1.下载maxwell
2.解压maxwell
3.修改MySQL的配置文件my.cnf

[root@hadoop000 ~]# cd /etc[root@hadoop000 etc]# vi my.cnf

[mysqld]server-id  = 1binlog_format = ROW

PS:binlog_format必须设为ROW模式,如果设为其他模式,比如binlog_format = STATEMENT,启动maxwell时会报错

[root@hadoop001 maxwell-1.17.1]# bin/maxwell --user='maxwell' --password='maxwell' --host='127.0.0.1' --producer=stdoutUsing kafka version: 1.0.008:57:08,194 WARN  MaxwellMetrics - Metrics will not be exposed: metricsReportingType not configured.08:57:08,521 INFO  TaskManager - Stopping 0 tasks08:57:08,521 ERROR TaskManager - cause: com.zendesk.maxwell.MaxwellCompatibilityError: variable binlog_format must be set to 'ROW'at com.zendesk.maxwell.MaxwellMysqlStatus.ensureVariableState(MaxwellMysqlStatus.java:42) ~[maxwell-1.17.1.jar:1.17.1]at com.zendesk.maxwell.MaxwellMysqlStatus.ensureReplicationMysqlState(MaxwellMysqlStatus.java:70) ~[maxwell-1.17.1.jar:1.17.1]at com.zendesk.maxwell.Maxwell.startInner(Maxwell.java:173) ~[maxwell-1.17.1.jar:1.17.1]at com.zendesk.maxwell.Maxwell.start(Maxwell.java:156) ~[maxwell-1.17.1.jar:1.17.1]at com.zendesk.maxwell.Maxwell.main(Maxwell.java:245) ~[maxwell-1.17.1.jar:1.17.1]08:57:08,525 INFO  TaskManager - Stopped all taskscom.zendesk.maxwell.MaxwellCompatibilityError: variable binlog_format must be set to 'ROW'at com.zendesk.maxwell.MaxwellMysqlStatus.ensureVariableState(MaxwellMysqlStatus.java:42)at com.zendesk.maxwell.MaxwellMysqlStatus.ensureReplicationMysqlState(MaxwellMysqlStatus.java:70)at com.zendesk.maxwell.Maxwell.startInner(Maxwell.java:173)at com.zendesk.maxwell.Maxwell.start(Maxwell.java:156)at com.zendesk.maxwell.Maxwell.main(Maxwell.java:245)

4.启动MySQL

[mysqladmin@hadoop000 ~]$ service mysql startStarting MySQL....                                         [  OK  ][mysqladmin@hadoop000 ~]$ mysql -uroot -pEnter password: Welcome to the MySQL monitor.  Commands end with ; or \g.Your MySQL connection id is 1Server version: 5.6.23-log MySQL Community Server (GPL)Copyright (c) 2000, 2015, Oracle and/or its affiliates. All rights reserved.Oracle is a registered trademark of Oracle Corporation and/or itsaffiliates. Other names may be trademarks of their respectiveowners.Type 'help;' or '\h' for help. Type '\c' to clear the current input statement.mysql> show variables like '%binlog%';+-----------------------------------------+----------------------+| Variable_name                           | Value                |+-----------------------------------------+----------------------+| binlog_cache_size                       | 32768                || binlog_checksum                         | CRC32                || binlog_direct_non_transactional_updates | OFF                  || binlog_error_action                     | IGNORE_ERROR         || binlog_format                           | ROW                  | //binlog_format已生效| binlog_gtid_simple_recovery             | OFF                  || binlog_max_flush_queue_time             | 0                    || binlog_order_commits                    | ON                   || binlog_row_image                        | FULL                 || binlog_rows_query_log_events            | OFF                  || binlog_stmt_cache_size                  | 32768                || binlogging_impossible_mode              | IGNORE_ERROR         || innodb_api_enable_binlog                | OFF                  || innodb_locks_unsafe_for_binlog          | ON                   || max_binlog_cache_size                   | 18446744073709547520 || max_binlog_size                         | 1073741824           || max_binlog_stmt_cache_size              | 18446744073709547520 || simplified_binlog_gtid_recovery         | OFF                  || sync_binlog                             | 0                    |+-----------------------------------------+----------------------+19 rows in set (0.00 sec)

