4、离线数仓数据同步策略（全量表数据同步、增量表数据同步、首日同步、采集通道脚本）

1、离线数仓同步数据

1.1 用户行为数据同步

1.1.1 数据通道

用户行为数据由Flume从Kafka直接同步到HDFS，由于离线数仓采用Hive的分区表按天统计，所以目标路径要包含一层日期。具体数据流向如下图所示。

1.1.2 日志消费Flume配置概述

按照规划，该Flume需将Kafka中topic_log的数据发往HDFS。并且对每天产生的用户行为日志进行区分，将不同天的数据发往HDFS不同天的路径。
此次选择KafkaSource、FileChannel、HDFSSink

2.1.3 日志消费Flume配置实操

1、创建Flume配置文件
在hadoop104节点的Flume的job目录下创建kafka_to_hdfs_log.conf

vim job/kafka_to_hdfs_log.conf 

2、配置内容如下

#定义组件
a1.sources=r1
a1.channels=c1
a1.sinks=k1

#配置source1
a1.sources.r1.type = org.apache.flume.source.kafka.KafkaSource
a1.sources.r1.batchSize = 5000
a1.sources.r1.batchDurationMillis = 2000
a1.sources.r1.kafka.bootstrap.servers = hadoop102:9092,hadoop103:9092,hadoop104:9092
a1.sources.r1.kafka.topics=topic_log
a1.sources.r1.interceptors = i1
a1.sources.r1.interceptors.i1.type = org.zhm.gmall.flume.interceptor.TimestampInterceptor$Builder

#配置channel
a1.channels.c1.type = file
a1.channels.c1.checkpointDir = /opt/module/flume-1.9.0/checkpoint/behavior1
a1.channels.c1.dataDirs = /opt/module/flume-1.9.0/data/behavior1
a1.channels.c1.maxFileSize = 2146435071
a1.channels.c1.capacity = 1000000
a1.channels.c1.keep-alive = 6

#配置sink
a1.sinks.k1.type = hdfs
a1.sinks.k1.hdfs.path = /origin_data/gmall/log/topic_log/%Y-%m-%d
a1.sinks.k1.hdfs.filePrefix = log
a1.sinks.k1.hdfs.round = false


a1.sinks.k1.hdfs.rollInterval = 10
a1.sinks.k1.hdfs.rollSize = 134217728
a1.sinks.k1.hdfs.rollCount = 0

#控制输出文件类型
a1.sinks.k1.hdfs.fileType = CompressedStream
a1.sinks.k1.hdfs.codeC = gzip

#组装 
a1.sources.r1.channels = c1
a1.sinks.k1.channel = c1

注：配置优化
1、FileChannel优化
通过配置dataDirs指向多个路径，每个路径对应不同的硬盘，增大Flume吞吐量。
checkpointDir和backupCheckpointDir也尽量配置在不同硬盘对应的目录中，保证checkpoint坏掉后，可以快速使用backupCheckpointDir恢复数据
2、HDFS Sink优化
（1）HDFS存入大量小文件，有什么影响？
元数据层面：每个小文件都有一份元数据，其中包括文件路径，文件名，所有者，所属组，权限，创建时间等，这些信息都保存在Namenode内存中。所以小文件过多，会占用Namenode服务器大量内存，影响Namenode性能和使用寿命
计算层面：默认情况下MR会对每个小文件启用一个Map任务计算，非常影响计算性能。同时也影响磁盘寻址时间。
（2）HDFS小文件处理
官方默认的这三个参数配置写入HDFS后会产生小文件，hdfs.rollInterval、hdfs.rollSize、hdfs.rollCount
基于以上hdfs.rollInterval=3600，hdfs.rollSize=134217728，hdfs.rollCount =0几个参数综合作用，效果如下：
（1）文件在达到128M时会滚动生成新文件
（2）文件创建超3600秒时会滚动生成新文件
3、编写Flume拦截器
（1）数据漂移问题

