首页 > 代码库 > MySQL Binlog Mixed模式记录成Row格式
MySQL Binlog Mixed模式记录成Row格式
概念:
binlog format有三种形式:Statement、Mixed、Row,具体的信息可以自行到网上搜查。
分析(本文碰到的案例):
查看MySQL binlog format
dba@192.168.111.4 : dba_test 02:33:39>show variables like ‘binlog_format%‘; +---------------+-------+| Variable_name | Value |+---------------+-------+| binlog_format | MIXED |+---------------+-------+
测试语句:
dba@192.168.111.4 : dba_test 02:24:14>create table tmp_test(id int,name varchar(64),age int,primary key(id)) engine = innodb;
Query OK, 0 rows affected (0.05 sec)
dba@192.168.111.4 : dba_test 02:24:23>insert into tmp_test values(1,‘aaa‘,11);
Query OK, 1 row affected (0.02 sec)
dba@192.168.111.4 : dba_test 02:25:17>insert into tmp_test values(2,‘bbb‘,22);
Query OK, 1 row affected (0.02 sec)
dba@192.168.111.4 : dba_test 02:25:23>insert into tmp_test values(3,‘ccc‘,33);
Query OK, 1 row affected (0.01 sec)
dba@192.168.111.4 : dba_test 02:25:28>insert into tmp_test values(4,‘ddd‘,44);
Query OK, 1 row affected (0.01 sec)
dba@192.168.111.4 : dba_test 02:25:34>insert into tmp_test values(5,‘eee‘,55);
Query OK, 1 row affected (0.01 sec)
dba@192.168.111.4 : dba_test 02:25:42>select * from tmp_test;+----+------+------+| id | name | age |+----+------+------+| 1 | aaa | 11 || 2 | bbb | 22 || 3 | ccc | 33 || 4 | ddd | 44 || 5 | eee | 55 |+----+------+------+5 rows in set (0.01 sec)
dba@192.168.111.4 : dba_test 02:25:50>create table tmp_test_bak(id int,name varchar(64),age int,primary key(id)) engine = innodb;
Query OK, 0 rows affected (0.03 sec)
dba@192.168.111.4 : dba_test 02:26:31>insert into tmp_test_bak select * from tmp_test; ###记录成了Row模式Query OK, 5 rows affected (0.03 sec)
Records: 5 Duplicates: 0 Warnings: 0
Binlog 记录图:
问题来了,我想要出来的binlog format是Statement,而不是Row。而一条insert into tb select * from ta的简单语句在Mixed模式下记录了Row模式的binlog。原因是什么?
首先确实在一些特定的情况下,Mixed会被转换成Row模式:
. 当 DML 语句更新一个 NDB 表时;
. 当函数中包含 UUID() 时;
. 2 个及以上包含 AUTO_INCREMENT 字段的表被更新时;
. 执行 INSERT DELAYED 语句时;
. 用 UDF 时;
. 视图中必须要求运用 row 时,例如建立视图时使用了 UUID() 函数;
上面来自网络,有兴趣的可以自己测试测试。而对于本文中的sql,符合不了上面的条件,但binlog也记录成了Row格式。所以还是很奇怪为什么binlog格式被转换了,日常工作的时候有遇到过执行一条sql,会报一个warning:
Warning: Unsafe statement written to the binary log using statement format since BINLOG_FORMAT = STATEMENT
难道因为这个导致转换的?因为上面的SQL可以重现,没有报warning,所以这个情况排除。根据经验想到了一个参数:innodb_locks_unsafe_for_binlog,看到里面讲到事务隔离级别,那就看看隔离级别的情况:
dba@192.168.111.4 : dba_test 05:46:56>select @@global.tx_isolation;+-----------------------+| @@global.tx_isolation |+-----------------------+| READ-COMMITTED |+-----------------------+1 row in set (0.01 sec)
dba@192.168.111.4 : dba_test 06:36:45>select @@session.tx_isolation;+------------------------+| @@session.tx_isolation |+------------------------+| READ-COMMITTED |+------------------------+1 row in set (0.01 sec)
看到隔离级别是提交读,即不可重复读。把事务隔离级别设置成默认的 REPEATABLE READ:
dba@192.168.111.4 : dba_test 06:41:02>set session transaction isolation level REPEATABLE READ;
Query OK, 0 rows affected (0.14 sec)
dba@192.168.111.4 : dba_test 06:41:42>select @@session.tx_isolation;+------------------------+| @@session.tx_isolation |+------------------------+| REPEATABLE-READ |+------------------------+1 row in set (0.00 sec)
再执行测试里的SQL,发现这时候Mixed的binlog记录了Statement格式,正常了,符合预期了。难道就是这个事务隔离级别的问题引起的?在手册里发现了这句:
NoteIn MySQL 5.7, when READ COMMITTED isolation level is used, or the deprecated innodb_locks_unsafe_for_binlog system variable is enabled,
there is no InnoDB gap locking except for foreign-key constraint checking and duplicate-key checking. Also, record locks for nonmatching
rows are released after MySQL has evaluated the WHERE condition.
If you use READ COMMITTED or enable innodb_locks_unsafe_for_binlog, you must use row-based binary logging.
展开可以看例子:
View Code
经过测试,在5.1、5.5、5.6都有这个情况,可能这个本身就不是问题。:)
MySQL Binlog Mixed模式记录成Row格式
声明:以上内容来自用户投稿及互联网公开渠道收集整理发布,本网站不拥有所有权,未作人工编辑处理,也不承担相关法律责任,若内容有误或涉及侵权可进行投诉: 投诉/举报 工作人员会在5个工作日内联系你,一经查实,本站将立刻删除涉嫌侵权内容。