MySQL replace into 错误案例
背景
* MySQL5.7
* ROW模式
* 表结构
CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
`col_3` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8
错误场景一
其他字段value莫名其妙的没了
- step1 初始化记录
mater:lc> REPLACE INTO test (col_1,col_2,col_3) values('a','a','a');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master:lc> REPLACE INTO test (col_1,col_2,col_3) values('b','b','b');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master:lc> REPLACE INTO test (col_1,col_2,col_3) values('c','c','c');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master > show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
`col_3` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8 |
mater > select * from test;
+----+-------+-------+-------+
| id | col_1 | col_2 | col_3 |
+----+-------+-------+-------+
| 1 | a | a | a |
| 2 | b | b | b |
| 3 | c | c | c |
+----+-------+-------+-------+
3 rows in set (0.00 sec)
- step2 构造错误场景
master:lc> replace into test(col_1,col_2) values('c','cc');
Query OK, 2 rows affected (0.00 sec)
dba:lc> select * from test;
+----+-------+-------+-------+
| id | col_1 | col_2 | col_3 |
+----+-------+-------+-------+
| 1 | a | a | a |
| 2 | b | b | b |
| 4 | c | cc | NULL |
+----+-------+-------+-------+
3 rows in set (0.00 sec)
- 总结
- col_3 的值，从原来的c，变成了NULL，天呐，数据不见了。 id 也变了。
- 用户原本的需求，应该是如果col_1='c' 存在，那么就改变col_2='cc'，其余的记录保持不变，结果id,col_3都变化了
- 解决方案就是：将replace into 改成 INSERT INTO … ON DUPLICATE KEY UPDATE
但是你以为这样就完美的解决了吗？ 马上就会带来另外一场灾难，请看下面的错误场景
错误场景二
ERROR 1062 (23000): Duplicate entry 'x' for key 'PRIMARY'
- step1 初始化记录
mater:lc> REPLACE INTO test (col_1,col_2) values('a','a');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master:lc> REPLACE INTO test (col_1,col_2) values('b','b');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master:lc> REPLACE INTO test (col_1,col_2) values('c','c');
Query OK, 1 row affected (0.00 sec) --注意，这里是影响了1条记录
master > show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8 |
slave > show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8 |- step2 构造错误场景
* master
mater:lc> REPLACE INTO test (col_1,col_2) values('c','cc');
Query OK, 2 rows affected (0.00 sec) --注意，这里是影响了两条记录
mater:lc> show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=5 DEFAULT CHARSET=utf8 |
master:lc> select * from test
+----+-------+-------+
| id | col_1 | col_2 |
+----+-------+-------+
| 1 | a | a |
| 2 | b | b |
| 4 | c | cc |
+----+-------+-------+
3 rows in set (0.00 sec)
* slave
slave:lc> show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8 |
slave:lc> select * from test
+----+-------+-------+
| id | col_1 | col_2 |
+----+-------+-------+
| 1 | a | a |
| 2 | b | b |
| 4 | c | cc |
+----+-------+-------+
3 rows in set (0.00 sec)
- step3 错误案例产生
* 假设有一天，master 挂了， 由slave 提升为 new mater
原slave:lc> show create table test
| test | CREATE TABLE `test` (
`id` int(10) unsigned NOT NULL AUTO_INCREMENT,
`col_1` varchar(100) DEFAULT NULL,
`col_2` varchar(100) DEFAULT NULL,
PRIMARY KEY (`id`),
UNIQUE KEY `col_1` (`col_1`)
) ENGINE=InnoDB AUTO_INCREMENT=4 DEFAULT CHARSET=utf8 |
原slave:lc> select * from test
+----+-------+-------+
| id | col_1 | col_2 |
+----+-------+-------+
| 1 | a | a |
| 2 | b | b |
| 4 | c | cc |
+----+-------+-------+
3 rows in set (0.00 sec)
===注意==
root:lc> REPLACE INTO test (col_1,col_2) values('d','d');
ERROR 1062 (23000): Duplicate entry '4' for key 'PRIMARY'
- 总结
* Row 模式，主从情况下，replace into 和 INSERT INTO … ON DUPLICATE KEY UPDATE 都会导致以上问题的发生
* 解决方案： 最后可以通过alter table auto_increment值解决，但是这样已经造成mater的表很长时间没有写入了。。。最后总结
- replace with unique key
1. 禁止 replace into (错误一，错误二 都会发生)
2. 禁止 INSERT INTO … ON DUPLICATE KEY UPDATE （错误二 会发生）- replace with primary key
1. 禁止 replace into (会发生错误场景一的案例，丢失部分字段数据)
2. 可以使用INSERT INTO … ON DUPLICATE KEY UPDATE 代替 replace into