MySQL GTID 错误处理汇总

MySQL GTID是在传统的mysql主从复制的基础之上演化而来的产物，即通过UUID加上事务ID的方式来确保每一个事物的唯一性。这样的操作方式使得我们不再需要关心所谓的log_file和log_Pos，只是简单的告诉从库，从哪个服务器上去找主库就OK了。简化了主从的搭建以及failover的过程，同时比传统的复制更加安全可靠。由于GTID是连续没有空洞的，因此主从库出现数据冲突时，可以通过注入空事物的方式进行跳过。本文主要讲述GTID主从架构的错误处理方式。

http://blog.csdn.net/leshami/article/details/52778480

一、GTID的相关特性

配置MySQL GTID 主从复制 
基于mysqldump搭建gtid主从

二、GTID如何跳过事务冲突

    很多无法预料的情形导致mysql主从发生事务冲突，主从失败或停止的情形，即需要修复主从
    对于GTID方式的主从架构而言，更多的是处理事务冲突来修复主从
    GTID不支持通过传统设置sql_slave_skip_counter方法来跳过事务
    方法：通过注入空事务来填补事务空洞，等同于传统复制的(set global sql_slave_skip_counter = 1)
    步骤：
            stop slave；
            set gtid_next='xxxxxxx:N'; --指定下一个事务执行的版本，即想要跳过的GTID
            begin;
            commit;  --注入一个空事物
            set gtid_next='AUTOMATIC' --自动的寻找GTID事务。
            start slave; --开始同步

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12

三、GTID事务冲突的几种常见类型

    1、主库新增记录，从库提示主键冲突
    2、主库对象可更新，从库无对应的对象可更新
    3、主库对象可删除，从库无对应的对象可删除
    4、通过延迟从修复主库意外删除的对象
    5、主库日志被purged的情形

• 1
• 2
• 3
• 4
• 5
• 6

四、示例演示

当前演示的主从架构图
# mysqlrplshow --master=root:pass@192.168.1.233:3306 --discover-slaves-login=root:pass --verbose
WARNING: Using a password on the command line interface can be insecure.
# master on 192.168.1.233: ... connected.
# Finding slaves for master: 192.168.1.233:3306

# Replication Topology Graph
192.168.1.233:3306 (MASTER)
   |
   +--- 192.168.1.245:3306 [IO: Yes, SQL: Yes] - (SLAVE)
   |
   +--- 192.168.1.247:3306 [IO: Yes, SQL: Yes] - (SLAVE)

(root@192.168.1.233)[tempdb]>show slave hosts;
+-----------+---------------+------+-----------+--------------------------------------+
| Server_id | Host          | Port | Master_id | Slave_UUID                           |
+-----------+---------------+------+-----------+--------------------------------------+
|       245 | 192.168.1.245 | 3306 |       233 | 78336cdc-8cfb-11e6-ba9f-000c29328504 |
|       247 | 192.168.1.247 | 3306 |       233 | 13a26fc1-555a-11e6-b5e0-000c292e1642 |
+-----------+---------------+------+-----------+--------------------------------------+

--演示的mysql版本
(root@192.168.1.233)[tempdb]>show variables like 'version';
+---------------+------------+
| Variable_name | Value      |
+---------------+------------+
| version       | 5.7.12-log |
+---------------+------------+

--查看gtid是否开启
(root@192.168.1.233)[tempdb]>show variables like '%gtid_mode%';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| gtid_mode     | ON    |
+---------------+-------+

--主库上面可以看到基于gtid的dump线程，如下
(root@192.168.1.233)[tempdb]>show processlist;
+----+------+-----------------------+--------+------------------+------+
| Id | User | Host                  | db     | Command          | Time |
+----+------+-----------------------+--------+------------------+------+
| 17 | repl | node245.edq.com:52685 | NULL   | Binlog Dump GTID | 2738 |
| 18 | repl | node247.edq.com:33516 | NULL   | Binlog Dump GTID | 2690 |
| 24 | root | localhost             | tempdb | Query            |    0 |
+----+------+-----------------------+--------+------------------+------+

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46

1、从库报主键重复(Errno: 1062)