（2）在com.zhm.gmall.flume.interceptor包下创建TimestampInterceptor类

package org.zhm.gmall.flume.interceptor;

/**
 * @ClassName TimestampInterceptor
 * @Description TODO
 * @Author Zouhuiming
 * @Date 2023/6/19 18:33
 * @Version 1.0
 */
import com.alibaba.fastjson.JSONObject;
import org.apache.flume.Context;
import org.apache.flume.Event;
import org.apache.flume.interceptor.Interceptor;
import java.nio.charset.StandardCharsets;
import java.util.List;
import java.util.Map;

public class TimestampInterceptor implements Interceptor {


    @Override
    public void initialize() {

    }

    @Override
    public Event intercept(Event event) {

        //1、获取header和body的数据
        Map<String, String> headers = event.getHeaders();
        String log = new String(event.getBody(), StandardCharsets.UTF_8);

        //2、将body的数据类型转成jsonObject类型（方便获取数据）
        JSONObject jsonObject = JSONObject.parseObject(log);

        //3、header中timestamp时间字段替换成日志生成的时间戳（解决数据漂移问题）
        String ts = jsonObject.getString("ts");
        headers.put("timestamp", ts);

        return event;
    }

    @Override
    public List<Event> intercept(List<Event> list) {
        for (Event event : list) {
            intercept(event);
        }
        return list;
    }

    @Override
    public void close() {

    }

    public static class Builder implements Interceptor.Builder {
        @Override
        public Interceptor build() {
            return new TimestampInterceptor();
        }

        @Override
        public void configure(Context context) {
        }
    }
}

(3) Repackage
(4) You need to put the packaged package into the /opt/module/flume/lib folder of hadoop104

2.1.4 Log consumption Flume test

1. Start Zookeeper, Kafka, hadoop cluster
2. Start log collection Flume

f1.sh start

3. Start the log consumption Flume of hadoop104

bin/flume-ng agent -n a1 -c conf/ -f job/kafka_to_hdfs_log.conf -Dflume.root.logger=info,console

4. Generate simulation data

 lg.sh 

5. Observe whether there is data in HDFS

2.1.5 Flume startup and shutdown script for log consumption

If the above tests pass, for convenience, create a start-stop script for Flume here.
1. Create the script f2.sh in the /home/atguigu/bin directory of the hadoop102 node

 vim f2.sh

2. Fill in the following content in the script

#!/bin/bash

case $1 in
"start")
        echo " --------启动 hadoop104 日志数据flume-------"
        ssh hadoop104 "nohup /opt/module/flume/bin/flume-ng agent -n a1 -c /opt/module/flume/conf -f /opt/module/flume/job/kafka_to_hdfs_log.conf >/dev/null 2>&1 &"
;;
"stop")

        echo " --------停止 hadoop104 日志数据flume-------"
        ssh hadoop104 "ps -ef | grep kafka_to_hdfs_log | grep -v grep |awk '{print \$2}' | xargs -n1 kill"
;;
esac

3. Increase script permissions

chmod 777 f2.sh

2.2 Business data synchronization

2.2.1 Overview of Data Synchronization Strategy

Business data is an important data source of the data warehouse. We need to extract data from the business database regularly every day, transfer it to the data warehouse, and then analyze and count the data.
In order to ensure the correctness of the statistical results, it is necessary to ensure that the data in the data warehouse is synchronized with the business database. The calculation cycle of the offline data warehouse is usually days, so the data synchronization cycle is usually days, that is, only once a day.
Data synchronization strategies include full synchronization and incremental synchronization.
Full synchronization is to synchronize all the data in the business database to the data warehouse every day. This is the easiest way to ensure data synchronization on both sides.

Incremental synchronization is to synchronize only new and changed data in business data to the data warehouse every day. Tables that use daily incremental synchronization usually need to perform a full synchronization on the first day. .