5.创建maxwell的数据库和用户

mysql> create database maxwell;mysql> GRANT ALL on maxwell.* to 'maxwell'@'%' identified by 'maxwell';mysql> GRANT SELECT, REPLICATION CLIENT, REPLICATION SLAVE on *.* to 'maxwell'@'%';mysql> flush privileges;    //一定记得刷新权限

6.在MySQL中创建一个测试数据库和表

//创建一个数据库mysql> create database hlwtest;mysql> grant all privileges on hlwtest.* to hlw@'%' identified by '123456';mysql> flush privileges;mysql> use hlwtest;Database changedmysql> show tables;Empty set (0.00 sec)//建表语句create table emp (    empno numeric(4) primary key,    ename varchar(10),    job varchar(9),    mgr numeric(4),    hiredate datetime,    sal numeric(7,2),    comm numeric(7,2),    deptno numeric(2));//查看数据库中的表mysql> show tables;+-------------------+| Tables_in_hlwtest |+-------------------+| emp               |+-------------------+mysql> desc emp;+----------+--------------+------+-----+---------+-------+| Field    | Type         | Null | Key | Default | Extra |+----------+--------------+------+-----+---------+-------+| empno    | decimal(4,0) | NO   | PRI | NULL    |       || ename    | varchar(10)  | YES  |     | NULL    |       || job      | varchar(9)   | YES  |     | NULL    |       || mgr      | decimal(4,0) | YES  |     | NULL    |       || hiredate | datetime     | YES  |     | NULL    |       || sal      | decimal(7,2) | YES  |     | NULL    |       || comm     | decimal(7,2) | YES  |     | NULL    |       || deptno   | decimal(2,0) | YES  |     | NULL    |       |+----------+--------------+------+-----+---------+-------+//导入测试数据mysql> LOAD DATA LOCAL INFILE '/root/data/emp.txt' INTO TABLE emp FIELDS TERMINATED BY '\t' LINES TERMINATED BY '\r\n';mysql> select * from emp;+-------+--------+-----------+------+---------------------+----------+---------+--------+| empno | ename  | job       | mgr  | hiredate            | sal      | comm    | deptno |+-------+--------+-----------+------+---------------------+----------+---------+--------+|  7369 | SMITH  | CLERK     | 7902 | 1980-12-17 00:00:00 |   800.00 |    0.00 |     20 ||  7499 | ALLEN  | SALESMAN  | 7698 | 1981-02-20 00:00:00 |  1600.00 |  300.00 |     30 ||  7521 | WARD   | SALESMAN  | 7698 | 1981-02-22 00:00:00 |  1250.00 |  500.00 |     30 ||  7566 | JONES  | MANAGER   | 7839 | 1981-04-02 00:00:00 |  2975.00 |    0.00 |     20 ||  7654 | MARTIN | SALESMAN  | 7698 | 1981-09-28 00:00:00 |  1250.00 | 1400.00 |     30 ||  7698 | BLAKE  | MANAGER   | 7839 | 1981-05-01 00:00:00 |  2850.00 |    0.00 |     30 ||  7782 | CLARK  | MANAGER   | 7839 | 1981-06-09 00:00:00 |  2450.00 |    0.00 |     10 ||  7788 | SCOTT  | ANALYST   | 7566 | 1987-04-19 00:00:00 |  3000.00 |    0.00 |     20 ||  7839 | KING   | PRESIDENT |    0 | 1981-11-17 00:00:00 |  5000.00 |    0.00 |     10 ||  7844 | TURNER | SALESMAN  | 7698 | 1981-09-08 00:00:00 |  1500.00 |    0.00 |     30 ||  7876 | ADAMS  | CLERK     | 7788 | 1987-05-23 00:00:00 |  1100.00 |    0.00 |     20 ||  7900 | JAMES  | CLERK     | 7698 | 1981-12-03 00:00:00 |   950.00 |    0.00 |     30 ||  7902 | FORD   | ANALYST   | 7566 | 1981-12-03 00:00:00 |  3000.00 |    0.00 |     20 ||  7934 | MILLER | CLERK     | 7782 | 1982-01-23 00:00:00 |  1300.00 |    0.00 |     10 ||  8888 | HIVE   | PROGRAM   | 7839 | 1988-01-23 00:00:00 | 10300.00 |    0.00 |   NULL |+-------+--------+-----------+------+---------------------+----------+---------+--------+