(root@Master)[tempdb]>create table t1 (
            -> id tinyint not null primary key,ename varchar(20),blog varchar(50));

(root@Master)[tempdb]>insert into t1 
            -> values(1,'leshami','http://blog.csdn.net/leshami');

(root@Master)[tempdb]>insert into t1 
            -> values(2,'robin','http://blog.csdn.net/robinson_0612');

(root@Master)[tempdb]>set sql_log_bin=off;

(root@Master)[tempdb]>delete from t1 where ename='robin';

(root@Master)[tempdb]>set sql_log_bin=on;

(root@Master)[tempdb]>insert into t1 
            -> values(2,'robin','http://blog.csdn.net/robinson_0612');

-- 从库状态报错，提示重复的primary key
(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
Last_Errno: 1062
Last_Error: Could not execute Write_rows event on table tempdb.t1; Duplicate entry '2' for key 'PRIMARY', 
                        Error_code: 1062; handler error HA_ERR_FOUND_DUPP_KEY; 
                        the event's master log node233-binlog.000004, end_log_pos 4426
Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-90
 Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-89
     Auto_Position: 1

-- 如下解决方案，可以通过删除重库的这条记录
(root@Slave)[tempdb]>stop slave;

(root@Slave)[tempdb]>delete from t1 where ename='robin';

(root@Slave)[tempdb]>start slave;

(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-90
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-90,
 78336cdc-8cfb-11e6-ba9f-000c29328504:1  --这里多了一个GTID，注意这个是从库上执行的，这里的UUID跟IP 245的UUID一致
                Auto_Position: 1
         Replicate_Rewrite_DB: 
                 Channel_Name: 
           Master_TLS_Version: 

(root@Slave)[tempdb]>show variables like '%uuid%';
+---------------+--------------------------------------+
| Variable_name | Value                                |
+---------------+--------------------------------------+
| server_uuid   | 78336cdc-8cfb-11e6-ba9f-000c29328504 |
+---------------+--------------------------------------+

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46
• 47
• 48
• 49
• 50
• 51
• 52

2、从库报找不到对应的被更新的记录(Errno: 1032)

--首先在从库上删除leshami这条记录
(root@Slave)[tempdb]>delete from t1 where ename='leshami';

--接下来再主库尝试更新leshami这条记录
(root@Master)[tempdb]>update t1 set 
            -> blog='http://blog.csdn.net/robinson_0612' where ename='leshami';

Query OK, 1 row affected (0.02 sec)
Rows matched: 1  Changed: 1  Warnings: 0

-- 查看从库状态
(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
Last_SQL_Errno: 1032
Last_SQL_Error: Could not execute Update_rows event on table tempdb.t1; Can't find record in 't1',
                                Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND;
                            the event's master log node233-binlog.000004, end_log_pos 4769
Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-91
Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-90,
        78336cdc-8cfb-11e6-ba9f-000c29328504:1-2

-- 通过mysqlbinlog在主服务器上寻找报错的binglog日志文件及位置，找到对应的SQL语句，如下所示
-- update中的where后面的部分为更新前的数据，set部分为更新后的数据，因此可以将更新前的数据插入到从库
# mysqlbinlog --no-defaults -v -v --base64-output=DECODE-ROWS /data/node233-binlog.000004|grep -A '10' 4769
#161009 13:46:34 server id 233 end_log_pos 4769 CRC32 0xb60df74e Update_rows: table id 147 flags: STMT_END_F
### UPDATE `tempdb`.`t1`
### WHERE
###   @1=1 /* TINYINT meta=0 nullable=0 is_null=0 */
###   @2='leshami' /* VARSTRING(20) meta=20 nullable=1 is_null=0 */
###   @3='http://blog.csdn.net/leshami' /* VARSTRING(50) meta=50 nullable=1 is_null=0 */
### SET
###   @1=1 /* TINYINT meta=0 nullable=0 is_null=0 */
###   @2='leshami' /* VARSTRING(20) meta=20 nullable=1 is_null=0 */
###   @3='http://blog.csdn.net/robinson_0612' /* VARSTRING(50) meta=50 nullable=1 is_null=0 */
# at 4769
#161009 13:46:34 server id 233  end_log_pos 4800 CRC32 0xa9669811       Xid = 1749
COMMIT/*!*/;
SET @@SESSION.GTID_NEXT= 'AUTOMATIC' /* added by mysqlbinlog */ /*!*/;
DELIMITER ;
# End of log file
/*!50003 SET COMPLETION_TYPE=@OLD_COMPLETION_TYPE*/;
/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=0*/;   

