原创：转载请说明出处谢谢！

上接

http://blog.itpub.net/7728585/viewspace-2133188/ 解析MYSQL BINLOG 二进制格式(1)--准备工作

http://blog.itpub.net/7728585/viewspace-2133189/ 解析MYSQL BINLOG 二进制格式(2)--FORMAT_DESCRIPTION_EVENT

http://blog.itpub.net/7728585/viewspace-2133321/ 解析MYSQL BINLOG 二进制格式(3)--QUERY_EVENT

http://blog.itpub.net/7728585/viewspace-2133429/ 解析MYSQL BINLOG 二进制格式(4)--TABLE_MAP_EVENT

http://blog.itpub.net/7728585/viewspace-2133463/ 解析MYSQL BINLOG 二进制格式(5)--WRITE_ROW_EVENT

http://blog.itpub.net/7728585/viewspace-2133469/ 解析MYSQL BINLOG 二进制格式(6)--UPDATE_ROW_EVENT/DELETE_ROW_EVENT

http://blog.itpub.net/7728585/viewspace-2133502/ 解析MYSQL BINLOG 二进制格式(7)--Xid_log_event/XID_EVENT

http://blog.itpub.net/7728585/viewspace-2133506/ 解析MYSQL BINLOG二进制格式(8)--GTID_LOG_EVENT/ANONYMOUS_GTID_LOG_EVENT及其他

1、工具原理和作用：

本工具是我通过学习binlog event写的一个能够直接读取binlog二进制文件的工具，使用语言C语言，编译器gcc，

他是通过读取各个主要字节的输出。因为mysqlbinlog输出有时候过于繁杂，infobin工具可以帮助初次定位，结合

mysqlbinlog进行分析,同时写这个工具也是为了验证学习event的正确性。

最后生成一个总的汇总，汇总信息是我最近更新的加入了：

如下部分：

-------------Total now--------------

Trx total[counts]:7665

Event total[counts]:183513

Max trx event size:8193(bytes) Pos:306929600[0X124B5FC0]

Avg binlog size(/sec):609537.375(bytes)[595.251(kb)]

Avg binlog size(/min):36572244.000(bytes)[35715.082(kb)]

--Piece view:

(1)Time:1487126981-1487127083(102(s)) piece:237110047(bytes)[231552.781(kb)]

(2)Time:1487127083-1487127083(0(s)) piece:237110047(bytes)[231552.781(kb)]

(3)Time:1487127083-1487127083(0(s)) piece:237110047(bytes)[231552.781(kb)]

(4)Time:1487127083-1487128812(1729(s)) piece:237110047(bytes)[231552.781(kb)]

(5)Time:1487128812-1487128926(114(s)) piece:237110047(bytes)[231552.781(kb)]

--Large than 1000(bytes) trx:

(1)Trx_size:815352746(bytes)[796242.938(kb)] trx_begin_p:1174606[0X11EC4E] trx_end_p:816527352[0X30AB37F8]

(2)Trx_size:3108(bytes)[3.035(kb)] trx_begin_p:817520533[0X30BA5F95] trx_end_p:817523641[0X30BA6BB9]

(3)Trx_size:3116(bytes)[3.043(kb)] trx_begin_p:817523706[0X30BA6BFA] trx_end_p:817526822[0X30BA7826]

(4)Trx_size:5910(bytes)[5.771(kb)] trx_begin_p:817526887[0X30BA7867] trx_end_p:817532797[0X30BA8F7D]

(5)Trx_size:11538(bytes)[11.268(kb)] trx_begin_p:817532862[0X30BA8FBE] trx_end_p:817544400[0X30BABCD0]

(6)Trx_size:22754(bytes)[22.221(kb)] trx_begin_p:817544465[0X30BABD11] trx_end_p:817567219[0X30BB15F3]

2、限制：

--只能使用在Little_endian上，编译是在LINUX gcc编译的

--load data infile event是没有检测的

--不能读取出row event的语句，因为没有写那么复杂

--可以读取出statement格式的语句，但是为了简洁做了35字节的截断，方便输出

这些东西在mysqlbinlog解析中都有。

3、使用方法和获取

获取可以通过百度云盘

http://pan.baidu.com/s/1jHIWUN0

获得，编译的只有LINUX64版本的

使用帮助如下：

[root@testmy data]# ./infobin

USAGE ERROR!

[Author]: gaopeng [QQ]:22389860 [blog]:http://blog.itpub.net/7728585/

--USAGE:./infobin [binlogfile] [piece] [bigtrxsize]

[binlogfile]:binlog file!