2.2.2 Data Synchronization Strategy Selection

synchronization strategy	advantage	shortcoming
full sync	simple logic	In some cases the efficiency is relatively low. For example, the amount of data in a table is relatively large, but the daily data change ratio is very low. If daily full synchronization is used for it, it will repeatedly synchronize and store a large amount of the same data
Incremental synchronization	High efficiency, no need to synchronize and store duplicate data	The logic is complex, and it is necessary to integrate the daily new and changed data with the original data before it can be used

根据上述对比，可以得出以下结论：通常情况，业务表数据量比较大，优先考虑增量，数据量比较小，优先考虑全量

下图为各表同步策略：

2.2.3 数据同步工具概述

数据同步工具种类繁多，大致可分为两类，一类是以DataX、Sqoop为代表的基于Select查询的离线、批量同步工具，另一类是以Maxwell、Canal为代表的基于数据库数据变更日志（例如MySQL的binlog，其会实时记录所有的insert、update以及delete操作）的实时流式同步工具。
全量同步通常使用DataX、Sqoop等基于查询的离线同步工具。而增量同步既可以使用DataX、Sqoop等工具，也可使用Maxwell、Canal等工具，下面对增量同步不同方案进行简要对比。

增量同步方案	DataX/Sqoop	Maxwell/Canal
对数据的要求	原理是基于查询，故若想通过select查询获取新增及变化数据，就要求数据表中存在create_time、update_time等字段，然后根据这些字段获取变更数据	要求数据库记录变更操作，例如MySQL需要开启binlog
数据的中间状态	由于是离线批量同步，故若一条数据在一天中变化多次，该方案只能获取最后一个状态，中间状态无法获取	由于是实时获取所有的数据变更操作所以可以获取变更数据的所有中间状态

2.2.5 全量表数据同步

1、数据同步工具DataX部署
DataX学习链接

2、数据通道
全量表数据由DataX从MySQL业务数据库直接同步到HDFS，具体数据流向如下图所示。

3、DataX配置文件
我们需要为每张全量表编写一个DataX的json配置文件，此处以activity_info为例，配置文件内容如下：

{
    "job": {
        "content": [
            {
                "reader": {
                    "name": "mysqlreader",
                    "parameter": {
                        "column": [
                            "id",
                            "activity_name",
                            "activity_type",
                            "activity_desc",
                            "start_time",
                            "end_time",
                            "create_time"
                        ],
                        "connection": [
                            {
                                "jdbcUrl": [
                                    "jdbc:mysql://hadoop102:3306/gmall"
                                ],
                                "table": [
                                    "activity_info"
                                ]
                            }
                        ],
                        "password": "000000",
                        "splitPk": "",
                        "username": "root"
                    }
                },
                "writer": {
                    "name": "hdfswriter",
                    "parameter": {
                        "column": [
                            {
                                "name": "id",
                                "type": "bigint"
                            },
                            {
                                "name": "activity_name",
                                "type": "string"
                            },
                            {
                                "name": "activity_type",
                                "type": "string"
                            },
                            {
                                "name": "activity_desc",
                                "type": "string"
                            },
                            {
                                "name": "start_time",
                                "type": "string"
                            },
                            {
                                "name": "end_time",
                                "type": "string"
                            },
                            {
                                "name": "create_time",
                                "type": "string"
                            }
                        ],
                        "compress": "gzip",
                        "defaultFS": "hdfs://hadoop102:8020",
                        "fieldDelimiter": "\t",
                        "fileName": "activity_info",
                        "fileType": "text",
                        "path": "${targetdir}",
                        "writeMode": "append"
                    }
                }
            }
        ],
        "setting": {
            "speed": {
                "channel": 1
            }
        }
    }
}

注：由于目标路径包含一层日期，用于对不同天的数据加以区分，故path参数并未写死，需在提交任务时通过参数动态传入，参数名称为targetdir。

4、 DataX配置文件生成脚本
（1）在~/bin目录下创建gen_import_config.py脚本

# ecoding=utf-8
import json
import getopt
import os
import sys
import MySQLdb

#MySQL相关配置，需根据实际情况作出修改
mysql_host = "hadoop102"
mysql_port = "3306"
mysql_user = "root"
mysql_passwd = "000000"