(root@Slave)[tempdb]>select * from t1;
+----+-------+------------------------------------+
| id | ename | blog                               |
+----+-------+------------------------------------+
|  2 | robin | http://blog.csdn.net/robinson_0612 |
+----+-------+------------------------------------+

(root@Slave)[tempdb]>stop slave sql_thread;

(root@Slave)[tempdb]>insert into t1 values(1,'leshami','http://blog.csdn.net/leshami');

(root@Slave)[tempdb]>start slave sql_thread;

(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-91
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-91,
                               78336cdc-8cfb-11e6-ba9f-000c29328504:1-3
                Auto_Position: 1

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46
• 47
• 48
• 49
• 50
• 51
• 52
• 53
• 54
• 55
• 56
• 57
• 58
• 59
• 60
• 61
• 62

3、从库找不到对应的被删除的记录(Errno: 1032)

-- 如果是在主库上删除记录，而从库上找不到对应的记录，则可以直接跳过该事务
-- 下面我们首选在从库上删除一条记录
(root@Slave)[tempdb]>delete from t1 where ename='robin';

-- 接下来在主库上删除该记录
(root@Master)[tempdb]>delete from t1 where ename='robin';

-- 从库端提示无法找到对应的记录
(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
Last_SQL_Error: Could not execute Delete_rows event on table tempdb.t1; Can't find record in 't1',
                Error_code: 1032; handler error HA_ERR_KEY_NOT_FOUND; 
                the event's master log node233-binlog.000004, end_log_pos 5070
Last_SQL_Error_Timestamp: 161009 15:08:06
    Master_SSL_Crl: 
Master_SSL_Crlpath: 
Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-92
 Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-91,
                    78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
     Auto_Position: 1      

-- 下面通过注入空事务来跳过
(root@Slave)[tempdb]>stop slave sql_thread;

(root@Slave)[tempdb]>set gtid_next='1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:92';

(root@Slave)[tempdb]>begin;commit;

(root@Slave)[tempdb]>set gtid_next='AUTOMATIC';

(root@Slave)[tempdb]>start slave sql_thread;

(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-92
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-92,
                               78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
                Auto_Position: 1
         Replicate_Rewrite_DB: 
                 Channel_Name: 
           Master_TLS_Version: 

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41

4、延迟从修复主库意外truncate

-- 主库上面新增表及记录             
(root@Master)[tempdb]>create table t2 (id tinyint not null primary key, 
        -> ename varchar(20),blog varchar(50));

(root@Master)[tempdb]>insert into t2  
            -> values(1,'leshami','http://blog.csdn.net/leshami');

(root@Master)[tempdb]>insert into t2  
            -> values(2,'robin','http://blog.csdn.net/robinson_0612');

(root@Master)[tempdb]>select * from t2;
+----+---------+------------------------------------+
| id | ename   | blog                               |
+----+---------+------------------------------------+
|  1 | leshami | http://blog.csdn.net/leshami       |
|  2 | robin   | http://blog.csdn.net/robinson_0612 |
+----+---------+------------------------------------+

--先将从库配置为延迟从
(root@Slave)[tempdb]>stop slave sql_thread;
Query OK, 0 rows affected (0.01 sec)

(root@Slave)[tempdb]>CHANGE MASTER TO MASTER_DELAY = 300;
Query OK, 0 rows affected (0.00 sec)

(root@Slave)[tempdb]>start slave sql_thread;
Query OK, 0 rows affected (0.02 sec)

(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
                    SQL_Delay: 300  

root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-99
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-99,
                               78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
                Auto_Position: 1