7.启动maxwell进程,首先使用stdout模式测试一下

[root@hadoop001 maxwell-1.17.1]# bin/maxwell --user='maxwell' --password='maxwell' --host='127.0.0.1' --producer=stdoutUsing kafka version: 1.0.019:04:48,030 WARN  MaxwellMetrics - Metrics will not be exposed: metricsReportingType not configured.19:04:48,365 INFO  SchemaStoreSchema - Creating maxwell database19:04:48,559 INFO  Maxwell - Maxwell v1.17.1 is booting (StdoutProducer), starting at Position[BinlogPosition[mysql-bin.000014:5999], lastHeartbeat=0]19:04:48,713 INFO  AbstractSchemaStore - Maxwell is capturing initial schema19:04:49,339 INFO  BinlogConnectorReplicator - Setting initial binlog pos to: mysql-bin.000014:599919:04:49,506 INFO  BinaryLogClient - Connected to 127.0.0.1:3306 at mysql-bin.000014/5999 (sid:6379, cid:10)19:04:49,506 INFO  BinlogConnectorLifecycleListener - Binlog connected.

8.向测试表中insert一条数据

mysql> insert into emp (empno,ename,job,mgr,hiredate,sal,comm,deptno) values (6001,'SIWA','DESIGNER',7001,'2019-03-08',1000,6000,40);

9.查看maxwell控制台

{"database":"hlwtest","table":"emp","type":"insert","ts":1552043107,"xid":444,"commit":true,"data":{"empno":6001,"ename":"SIWA","job":"DESIGNER","mgr":7001,"hiredate":"2019-03-08 00:00:00","sal":1000.00,"comm":6000.00,"deptno":40}}

10.再insert一条数据,查看下binlog日志内容

//MySQLmysql> update emp set sal=500 where empno=6001;//Maxwell{"database":"hlwtest","table":"emp","type":"update","ts":1552090904,"xid":138,"commit":true,"data":{"empno":6001,"ename":"SIWA","job":"DESIGNER","mgr":7001,"hiredate":"2019-03-08 00:00:00","sal":500.00,"comm":6000.00,"deptno":40},"old":{"sal":1000.00}}

首先使用show binlog events的方式查看

//MySQLmysql> show master status;+------------------+----------+--------------+------------------+-------------------+| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |+------------------+----------+--------------+------------------+-------------------+| mysql-bin.000016 |    40967 |              |                  |                   |+------------------+----------+--------------+------------------+-------------------+mysql> show binlog events in 'mysql-bin.000016' from 3954 limit 3;+------------------+------+-------------+-----------+-------------+--------------------------------+| Log_name         | Pos  | Event_type  | Server_id | End_log_pos | Info                           |+------------------+------+-------------+-----------+-------------+--------------------------------+| mysql-bin.000016 | 3954 | Table_map   |         1 |        4025 | table_id: 71 (hlwtest.emp)     || mysql-bin.000016 | 4025 | Update_rows |         1 |        4127 | table_id: 71 flags: STMT_END_F || mysql-bin.000016 | 4127 | Xid         |         1 |        4158 | COMMIT /* xid=138 */           |+------------------+------+-------------+-----------+-------------+--------------------------------+

`
从解析的binlog中可以看出row模式下,DML操作会记录为:TABLE_MAP_EVENT+ROW_LOG_EVENT(包括WRITE_ROWS_EVENT ,UPDATE_ROWS_EVENT,DELETE_ROWS_EVENT).