#HDFS NameNode相关配置，需根据实际情况作出修改
hdfs_nn_host = "hadoop102"
hdfs_nn_port = "8020"

#生成配置文件的目标路径，可根据实际情况作出修改
output_path = "/opt/module/datax/job/import"


def get_connection():
    return MySQLdb.connect(host=mysql_host, port=int(mysql_port), user=mysql_user, passwd=mysql_passwd)


def get_mysql_meta(database, table):
    connection = get_connection()
    cursor = connection.cursor()
    sql = "SELECT COLUMN_NAME,DATA_TYPE from information_schema.COLUMNS WHERE TABLE_SCHEMA=%s AND TABLE_NAME=%s ORDER BY ORDINAL_POSITION"
    cursor.execute(sql, [database, table])
    fetchall = cursor.fetchall()
    cursor.close()
    connection.close()
    return fetchall


def get_mysql_columns(database, table):
    return map(lambda x: x[0], get_mysql_meta(database, table))


def get_hive_columns(database, table):
    def type_mapping(mysql_type):
        mappings = {
    
    
            "bigint": "bigint",
            "int": "bigint",
            "smallint": "bigint",
            "tinyint": "bigint",
            "decimal": "string",
            "double": "double",
            "float": "float",
            "binary": "string",
            "char": "string",
            "varchar": "string",
            "datetime": "string",
            "time": "string",
            "timestamp": "string",
            "date": "string",
            "text": "string"
        }
        return mappings[mysql_type]

    meta = get_mysql_meta(database, table)
    return map(lambda x: {
    
    "name": x[0], "type": type_mapping(x[1].lower())}, meta)


def generate_json(source_database, source_table):
    job = {
    
    
        "job": {
    
    
            "setting": {
    
    
                "speed": {
    
    
                    "channel": 3
                },
                "errorLimit": {
    
    
                    "record": 0,
                    "percentage": 0.02
                }
            },
            "content": [{
    
    
                "reader": {
    
    
                    "name": "mysqlreader",
                    "parameter": {
    
    
                        "username": mysql_user,
                        "password": mysql_passwd,
                        "column": get_mysql_columns(source_database, source_table),
                        "splitPk": "",
                        "connection": [{
    
    
                            "table": [source_table],
                            "jdbcUrl": ["jdbc:mysql://" + mysql_host + ":" + mysql_port + "/" + source_database]
                        }]
                    }
                },
                "writer": {
    
    
                    "name": "hdfswriter",
                    "parameter": {
    
    
                        "defaultFS": "hdfs://" + hdfs_nn_host + ":" + hdfs_nn_port,
                        "fileType": "text",
                        "path": "${targetdir}",
                        "fileName": source_table,
                        "column": get_hive_columns(source_database, source_table),
                        "writeMode": "append",
                        "fieldDelimiter": "\t",
                        "compress": "gzip"
                    }
                }
            }]
        }
    }
    if not os.path.exists(output_path):
        os.makedirs(output_path)
    with open(os.path.join(output_path, ".".join([source_database, source_table, "json"])), "w") as f:
        json.dump(job, f)


def main(args):
    source_database = ""
    source_table = ""

    options, arguments = getopt.getopt(args, '-d:-t:', ['sourcedb=', 'sourcetbl='])
    for opt_name, opt_value in options:
        if opt_name in ('-d', '--sourcedb'):
            source_database = opt_value
        if opt_name in ('-t', '--sourcetbl'):
            source_table = opt_value

    generate_json(source_database, source_table)


if __name__ == '__main__':
    main(sys.argv[1:])

