在数据库日常维护中,开发人员是最让人头痛的,很多时候都会由于SQL语句 写的有问题导致服务器出问题,导致资源耗尽。最危险的操作就是在做DML操作的时候忘加where条件,导致全表更新,这是作为运维或者DBA的我们改如 何处理呢?下面我分别针对update和delete操作忘加where条件导致全表更新的处理方法。
一. update 忘加where条件误操作恢复数据(binglog格式必须是ROW)
1.创建测试用的数据表
create table t1 (
id int unsigned not null auto_increment,
name char(20) not null,
sex enum('f','m') not null default 'm',
address varchar(30) not null,
primary key(id)
);
2.插入测试数据
insert into t1 (name,sex,address)values('daiiy','m','guangzhou');
insert into t1 (name,sex,address)values('tom','f','shanghai');
insert into t1 (name,sex,address)values('liany','m','beijing');
insert into t1 (name,sex,address)values('lilu','m','zhuhai');
3.现在需要将id等于2的用户的地址改为zhuhai,update时没有添加where条件
mysql> select * from t1;
+----+-------+-----+-----------+
| id | name | sex | address |
+----+-------+-----+-----------+
| 1 | daiiy | m | guangzhou |
| 2 | tom | f | shanghai |
| 3 | liany | m | beijing |
| 4 | lilu | m | zhuhai |
+----+-------+-----+-----------+
4 rows in set (0.00 sec)
mysql> update t1 set address='zhuhai';
mysql> select * from t1;
+----+-------+-----+---------+
| id | name | sex | address |
+----+-------+-----+---------+
| 1 | daiiy | m | zhuhai |
| 2 | tom | f | zhuhai |
| 3 | liany | m | zhuhai |
| 4 | lilu | m | zhuhai |
+----+-------+-----+---------+
4 rows in set (0.00 sec)
4.开始恢复,在线上的话,应该比较复杂,要先进行锁表,以免数据再次被污染。(锁表,查看正在写哪个二进制日志)
mysql> lock tables t1 read ;
mysql> show master status;
+------------------+----------+--------------+------------------+-------------------+
| File | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |
+------------------+----------+--------------+------------------+-------------------+
| mysql-bin.000005 | 2309 | | | |
+------------------+----------+--------------+------------------+-------------------+
1 row in set (0.00 sec)
5.分析二进制日志,并且在其中找到相关记录,在更新时是address='zhuhai',我们可以在日志中过滤出来。
[root@node1 binlog]# mysqlbinlog --no-defaults -v -v --base64-output=DECODE-ROWS mysql-bin.000005 | grep -B 15 -i --color 'zhuhai'
COMMIT/*!*/;
# at 1790
#160302 14:43:37 server id 1 end_log_pos 1862 CRC32 0xe5d422f5 Query thread_id=44 exec_time=0 error_code=0
SET TIMESTAMP=1456901017/*!*/;
BEGIN
/*!*/;
# at 1862
#160302 14:43:37 server id 1 end_log_pos 1916 CRC32 0x39917b1c Table_map: `arun`.`t1` mapped to number 156
# at 1916
#160302 14:43:37 server id 1 end_log_pos 1969 CRC32 0x31a19175 Write_rows: table id 156 flags: STMT_END_F
### INSERT INTO `arun`.`t1`
### SET
### @1=4 /* INT meta=0 nullable=0 is_null=0 */
### @2='lilu' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
--
/*!*/;
# at 2072
#160302 14:50:00 server id 1 end_log_pos 2126 CRC32 0xd2434753 Table_map: `arun`.`t1` mapped to number 156
# at 2126
#160302 14:50:00 server id 1 end_log_pos 2278 CRC32 0xfb36ef03 Update_rows: table id 156 flags: STMT_END_F
### UPDATE `arun`.`t1`
### WHERE
### @1=1 /* INT meta=0 nullable=0 is_null=0 */
### @2='daiiy' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='guangzhou' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=1 /* INT meta=0 nullable=0 is_null=0 */
### @2='daiiy' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### UPDATE `arun`.`t1`
### WHERE
### @1=2 /* INT meta=0 nullable=0 is_null=0 */
### @2='tom' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=1 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='shanghai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=2 /* INT meta=0 nullable=0 is_null=0 */
### @2='tom' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=1 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### UPDATE `arun`.`t1`
### WHERE
### @1=3 /* INT meta=0 nullable=0 is_null=0 */
### @2='liany' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='beijing' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=3 /* INT meta=0 nullable=0 is_null=0 */
### @2='liany' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
可以看见里面记录了每一行的变化,这也是binglog格式要一定是row才行的原因。其中@1,@2,@3,@4,分别对应表中 id,name,sex,address字段。相信大家看到这里有点明白了吧,对,没错,你猜到了,我们将相关记录转换为sql语句,重新导入数据库。