为什么一个update在ROW模式下需要分解成两个event:
一个Table_map,一个Update_rows。我们想象一下,一个update如果更新了10000条数据,那么对应的表结构信息是否需要记录10000次?其实是对同一个表的操作,所以这里binlog只是记录了一个Table_map用于记录表结构相关信息,而后面的Update_rows记录了更新数据的行信息,注意此表中的table_id是会变化的。

然后使用/usr/local/mysql/bin目录下的mysqlbinlog工具解析binlog日志

[mysqladmin@hadoop001 bin]$ mysqlbinlog --start-position="3954" --stop-position="4158" /usr/local/mysql/arch/mysql-bin.000016/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=1*/;/*!40019 SET @@session.max_insert_delayed_threads=0*/;/*!50003 SET @OLD_COMPLETION_TYPE=@@COMPLETION_TYPE,COMPLETION_TYPE=0*/;DELIMITER /*!*/;# at 3954#190309  8:21:44 server id 1  end_log_pos 4025 CRC32 0xf962b4b1         Table_map: `hlwtest`.`emp` mapped to number 71# at 4025#190309  8:21:44 server id 1  end_log_pos 4127 CRC32 0x5013e9f3         Update_rows: table id 71 flags: STMT_END_FBINLOG 'GAeDXBMBAAAARwAAALkPAAAAAEcAAAAAAAEAB2hsd3Rlc3QAA2VtcAAI9g8P9hL29vYPBAAKAAkABAAABwIHAgIA/rG0Yvk=GAeDXB8BAAAAZgAAAB8QAAAAAEcAAAAAAAEAAgAI//8Al3EEU0lXQQhERVNJR05FUptZmaKQAACAA+gAgBdwAKgAl3EEU0lXQQhERVNJR05FUptZmaKQAACAAfQAgBdwAKjz6RNQ'/*!*/;# at 4127#190309  8:21:44 server id 1  end_log_pos 4158 CRC32 0xfc03c0c3         Xid = 138COMMIT/*!*/;DELIMITER ;# End of log fileROLLBACK /* added by mysqlbinlog */;/*!50003 SET COMPLETION_TYPE=@OLD_COMPLETION_TYPE*/;/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=0*/;

完全看不懂,这时需要添加参数(--base64-output=decode-rows -v)对输出结果解码

[mysqladmin@hadoop001 bin]$ mysqlbinlog --base64-output=decode-rows -v --start-position="3954" --stop-position="4158" /usr/local/mysql/arch/mysql-bin.000016/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=1*/;/*!40019 SET @@session.max_insert_delayed_threads=0*/;/*!50003 SET @OLD_COMPLETION_TYPE=@@COMPLETION_TYPE,COMPLETION_TYPE=0*/;DELIMITER /*!*/;# at 3954#190309  8:21:44 server id 1  end_log_pos 4025 CRC32 0xf962b4b1         Table_map: `hlwtest`.`emp` mapped to number 71# at 4025#190309  8:21:44 server id 1  end_log_pos 4127 CRC32 0x5013e9f3         Update_rows: table id 71 flags: STMT_END_F### UPDATE `hlwtest`.`emp`### WHERE###   @1=6001###   @2='SIWA'###   @3='DESIGNER'###   @4=7001###   @5='2019-03-08 00:00:00'###   @6=1000.00###   @7=6000.00###   @8=40### SET###   @1=6001###   @2='SIWA'###   @3='DESIGNER'###   @4=7001###   @5='2019-03-08 00:00:00'###   @6=500.00###   @7=6000.00###   @8=40# at 4127#190309  8:21:44 server id 1  end_log_pos 4158 CRC32 0xfc03c0c3         Xid = 138COMMIT/*!*/;DELIMITER ;# End of log fileROLLBACK /* added by mysqlbinlog */;/*!50003 SET COMPLETION_TYPE=@OLD_COMPLETION_TYPE*/;/*!50530 SET @@SESSION.PSEUDO_SLAVE_MODE=0*/;