[piece]:how many piece will split,is a Highly balanced histogram,

find which time generate biggest binlog.(must:piece<2000000)

[bigtrxsize](bytes):larger than this size trx will view.(must:trx>256(bytes))

如：./infobin test.000200 5 257

5：代表是piece，这是一个高度均衡直方图的桶数量，后面会看到输出解释

257：是大约257 bytes的事物输出到最后

4、输出解析

我们以MYSQL 5.7 row format格式为例，我做的语句是

mysql> select * from testnull2;

+------+---------+---------+

| id   | name1   | name2   |

+------+---------+---------+

|    1 | gaopeng | gaopeng |

|    1 | gaopeng | gaopeng |

+------+---------+---------+

2 rows in set (0.00 sec)

mysql> create table testtool like testnull2;

Query OK, 0 rows affected (0.26 sec)

mysql> insert into testtool select * from testnull2;

Query OK, 2 rows affected (0.03 sec)

Records: 2  Duplicates: 0  Warnings: 0

使用./infobin test.000200 得到的输出如下：

[root@testmy data]# ./infobin test.000200 5 512

Check is Little_endian

Author: gaopeng QQ:22389860 Mail: gaopp_200217@163.com

Waring: This tool only Little_endian platform!

Little_endian check ok!!!

-------------Now begin--------------

Check Mysql Version is:5.7.13-log

Check Mysql binlog format ver is:V4

Warning:Check This binlog is not closed!

Check This binlog total size:664(bytes)

Note:load data infile not check!

------------Detail now--------------

>Format description log Event:Pos:4(0X4) N_pos:123(0X7b) Time:1487035963 Event_size:119(bytes)

>Previous gtid Event:Pos:123(0X7b) N_pos:194(0Xc2) Time:1487035963 Event_size:71(bytes)

>Gtid Event:Pos:194(0Xc2) N_pos:259(0X103) Time:1487035999 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100463

-->Query Event:Pos:259(0X103) N_Pos:362(0X16a) Time:1487035999 Event_size:103(bytes)

Exe_time:0  Use_db:test Statment(35b-trun):create table testtool like testnull Gno:1100463

>Gtid Event:Pos:362(0X16a) N_pos:427(0X1ab) Time:1487036014 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100464

-->Query Event:Pos:427(0X1ab) N_Pos:499(0X1f3) Time:1487036014 Event_size:72(bytes)

Exe_time:0  Use_db:test Statment(35b-trun):BEGIN /*!Trx begin!*/ Gno:1100464

---->Map Event:Pos499(0X1f3) N_pos:556(0X22c) Time:1487036014 Event_size:57(bytes)

TABLE_ID:346 DB_NAME:test TABLE_NAME:testtool Gno:1100464

------>Insert Event:Pos:556(0X22c) N_pos:633(0X279) Time:1487036014 Event_size:77(bytes)

Dml on table: test.testtool  table_id:346 Gno:1100464

>Xid Event:Pos:633(0X279) N_Pos:664(0X298) Time:1487036014 Event_size:31(bytes)

COMMIT; /*!Trx end*/ Gno:1100464

-------------Total now--------------

Trx total[counts]:1

Event total[counts]:10

Max trx event size:119(bytes) Pos:4[0X4]

Avg binlog size(/sec):5.007(bytes)[0.005(kb)]

Avg binlog size(/min):300.426(bytes)[0.293(kb)]

--Piece view:

(1)Time:1487035963-1487035999(36(s)) piece:141(bytes)[0.138(kb)]

(2)Time:1487035999-1487036014(15(s)) piece:141(bytes)[0.138(kb)]

(3)Time:1487036014-1487036014(0(s)) piece:141(bytes)[0.138(kb)]

(4)Time:1487036014-1487036014(0(s)) piece:141(bytes)[0.138(kb)]

(5)Time:1487036014-1487036104(90(s)) piece:141(bytes)[0.138(kb)]

--Large than 257(bytes) trx:

No trx find!

now begin部分：

一目了然需要说明一点Warning:Check This binlog is not closed!说明这个binlog是当前正在使用binlog

这个和mysqlbinlog

# at 4

#170214  9:32:43 server id 93157  end_log_pos 123 CRC32 0xc5518ba9      Start: binlog v 4, server v 5.7.13-log created 170214  9:32:43

# Warning: this binlog is either in use or was not closed properly.

中是一个意思

Detail now部分：

*1、event都以>开始，但是一个事物的event我使用--> ----> ------->来进行区别化更加利于阅读，如果

仔细研究过event这些event一定不会陌生

*2、

Pos:当前event位置

N_pos:下一个event位置，

Gtid: 当然就是GTID如果是匿名事物就是ANONYMOUS 其GTID为0

Time:新纪元时间以来的秒数 可以用LINUX命令换算 如：date -s "@1487035999"