6.处理分析处理的二进制日志
[root@node1 binlog]# mysqlbinlog --no-defaults -v -v --base64-output=DECODE-ROWS mysql-bin.000005 | sed -n '/# at 2126/,/COMMIT/p' > t1.txt
[root@node1 binlog]# cat t1.txt
# at 2126
#160302 14:50:00 server id 1 end_log_pos 2278 CRC32 0xfb36ef03 Update_rows: table id 156 flags: STMT_END_F
### UPDATE `arun`.`t1`
### WHERE
### @1=1 /* INT meta=0 nullable=0 is_null=0 */
### @2='daiiy' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='guangzhou' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=1 /* INT meta=0 nullable=0 is_null=0 */
### @2='daiiy' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### UPDATE `arun`.`t1`
### WHERE
### @1=2 /* INT meta=0 nullable=0 is_null=0 */
### @2='tom' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=1 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='shanghai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=2 /* INT meta=0 nullable=0 is_null=0 */
### @2='tom' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=1 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### UPDATE `arun`.`t1`
### WHERE
### @1=3 /* INT meta=0 nullable=0 is_null=0 */
### @2='liany' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='beijing' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### SET
### @1=3 /* INT meta=0 nullable=0 is_null=0 */
### @2='liany' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
# at 2278
#160302 14:50:00 server id 1 end_log_pos 2309 CRC32 0x58165775 Xid = 1466
COMMIT/*!*/;
这里sed有点复杂,需要童鞋们好好自己研究研究,这里我就不多说了。
[root@node1 binlog]# sed '/WHERE/{:a;N;/SET/!ba;s/\([^\n]*\)\n\(.*\)\n\(.*\)/\3\n\2\n\1/}' t1.txt |sed -r '/WHERE/{:a;N;/@4/!ba;s/### @2.*//g}' | sed 's/### //g;s/\/\*.*/,/g' | sed '/WHERE/{:a;N;/@1/!ba;s/,/;/g};s/#.*//g;s/COMMIT,//g' | sed '/^$/d' > recover.sql
[root@node1 binlog]# cat recover.sql
UPDATE `arun`.`t1`
SET
@1=1 ,
@2='daiiy' ,
@3=2 ,
@4='guangzhou' ,
WHERE
@1=1 ;
UPDATE `arun`.`t1`
SET
@1=2 ,
@2='tom' ,
@3=1 ,
@4='shanghai' ,
WHERE
@1=2 ;
UPDATE `arun`.`t1`
SET
@1=3 ,
@2='liany' ,
@3=2 ,
@4='beijing' ,
WHERE
@1=3 ;
将文件中的@1,@2,@3,@4替换为t1表中id,name,sex,address字段,并删除最后字段的","号
[root@node1 binlog]# sed -i 's/@1/id/g;s/@2/name/g;s/@3/sex/g;s/@4/address/g' recover.sql
[root@node1 binlog]# sed -i -r 's/(address=.*),/\1/g' recover.sql
[root@node1 binlog]# cat recover.sql
UPDATE `arun`.`t1`
SET
id=1 ,
name='daiiy' ,
sex=2 ,
address='guangzhou'
WHERE
id=1 ;
UPDATE `arun`.`t1`
SET
id=2 ,
name='tom' ,
sex=1 ,
address='shanghai'
WHERE
id=2 ;
UPDATE `arun`.`t1`
SET
id=3 ,
name='liany' ,
sex=2 ,
address='beijing'
WHERE
id=3 ;
7.到这里日志就处理好了,现在导入即可(导入数据后,解锁表);
mysql> UNLOCK TABLES;
mysql> source /data/log/mysql/binlog/recover.sql;
Query OK, 1 row affected (0.03 sec)
Rows matched: 1 Changed: 1 Warnings: 0
Query OK, 1 row affected (0.01 sec)
Rows matched: 1 Changed: 1 Warnings: 0
Query OK, 1 row affected (0.03 sec)
Rows matched: 1 Changed: 1 Warnings: 0
可以看见数据已经完全恢复,这种方法的优点是快速,方便。
二. delete 忘加where条件误删除恢复(binglog格式必须是ROW)
其实这和update忘加条件差不多,不过这处理更简单,这里就用上面那张表做测试吧
1.模拟误删除数据
mysql> select * from t1;
+----+-------+-----+-----------+
| id | name | sex | address |