可以看到Mysql每次列的修改(update)都需要记录表中所有列的值。这样就存在一个问题,如果表中包含很多的大字段,表的单行长度就会非常长,这样每次update就会导致大量的 binlog空间生成。针对这个问题,在mysql 5.6中进行了改进,复制支持”row image control” ,只记录修改的列而不是行中所有的列,这对一些包含 BLOGs 字段的数据来说可以节省很大的处理能力,因此此项改进不仅节省了磁盘空间,同时也提升了性能。

11.当数据库的binlog format 是statement 模式时,查看下binlog日志

mysql> update emp set sal=501 where empno=6001;mysql> show master status;+------------------+----------+--------------+------------------+-------------------+| File             | Position | Binlog_Do_DB | Binlog_Ignore_DB | Executed_Gtid_Set |+------------------+----------+--------------+------------------+-------------------+| mysql-bin.000017 |      355 |              |                  |                   |+------------------+----------+--------------+------------------+-------------------+1 row in set (0.00 sec)mysql> show binlog events in 'mysql-bin.000017';+------------------+-----+-------------+-----------+-------------+--------------------------------------------------------+| Log_name         | Pos | Event_type  | Server_id | End_log_pos | Info                                                   |+------------------+-----+-------------+-----------+-------------+--------------------------------------------------------+| mysql-bin.000017 |   4 | Format_desc |         1 |         120 | Server ver: 5.6.23-log, Binlog ver: 4                  || mysql-bin.000017 | 120 | Query       |         1 |         205 | BEGIN                                                  || mysql-bin.000017 | 205 | Query       |         1 |         324 | use `hlwtest`; update emp set sal=501 where empno=6001 || mysql-bin.000017 | 324 | Xid         |         1 |         355 | COMMIT /* xid=24 */                                    |+------------------+-----+-------------+-----------+-------------+--------------------------------------------------------+4 rows in set (0.01 sec)//使用/usr/local/mysql/bin目录下的mysqlbinlog工具解析binlog日志[mysqladmin@hadoop001 ~]$ /usr/local/mysql/binmysqlbinlog /usr/local/mysql/arch/mysql-bin.000017...# at 205#190309  9:00:29 server id 1  end_log_pos 324 CRC32 0x1c4a065d  Query   thread_id=3     exec_time=0     error_code=0use `hlwtest`/*!*/;SET TIMESTAMP=1552093229/*!*/;update emp set sal=501 where empno=6001/*!*/;# at 324#190309  9:00:29 server id 1  end_log_pos 355 CRC32 0xf7071a73  Xid = 24COMMIT/*!*/;...

作者:白面葫芦娃92原文:https://www.jianshu.com/p/bbfdb5efef65

回归原创文章:

若泽数据-高级班&线下班报名

高级班学员高薪offer32w,你比他高吗?

捷报: 高级班学员月薪22K及上周3家offer的面试题

刚出炉的3家大数据面试题(含高级),你会吗?

捷报:刚出炉年薪30w的offer和面试题

捷报:高级班学员年薪37.4W的offer及3家面试题

加我,进大数据V群

日志读取_实时数仓之Maxwell读取MySQL binlog日志相关推荐

  1. adb实时获取屏幕_实时数仓 | 你需要的是一款合适且强大的OLAP数据库(上)

    欢迎扫码关注我的公众号,回复[JAVAPDF]可以获得一份200页秋招面试题! 前言 今年有个现象,实时数仓建设突然就被大家所关注.我个人在公众号也写过和转载过几篇关于实时数据仓库的文章和方案. 但是 ...

  2. 数据仓库—stg层_数据仓库之Hive快速入门 - 离线实时数仓架构

    数据仓库VS数据库 数据仓库的定义: 数据仓库是将多个数据源的数据经过ETL(Extract(抽取).Transform(转换).Load(加载))理之后,按照一定的主题集成起来提供决策支持和联机分析 ...

  3. 聊聊实时数仓架构设计

    什么是实时数仓 首先需要明确什么是实时数仓,百度百科与维基百科都没有给出具体说明,哪究竟什么才是实时数仓呢?是不是可以通过实时流实时获取数据就是实时数仓?或者说流批一体就是实时数仓?在或者全面采用实时 ...