注：（1）安装Python Mysql驱动

sudo yum install -y MySQL-python

（2）脚本使用说明

python gen_import_config.py -d database -t table

##通过-d传入数据库名，-t传入表名，执行上述命令即可生成该表的DataX同步配置文件。

（2）在~/bin目录下创建gen_import_config.sh脚本

vim ~/bin/gen_import_config.sh

脚本内容如下

#!/bin/bash

python ~/bin/gen_import_config.py -d gmall -t activity_info
python ~/bin/gen_import_config.py -d gmall -t activity_rule
python ~/bin/gen_import_config.py -d gmall -t base_category1
python ~/bin/gen_import_config.py -d gmall -t base_category2
python ~/bin/gen_import_config.py -d gmall -t base_category3
python ~/bin/gen_import_config.py -d gmall -t base_dic
python ~/bin/gen_import_config.py -d gmall -t base_province
python ~/bin/gen_import_config.py -d gmall -t base_region
python ~/bin/gen_import_config.py -d gmall -t base_trademark
python ~/bin/gen_import_config.py -d gmall -t cart_info
python ~/bin/gen_import_config.py -d gmall -t coupon_info
python ~/bin/gen_import_config.py -d gmall -t sku_attr_value
python ~/bin/gen_import_config.py -d gmall -t sku_info
python ~/bin/gen_import_config.py -d gmall -t sku_sale_attr_value
python ~/bin/gen_import_config.py -d gmall -t spu_info

（3）为gen_import_config.sh脚本增加执行权限

chmod 777 ~/bin/gen_import_config.sh 

（4）执行gen_import_config.sh脚本，生成配置文件

gen_import_config.sh 

（5）观察生成的配置文件

5、 测试生成的DataX配置文件
以activity_info为例，测试用脚本生成的配置文件是否可用。
1、创建目标路径
由于DataX同步任务要求目标路径提前存在，故需手动创建路径，当前activity_info表的目标路径应为/origin_data/gmall/db/activity_info_full/2020-06-14

hadoop fs -mkdir /origin_data/gmall/db/activity_info_full/2020-06-14

2、执行DataX同步命令

python /opt/module/datax/bin/datax.py -p"-Dtargetdir=/origin_data/gmall/db/activity_info_full/2020-06-14" /opt/module/datax/job/import/gmall.activity_info.json

3、观察同步结果

6、全量表数据同步脚本
为方便使用以及后续的任务调度，此处编写一个全量表数据同步脚本。

（1）在~/bin目录创建mysql_to_hdfs_full.sh
脚本内容如下

#!/bin/bash

DATAX_HOME=/opt/module/datax

# 如果传入日期则do_date等于传入的日期，否则等于前一天日期
if [ -n "$2" ] ;then
    do_date=$2
else
    do_date=`date -d "-1 day" +%F`
fi

#处理目标路径，此处的处理逻辑是，如果目标路径不存在，则创建；若存在，则清空，目的是保证同步任务可重复执行
handle_targetdir() {
    
    
  hadoop fs -test -e $1
  if [[ $? -eq 1 ]]; then
    echo "路径$1不存在，正在创建......"
    hadoop fs -mkdir -p $1
  else
    echo "路径$1已经存在"
    fs_count=$(hadoop fs -count $1)
    content_size=$(echo $fs_count | awk '{print $3}')
    if [[ $content_size -eq 0 ]]; then
      echo "路径$1为空"
    else
      echo "路径$1不为空，正在清空......"
      hadoop fs -rm -r -f $1/*
    fi
  fi
}

#数据同步
import_data() {
    
    
  datax_config=$1
  target_dir=$2

  handle_targetdir $target_dir
  python $DATAX_HOME/bin/datax.py -p"-Dtargetdir=$target_dir" $datax_config
}