+----+-------+-----+-----------+
| 1 | daiiy | m | guangzhou |
| 2 | tom | f | shanghai |
| 3 | liany | m | beijing |
| 4 | lilu | m | zhuhai |
+----+-------+-----+-----------+
4 rows in set (0.00 sec)
mysql> delete from t1;
Query OK, 4 rows affected (0.02 sec)
mysql> select * from t1;
Empty set (0.00 sec)
2.在binglog中去查找相关记录
[root@node1 binlog]# mysqlbinlog --no-defaults -v -v --base64-output=decode-rows mysql-bin.000005 |sed -n '/### DELETE FROM `arun`.`t1`/,/COMMIT/p'
### DELETE FROM `arun`.`t1`
### WHERE
### @1=1 /* INT meta=0 nullable=0 is_null=0 */
### @2='daiiy' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='guangzhou' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### DELETE FROM `arun`.`t1`
### WHERE
### @1=2 /* INT meta=0 nullable=0 is_null=0 */
### @2='tom' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=1 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='shanghai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### DELETE FROM `arun`.`t1`
### WHERE
### @1=3 /* INT meta=0 nullable=0 is_null=0 */
### @2='liany' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='beijing' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
### DELETE FROM `arun`.`t1`
### WHERE
### @1=4 /* INT meta=0 nullable=0 is_null=0 */
### @2='lilu' /* STRING(60) meta=65084 nullable=0 is_null=0 */
### @3=2 /* ENUM(1 byte) meta=63233 nullable=0 is_null=0 */
### @4='zhuhai' /* VARSTRING(90) meta=90 nullable=0 is_null=0 */
# at 3244
#160302 16:33:31 server id 1 end_log_pos 3275 CRC32 0xed9c2a39 Xid = 1484
COMMIT/*!*/;
[root@node1 binlog]# mysqlbinlog --no-defaults -v -v --base64-output=decode-rows mysql-bin.000005 |sed -n '/### DELETE FROM `arun`.`t1`/,/COMMIT/p' > delete.txt
3.将记录转换为SQL语句
[root@node1 binlog]# cat delete.txt | sed -n '/###/p' | sed 's/### //g;s/\/\*.*/,/g;s/DELETE FROM/INSERT INTO/g;s/WHERE/SELECT/g;' | sed -r 's/(@4.*),/\1;/g' | sed 's/@[1-9]=//g' > t1.sql
[root@node1 binlog]# cat t1.sql
INSERT INTO `arun`.`t1`
SELECT
1 ,
'daiiy' ,
2 ,
'guangzhou' ;
INSERT INTO `arun`.`t1`
SELECT
2 ,
'tom' ,
1 ,
'shanghai' ;
INSERT INTO `arun`.`t1`
SELECT
3 ,
'liany' ,
2 ,
'beijing' ;
INSERT INTO `arun`.`t1`
SELECT
4 ,
'lilu' ,
2 ,
'zhuhai' ;
4.导入数据,验证数据完整性
mysql> source /data/log/mysql/binlog/t1.sql;
Query OK, 1 row affected (0.02 sec)
Records: 1 Duplicates: 0 Warnings: 0
Query OK, 1 row affected (0.01 sec)
Records: 1 Duplicates: 0 Warnings: 0
Query OK, 1 row affected (0.01 sec)
Records: 1 Duplicates: 0 Warnings: 0
Query OK, 1 row affected (0.01 sec)
Records: 1 Duplicates: 0 Warnings: 0
mysql> select * from t1;
+----+-------+-----+-----------+
| id | name | sex | address |
+----+-------+-----+-----------+
| 1 | daiiy | m | guangzhou |
| 2 | tom | f | shanghai |
| 3 | liany | m | beijing |
| 4 | lilu | m | zhuhai |
+----+-------+-----+-----------+
4 rows in set (0.00 sec)
到这里数据就完整回来了。将binglog格式设置为row有利有弊,好处是记录了每一行的实际变化,在主从复制时也不容易出问题。但是 由于记录每行的变化,会占用大量磁盘,主从复制时带宽占用会有所消耗。到底是使用row还是mixed,需要在实际工作中自己去衡量,但从整体上来 说,binglog的格式设置为row,都是不二的选择。
总结:
所以在数据库操作的过程中我们需要格外小心,当然开发那边我们需要做好权限的控制,不过有一个参数可以解决我们的问题,让我们不用担心类似的问题发生:
在[mysql]段落开启这个参数:
safe-updates
这样当我们在做DML操作时忘记加where条件时,mysqld服务器是不会执行操作的:
还是刚才的表,执行delete操作==>
mysql> delete from t1;
ERROR 1175 (HY000): You are using safe update mode and you tried to update a table without a WHERE that uses a KEY column
ERROR 1175 (HY000): You are using safe update mode and you tried to update a table without a WHERE that uses a KEY column
转自
MySQL 误操作后数据恢复(update,delete忘加where条件) - arun_yh - 博客园 http://www.cnblogs.com/itcomputer/articles/5234893.html