  4. 几种实时数仓架构设计思路

    01 什么是实时数仓 首先需要明确什么是实时数仓,百度百科与维基百科都没有给出具体说明,哪究竟什么才是实时数仓呢?是不是可以通过实时流实时获取数据就是实时数仓?或者说流批一体就是实时数仓?在或者全面采 ...

  5. 数仓 调度_网易实时数仓实践

    分享嘉宾:马进 网易杭研 技术专家 编辑整理:张满意 出品平台:DataFunTalk 导读:随着大数据技术的进步,各种计算框架的涌现,数据仓库相关技术难题已经从离线数仓逐渐过渡到实时数仓,越来越多的 ...

  6. 【实时数仓】DWD层需求分析及实现思路、idea环境搭建、实现DWD层处理用户行为日志的功能

    文章目录 一 DWD层需求分析及实现思路 1 分层需求分析 2 每层的职能 3 DWD层职能详细介绍 (1)用户行为日志数据 (2)业务数据 4 DWD层数据准备实现思路 二 环境搭建 1 创建mav ...

  7. hive增量表和全量表_基于 Flink + Hive 构建流批一体准实时数仓

    基于 Hive 的离线数仓往往是企业大数据生产系统中不可缺少的一环.Hive 数仓有很高的成熟度和稳定性,但由于它是离线的,延时很大.在一些对延时要求比较高的场景,需要另外搭建基于 Flink 的实时 ...

  8. 怎样不停请求接口实现实时刷新_快狗打车实时数仓和基于Hologres的数据服务建设...

    前言 数据的实时化是最近几年数据行业很重要的趋势,我们在去年底也建立起新一代的实时数仓,但是在数据应用上一直没有取得很大的突破,我们希望实时数仓不仅仅是支撑大屏.核心实时报表.个别实时应用等简单的场景 ...

  9. 数据查询和业务流分开_滴滴实时数仓逐层剖解:实时与离线数据误差0.5%

    原标题:滴滴实时数仓逐层剖解:实时与离线数据误差< 作者介绍 潘澄,资深软件开发工程师.负责实时数据仓库建设,多年数据相关工作经验,专注数据建模.数据仓库.实时数据技术等领域. 朱峰,高级软件开 ...

最新文章

  1. Python 删除list 里面的空值
  2. 剑指offer-数组中出现次数超过一半的数字
  3. Pat甲级 1001 A+B Format
  4. Python里面数组拼接方法介绍
  5. java 线程安全的原因_Java并发编程——线程安全性深层原因
  6. CentOS7 DNS的添加
  7. 如何用Python发送通知到微信?
  8. python3中的property使用方法
  9. 链表C++ | 从尾部打印头部(使用栈、递归实现)_3
  10. Python数据结构:二分查找
  11. smc数显压力表设定方法_自动增压泵不停止工作原因及解决方法
  12. element-ui 源码学习
  13. linux内存基础知识和相关调优方案
  14. 部署和调优 2.1 squid正向代理
  15. 基于线程池技术的web服务器
  16. 20135202闫佳歆——信息安全系统设计基础第七周学习总结
  17. RabbitMQ 安装使用,Centos系统安装RabbitMQ、Docker安装启动RabbitMQ
  18. 图像识别,ocr 技术,有兴趣的可以了解一下
  19. 存储基础:LUN、ThickLUN与Thin LUN的联系和区别
  20. 用 PHP 进行 HTTP 认证,Basic Auth

热门文章

  1. 搜狗赴美上市:首日涨3.8%,王小川哽咽感谢母亲
  2. java 发送邮件带附件
  3. Pdftk - The PDF Toolkit
  4. 高三!忏悔ing~(在这儿的第一篇随笔)
  5. 徐明星斥资超4亿香港买壳:史玉柱女儿现身股东行列
  6. 华为p20:拍美景,听讲解,旅行更智能
  7. Linux文件的三个时间概念
  8. Centos下oracle11g R2的启动与关闭监听、数据库
  9. iPhone Instruments工具使用_检测内存泄露(转)
  10. Hadoop系列之九:Hadoop集群伪分布式模式的实现详解