Event_size：这个event有多大

Gno:gtid的事物号部分，我用来标示它们是一个事物

TABLE_ID：是行格式特有的，这个用来保证slave复制的正确性

Use_db: use database 默认当前在哪个数据下，是query event特有的

DB_NAME: 这是map event特有的，也是行格式特有的，记录的是表所在的数据库，和Use_db有区别，

Statment(35b-trun)：在query event中记录的语句为了方便输出将语句做35字节阶段

/*!Trx begin!*/：表示这是一个事物的开始，如果是gtid模式需要向前推一个event，因为gtid event也算到事物中

/*!Trx end*/：自然就是事物的结束点

mysqlbinlog中也是一致的比如：

>Gtid Event:Pos:194(0Xc2) N_pos:259(0X103) Time:1487035999 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100463

[root@testmy ~]# date -s "@1487035999"

Tue Feb 14 09:33:19 CST 2017

对应mysqlbinlog的如下部分：

# at 194

#170214  9:33:19 server id 93157  end_log_pos 259 CRC32 0xb664a0c6      GTID [commit=yes]

SET @@SESSION.GTID_NEXT= '4a6f2a67-5d87-11e6-a6bd-000c29a879a3:1100463'/*!*/;

关于 TABLE_ID Use_db DB_NAME 必须着重说明：

在语句模式下：

use mysql

insert into test.testtool select * from test.testnull2;

这种语句和salve 端--replicate-do-db 类参数一起出现可能呢出现问题，因为语句模式只有query event 而没有map event，

导致过滤的时候这种语句因为它是在mysql下：

如下：

>Gtid Event:Pos:800(0X320) N_pos:865(0X361) Time:1487036301 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100467

-->Query Event:Pos:865(0X361) N_Pos:945(0X3b1) Time:1487036301 Event_size:80(bytes)

Exe_time:0  Use_db:mysql Statment(35b-trun):BEGIN /*!Trx begin!*/ Gno:1100467

-->Query Event:Pos:945(0X3b1) N_Pos:1074(0X432) Time:1487036301 Event_size:129(bytes)

Exe_time:0  Use_db:mysql Statment(35b-trun):insert into test.testtool select *  Gno:1100467

>Xid Event:Pos:1074(0X432) N_Pos:1105(0X451) Time:1487036301 Event_size:31(bytes)

COMMIT; /*!Trx end*/ Gno:1100467

注意这里Use_db:mysql

同样的语句在row格式下：

>Gtid Event:Pos:497(0X1f1) N_pos:562(0X232) Time:1487036269 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100466

-->Query Event:Pos:562(0X232) N_Pos:635(0X27b) Time:1487036269 Event_size:73(bytes)

Exe_time:0  Use_db:mysql Statment(35b-trun):BEGIN /*!Trx begin!*/ Gno:1100466

---->Map Event:Pos635(0X27b) N_pos:692(0X2b4) Time:1487036269 Event_size:57(bytes)

TABLE_ID:346 DB_NAME:test TABLE_NAME:testtool Gno:1100466

------>Insert Event:Pos:692(0X2b4) N_pos:769(0X301) Time:1487036269 Event_size:77(bytes)

Dml on table: test.testtool  table_id:346 Gno:1100466

>Xid Event:Pos:769(0X301) N_Pos:800(0X320) Time:1487036269 Event_size:31(bytes)

COMMIT; /*!Trx end*/ Gno:1100466

注意Use_db:mysql DB_NAME:test Dml on table: test.testtool ,记录的正确性就是table id联系的。

这也保证row格式下复制的正确性

但是DDL语句始终是statement模式。

这一点在MYSQL 官方文档也有说明：

Statement-based replication.  Tell the slave SQL thread to restrict replication to statements where

the default database (that is, the one selected by USE) is db_name. To specify more than one database,

use this option multiple times, once for each database; however, doing so does notreplicate crossdatabase statements

such as UPDATE some_db.some_tableSET foo='bar'while a different

database (or no database) is selected.

Warning

To specify multiple databases you mustuse multiple instances of this option.

Because database names can contain commas, if you supply a comma

separated list then the list will be treated as the name of a single database.

An example of what does not work as you might expect when using statement-based replication: If

the slave is started with --replicate-do-db=salesand you issue the following statements on the

master, the UPDATE statement is not replicated:

USE prices;

UPDATE sales.january SET amount=amount+1000;

The main reason for this “check just the default database”behavior is that it is difficult from the statement

alone to know whether it should be replicated (for example, if you are using multiple-table DELETE

statements or multiple-table UPDATEstatements that act across multiple databases). It is also faster to

check only the default database rather than all databases if there is no need.