--查看主库上的binglog gtid
(root@Master)[tempdb]>show master statusG
*************************** 1. row ***************************
             File: node233-binlog.000004
         Position: 6970
     Binlog_Do_DB: 
 Binlog_Ignore_DB: 
Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-99
1 row in set (0.00 sec)

--在主库上truncate t2
(root@Master)[tempdb]>truncate table t2;
Query OK, 0 rows affected (0.03 sec)

--再次查看主库上的binglog gtid，有99变成了100，这个100即是我们需要跳过的ID
(root@Master)[tempdb]>show master statusG
*************************** 1. row ***************************
             File: node233-binlog.000004
         Position: 7121
     Binlog_Do_DB: 
 Binlog_Ignore_DB: 
Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-100
1 row in set (0.00 sec)

--从库上跳过被意外truncate的事务
(root@Slave)[tempdb]>stop slave sql_thread;
Query OK, 0 rows affected (0.01 sec)

(root@Slave)[tempdb]>set gtid_next='1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:100';
Query OK, 0 rows affected (0.00 sec)

(root@Slave)[tempdb]>begin;commit;
Query OK, 0 rows affected (0.00 sec)

Query OK, 0 rows affected (0.01 sec)

(root@Slave)[tempdb]>set gtid_next='AUTOMATIC';
Query OK, 0 rows affected (0.00 sec)

(root@Slave)[tempdb]>start slave sql_thread;
Query OK, 0 rows affected (0.02 sec)

(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: Master
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: node233-binlog.000004
          Read_Master_Log_Pos: 7121
               Relay_Log_File: node245-relay-bin.000003
                Relay_Log_Pos: 2982
        Relay_Master_Log_File: node233-binlog.000004
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes
             ...........................         
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-100
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-100,
                                                             78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
                Auto_Position: 1

-- 很多时候我们并不知道表何时被truncate，因此可以从binlog日志得到其gtid
-- 如下所示，可以得到这串 SET @@SESSION.GTID_NEXT= '1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:100'
-- 100即为这个truncate对应的gtid的事务号
# mysqlbinlog --no-defaults -v -v --base64-output=DECODE-ROWS /data/node233-binlog.000004|grep -i 
> "truncate table t2" -A3 -B10  
###   @3='http://blog.csdn.net/robinson_0612' /* VARSTRING(50) meta=50 nullable=1 is_null=0 */
# at 6939
#161009 18:04:58 server id 233  end_log_pos 6970 CRC32 0x71c5121c     Xid = 1775
COMMIT/*!*/;
# at 6970
#161009 18:08:42 server id 233 end_log_pos 7035 CRC32 0x00ba9437 GTID last_committed=26 sequence_number=27
SET @@SESSION.GTID_NEXT= '1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:100'/*!*/;
# at 7035
#161009 18:08:42 server id 233 end_log_pos 7121 CRC32 0x5a8b9723 Query thread_id=26 exec_time=0 error_code=0
SET TIMESTAMP=1476007722/*!*/;
truncate table t2
/*!*/;
SET @@SESSION.GTID_NEXT= 'AUTOMATIC' /* added by mysqlbinlog */ /*!*/;
DELIMITER ;

• 1
• 2
• 3
• 4
• 5
• 6
• 7
• 8
• 9
• 10
• 11
• 12
• 13
• 14
• 15
• 16
• 17
• 18
• 19
• 20
• 21
• 22
• 23
• 24
• 25
• 26
• 27
• 28
• 29
• 30
• 31
• 32
• 33
• 34
• 35
• 36
• 37
• 38
• 39
• 40
• 41
• 42
• 43
• 44
• 45
• 46
• 47
• 48
• 49
• 50
• 51
• 52
• 53
• 54
• 55
• 56
• 57
• 58
• 59
• 60
• 61
• 62
• 63
• 64
• 65
• 66
• 67
• 68
• 69
• 70
• 71
• 72
• 73
• 74
• 75
• 76
• 77
• 78
• 79
• 80
• 81
• 82
• 83
• 84
• 85
• 86
• 87
• 88
• 89
• 90
• 91
• 92
• 93
• 94
• 95
• 96
• 97
• 98
• 99
• 100
• 101
• 102
• 103
• 104
• 105
• 106
• 107
• 108
• 109
• 110
• 111
• 112
• 113
• 114
• 115
• 116
• 117
• 118
• 119
• 120
• 121
• 122

