SQL_Running_State: Slave has read all relay log; waiting for more updates
17. ……………………………………………………
18. Channel_Name: master_1
19. Master_TLS_Version:
20. *************************** 2. row ***************************
21. Slave_IO_State: Waiting for master to send event
22. Master_Host: 192.168.1.186
23. Master_User: repl
24. Master_Port: 3306
25. Connect_Retry: 60
26. ……………………………………………………
27. Slave_IO_Running: Yes
28. Slave_SQL_Running: Yes
29. ……………………………………………………
30. Master_Server_Id: 186
31. Master_UUID: 53774f2d-7e14-11e6-8900-000c298e914c
32. Master_Info_File: mysql.slave_master_info
33. SQL_Delay: 0
34. SQL_Remaining_Delay: NULL
35. Slave_SQL_Running_State: Slave has read all relay log; waiting for more updates
36. ……………………………………………………
37. Channel_Name: master_2
38. Master_TLS_Version:
39. 2 rows in set (0.00 sec)

〇 測試：
master_1上操作：

1. master_1> CREATE TABLE master_1.test_table(id int);
2. Query OK, 0 rows affected (0.05 sec)
4. master_1> INSERT INTO master_1.test_table SELECT 666666;
5. Query OK, 1 row affected (0.01 sec)
6. Records: 1 Duplicates: 0 Warnings: 0

master_2上操作：

1. master_2> CREATE TABLE master_2.test_table(massage varchar(16));
2. Query OK, 0 rows affected (0.02 sec)
4. master_2> INSERT INTO master_2.test_table SELECT ‘嘿嘿嘿‘;
5. Query OK, 1 row affected (0.00 sec)
6. Records: 1 Duplicates: 0 Warnings: 0
8. master_2> INSERT INTO master_2.test_table SELECT ‘三陽之炎‘;
9. Query OK, 1 row affected (0.00 sec)
10. Records: 1 Duplicates: 0 Warnings: 0

slave上操作：

1. salve> SELECT id FROM master_1.test_table;
2. +--------+
3. | id |
4. +--------+
5. | 666666 |
6. +--------+
7. 1 row in set (0.00 sec)
10. salve> SELECT massage FROM master_2.test_table;
11. +--------------+
12. | massage |
13. +--------------+
14. | 嘿嘿嘿 |
15. | 三陽之炎 |
16. +--------------+
17. 2 rows in set (0.00 sec)

〇 其他相關語法：

1. START/STOP/RESET ALL/RESET SLAVE FOR CHANNEL ‘XXX‘;
3. SHOW SLAVE STATUS FOR CHANNEL ‘XXX‘;

ps.
與上述傳統position方式類似，GTID方式配置起來也類似，開啟GTID後，需要註意使用FOR CHANNEL ‘xxx‘關鍵字即可，比如：

1. CHANGE MASTER TO
2. MASTER_HOST=‘‘,
3. MASTER_USER=‘repl‘,
4. MASTER_PORT=3306,
5. MASTER_PASSWORD=‘repl‘,
6. MASTER_AUTO_POSITION = 1
7. FOR CHANNEL ‘master_1‘;

多臺主機的schema名字不可以一樣，（比如master_1為db_00 ... db_09共10庫，master_2為db_10 ... db_19，master_3為db_20 ... db_29 ……）

MySQL5.6-》5.7