*3、如何辨别是一个事物

这就是row模式下的一个事物(算上GTID生成event)：

>Gtid Event:Pos:362(0X16a) N_pos:427(0X1ab) Time:1487036014 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100464

-->Query Event:Pos:427(0X1ab) N_Pos:499(0X1f3) Time:1487036014 Event_size:72(bytes)

Exe_time:0  Use_db:test Statment(35b-trun):BEGIN /*!Trx begin!*/ Gno:1100464

---->Map Event:Pos499(0X1f3) N_pos:556(0X22c) Time:1487036014 Event_size:57(bytes)

TABLE_ID:346 DB_NAME:test TABLE_NAME:testtool Gno:1100464

------>Insert Event:Pos:556(0X22c) N_pos:633(0X279) Time:1487036014 Event_size:77(bytes)

Dml on table: test.testtool  table_id:346 Gno:1100464

>Xid Event:Pos:633(0X279) N_Pos:664(0X298) Time:1487036014 Event_size:31(bytes)

COMMIT; /*!Trx end*/ Gno:1100464

这是statement模式下的一个事物(算上GTID生成event)：

>Gtid Event:Pos:800(0X320) N_pos:865(0X361) Time:1487036301 Event_size:65(bytes)

Gtid:4a6f2a67-5d87-11e6-a6bd-0c29a879a3:1100467

-->Query Event:Pos:865(0X361) N_Pos:945(0X3b1) Time:1487036301 Event_size:80(bytes)

Exe_time:0  Use_db:mysql Statment(35b-trun):BEGIN /*!Trx begin!*/ Gno:1100467

-->Query Event:Pos:945(0X3b1) N_Pos:1074(0X432) Time:1487036301 Event_size:129(bytes)

Exe_time:0  Use_db:mysql Statment(35b-trun):insert into test.testtool select *  Gno:1100467

>Xid Event:Pos:1074(0X432) N_Pos:1105(0X451) Time:1487036301 Event_size:31(bytes)

COMMIT; /*!Trx end*/ Gno:1100467

Total now 部分：

Trx total[counts]:1 总的事物个数

Event total[counts]:10 总的event个数

Avg binlog size(/sec):5.007(bytes)[0.005(kb)]

Avg binlog size(/min):300.426(bytes)[0.293(kb)]

平均每秒每分生成的binlog大小

--Piece view:

(1)Time:1487035963-1487035999(36(s)) piece:141(bytes)[0.138(kb)]

(2)Time:1487035999-1487036014(15(s)) piece:141(bytes)[0.138(kb)]

(3)Time:1487036014-1487036014(0(s)) piece:141(bytes)[0.138(kb)]

(4)Time:1487036014-1487036014(0(s)) piece:141(bytes)[0.138(kb)]

(5)Time:1487036014-1487036104(90(s)) piece:141(bytes)[0.138(kb)]

根据指定piece大小得到一个高度均衡直方图，这个直方图用于发现是否有某个时间段生成binlog特别大，

如上明显1487036014这个时间的event 比较多，可能是大事物。

--Large than 257(bytes) trx:

Node trx find!

大约257 BYTES个事务，这里测试的没有如果有如下：

--Large than 1000(bytes) trx:

(1)Trx_size:815352746(bytes)[796242.938(kb)] trx_begin_p:1174606[0X11EC4E] trx_end_p:816527352[0X30AB37F8]

(2)Trx_size:3108(bytes)[3.035(kb)] trx_begin_p:817520533[0X30BA5F95] trx_end_p:817523641[0X30BA6BB9]

(3)Trx_size:3116(bytes)[3.043(kb)] trx_begin_p:817523706[0X30BA6BFA] trx_end_p:817526822[0X30BA7826]

(4)Trx_size:5910(bytes)[5.771(kb)] trx_begin_p:817526887[0X30BA7867] trx_end_p:817532797[0X30BA8F7D]

(5)Trx_size:11538(bytes)[11.268(kb)] trx_begin_p:817532862[0X30BA8FBE] trx_end_p:817544400[0X30BABCD0]

(6)Trx_size:22754(bytes)[22.221(kb)] trx_begin_p:817544465[0X30BABD11] trx_end_p:817567219[0X30BB15F3]

(7)Trx_size:45226(bytes)[44.166(kb)] trx_begin_p:817567284[0X30BB1634] trx_end_p:817612510[0X30BBC6DE]

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/7728585/viewspace-2133534/，如需转载，请注明出处，否则将追究法律责任。