5、主库binlog被purge的情形(Errno: 1236)

-- 首先停止从库，模拟从库被意外宕机
(root@Slave)[tempdb]>stop slave;
Query OK, 0 rows affected (0.08 sec)

--在主库上进行相应的操作
--此时主库上的gtid_purged为空
(root@Master)[tempdb]>show variables like '%gtid_purged%';
+---------------+-------+
| Variable_name | Value |
+---------------+-------+
| gtid_purged   |       |
+---------------+-------+

--查看主库binlog
(root@Master)[tempdb]>show binary logs;
+-----------------------+-----------+
| Log_name              | File_size |
+-----------------------+-----------+
| node233-binlog.000001 |   1362104 |
| node233-binlog.000002 |      1331 |
| node233-binlog.000003 |       217 |
| node233-binlog.000004 |      7121 |
+-----------------------+-----------+

(root@Master)[tempdb]>select * from t1;
+----+---------+------------------------------------+
| id | ename   | blog                               |
+----+---------+------------------------------------+
|  1 | leshami | http://blog.csdn.net/robinson_0612 |
|  2 | robin   | http://blog.csdn.net/leshami       |
+----+---------+------------------------------------+

--从主库删除记录
(root@Master)[tempdb]>delete from t1;

--切换日志
(root@Master)[tempdb]>flush logs;

--新增记录
(root@Master)[tempdb]>insert into t1 values(1,
    -> 'xuputi','http://blog.csdn.net/leshami');

(root@Master)[tempdb]>show binary logs;
+-----------------------+-----------+
| Log_name              | File_size |
+-----------------------+-----------+
| node233-binlog.000001 |   1362104 |
| node233-binlog.000002 |      1331 |
| node233-binlog.000003 |       217 |
| node233-binlog.000004 |      7513 |
| node233-binlog.000005 |       490 |
+-----------------------+-----------+

--清理binlog
(root@Master)[tempdb]>purge binary logs to 'node233-binlog.000005';
Query OK, 0 rows affected (0.01 sec)

(root@Master)[tempdb]>show binary logs;
+-----------------------+-----------+
| Log_name              | File_size |
+-----------------------+-----------+
| node233-binlog.000005 |       490 |
+-----------------------+-----------+

--此时可以看到相应的gtid_purged值
(root@Master)[tempdb]>show variables like '%gtid_purged%';
+---------------+--------------------------------------------+
| Variable_name | Value                                      |
+---------------+--------------------------------------------+
| gtid_purged   | 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-101 |
+---------------+--------------------------------------------+

--下面启动从库
(root@Slave)[tempdb]>start slave;
Query OK, 0 rows affected (0.00 sec)

--从库状态提示有日志被purged
(root@Slave)[tempdb]>show slave statusG
*************************** 1. row ***************************
               Slave_IO_State: 
                  Master_Host: Master
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: node233-binlog.000004
          Read_Master_Log_Pos: 7121
               Relay_Log_File: node245-relay-bin.000003
                Relay_Log_Pos: 3133
        Relay_Master_Log_File: node233-binlog.000004
             Slave_IO_Running: No
            Slave_SQL_Running: Yes
                    ...............
                Last_IO_Errno: 1236
                Last_IO_Error: Got fatal error 1236 from master when reading data from binary log:
                'The slave is connecting using CHANGE MASTER TO MASTER_AUTO_POSITION = 1, 
                 but the master has purged binary logs containing GTIDs that the slave requires.'
                       ..................
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-100
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-100,
                               78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
                Auto_Position: 1

-- 从库上gtid_purged参数，此时为75
(root@Slave)[tempdb]>show variables like '%gtid_purged%';
+---------------+-------------------------------------------+
| Variable_name | Value                                     |
+---------------+-------------------------------------------+
| gtid_purged   | 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-75 |
+---------------+-------------------------------------------+                

--停止从库
(root@Slave)[tempdb]>stop slave;
Query OK, 0 rows affected (0.01 sec)

--下面尝试使用gtid_purged进行跳过事务，，如下，提示仅仅当GLOBAL.GTID_EXECUTED为空才能被设置
(root@Slave)[tempdb]>set global gtid_purged = '1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-101';
ERROR 1840 (HY000): @@GLOBAL.GTID_PURGED can only be set when @@GLOBAL.GTID_EXECUTED is empty.

--如下查看，已经存在被执行的gtid，即gtid_executed肯定是不为空，且这些gtid记录在从库的binary log中
(root@Slave)[tempdb]>show global variables like '%gtid_executed%'G
*************************** 1. row ***************************
Variable_name: gtid_executed
        Value: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-100,
               78336cdc-8cfb-11e6-ba9f-000c29328504:1-4
*************************** 2. row ***************************
Variable_name: gtid_executed_compression_period
        Value: 1000

--下面我们在从库上reset master，即清空从库binlog
(root@Slave)[tempdb]>reset master;
Query OK, 0 rows affected (0.05 sec)

--再次查看gtid_executed已经为空值
(root@Slave)[tempdb]>show global variables like '%gtid_executed%'G
*************************** 1. row ***************************
Variable_name: gtid_executed
        Value: 
*************************** 2. row ***************************
Variable_name: gtid_executed_compression_period
        Value: 1000

--此时再次设置gtid_purged的值
(root@Slave)[tempdb]>set global gtid_purged = '1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-101';
Query OK, 0 rows affected (0.01 sec)

--启动从库
(root@Slave)[tempdb]>start slave;
Query OK, 0 rows affected (0.03 sec)

--提示有重复记录，如下所示
--是由于我们在从库停止期间delete这个事务没有被从库的relay log接受到
--其次主从的binlog又被purged，而且从库启动后，执行了gtid_purged，因此主库上新增的记录在从库上提示主键重复
(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: Master
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: node233-binlog.000005
          Read_Master_Log_Pos: 490
               Relay_Log_File: node245-relay-bin.000004
                Relay_Log_Pos: 417
        Relay_Master_Log_File: node233-binlog.000005
             Slave_IO_Running: Yes
            Slave_SQL_Running: No
                ................
               Last_SQL_Error: Could not execute Write_rows event on table tempdb.t1; 
 Duplicate entry '1' for key 'PRIMARY', Error_code: 1062;
 handler error HA_ERR_FOUND_DUPP_KEY; the event's master log node233-binlog.000005, end_log_pos 459
           Retrieved_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:76-100:102
            Executed_Gtid_Set: 1b64c25d-8d2b-11e6-9ac0-000c29b82d0d:1-101
                Auto_Position: 1

--在从库上删除id为1的记录
(root@Slave)[tempdb]>delete from t1 where id=1;
Query OK, 1 row affected (0.05 sec)

--启动从库的sql_thread线程
(root@Slave)[tempdb]>start slave sql_thread;
Query OK, 0 rows affected (0.02 sec)

--再次查看正常
(root@Slave)[tempdb]>show slave status G
*************************** 1. row ***************************
               Slave_IO_State: Waiting for master to send event
                  Master_Host: Master
                  Master_User: repl
                  Master_Port: 3306
                Connect_Retry: 60
              Master_Log_File: node233-binlog.000005
          Read_Master_Log_Pos: 490
               Relay_Log_File: node245-relay-bin.000004
                Relay_Log_Pos: 713
        Relay_Master_Log_File: node233-binlog.000005
             Slave_IO_Running: Yes
            Slave_SQL_Running: Yes

--上面的这个示例，主要是演示我们使用gtid_purged方式来达到跳过事务的目的
--事实上，主从的数据已经不一致了，应根据实际的需要考虑是否进行相应的修复

五、小结

1、GTID是全局事务ID，简化了主从架构的部署使得从库不再需要关心log_file和log_pos 
2、由于事务ID的唯一性，使得将其他从库的GTID应用到其它从库成为可能，即提供了便利的failover 
3、GTID是连续的，非空洞性的，因此，对于冲突的情形，需要注入空的事务来实现 
4、可以通过配置延迟从来避免主库上意外的删除对象导致的人为错误