case $1 in
"activity_info")
  import_data /opt/module/datax/job/import/gmall.activity_info.json /origin_data/gmall/db/activity_info_full/$do_date
  ;;
"activity_rule")
  import_data /opt/module/datax/job/import/gmall.activity_rule.json /origin_data/gmall/db/activity_rule_full/$do_date
  ;;
"base_category1")
  import_data /opt/module/datax/job/import/gmall.base_category1.json /origin_data/gmall/db/base_category1_full/$do_date
  ;;
"base_category2")
  import_data /opt/module/datax/job/import/gmall.base_category2.json /origin_data/gmall/db/base_category2_full/$do_date
  ;;
"base_category3")
  import_data /opt/module/datax/job/import/gmall.base_category3.json /origin_data/gmall/db/base_category3_full/$do_date
  ;;
"base_dic")
  import_data /opt/module/datax/job/import/gmall.base_dic.json /origin_data/gmall/db/base_dic_full/$do_date
  ;;
"base_province")
  import_data /opt/module/datax/job/import/gmall.base_province.json /origin_data/gmall/db/base_province_full/$do_date
  ;;
"base_region")
  import_data /opt/module/datax/job/import/gmall.base_region.json /origin_data/gmall/db/base_region_full/$do_date
  ;;
"base_trademark")
  import_data /opt/module/datax/job/import/gmall.base_trademark.json /origin_data/gmall/db/base_trademark_full/$do_date
  ;;
"cart_info")
  import_data /opt/module/datax/job/import/gmall.cart_info.json /origin_data/gmall/db/cart_info_full/$do_date
  ;;
"coupon_info")
  import_data /opt/module/datax/job/import/gmall.coupon_info.json /origin_data/gmall/db/coupon_info_full/$do_date
  ;;
"sku_attr_value")
  import_data /opt/module/datax/job/import/gmall.sku_attr_value.json /origin_data/gmall/db/sku_attr_value_full/$do_date
  ;;
"sku_info")
  import_data /opt/module/datax/job/import/gmall.sku_info.json /origin_data/gmall/db/sku_info_full/$do_date
  ;;
"sku_sale_attr_value")
  import_data /opt/module/datax/job/import/gmall.sku_sale_attr_value.json /origin_data/gmall/db/sku_sale_attr_value_full/$do_date
  ;;
"spu_info")
  import_data /opt/module/datax/job/import/gmall.spu_info.json /origin_data/gmall/db/spu_info_full/$do_date
  ;;
"all")
  import_data /opt/module/datax/job/import/gmall.activity_info.json /origin_data/gmall/db/activity_info_full/$do_date
  import_data /opt/module/datax/job/import/gmall.activity_rule.json /origin_data/gmall/db/activity_rule_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_category1.json /origin_data/gmall/db/base_category1_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_category2.json /origin_data/gmall/db/base_category2_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_category3.json /origin_data/gmall/db/base_category3_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_dic.json /origin_data/gmall/db/base_dic_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_province.json /origin_data/gmall/db/base_province_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_region.json /origin_data/gmall/db/base_region_full/$do_date
  import_data /opt/module/datax/job/import/gmall.base_trademark.json /origin_data/gmall/db/base_trademark_full/$do_date
  import_data /opt/module/datax/job/import/gmall.cart_info.json /origin_data/gmall/db/cart_info_full/$do_date
  import_data /opt/module/datax/job/import/gmall.coupon_info.json /origin_data/gmall/db/coupon_info_full/$do_date
  import_data /opt/module/datax/job/import/gmall.sku_attr_value.json /origin_data/gmall/db/sku_attr_value_full/$do_date
  import_data /opt/module/datax/job/import/gmall.sku_info.json /origin_data/gmall/db/sku_info_full/$do_date
  import_data /opt/module/datax/job/import/gmall.sku_sale_attr_value.json /origin_data/gmall/db/sku_sale_attr_value_full/$do_date
  import_data /opt/module/datax/job/import/gmall.spu_info.json /origin_data/gmall/db/spu_info_full/$do_date
  ;;
esac

（2）为mysql_to_hdfs_full.sh增加执行权限

chmod 777 ~/bin/mysql_to_hdfs_full.sh 

（3）测试同步脚本

mysql_to_hdfs_full.sh all 2020-06-14

（4）检查同步结果
查看HDFS目表路径是否出现全量表数据，全量表共15张。

2.2.6 增量表数据同步

1. Data channel

2. Flume configuration
(1) Flume configuration overview
Flume needs to transfer the data of the topic_db topic in Kafka to HDFS, so it needs to use KafkaSource and HDFSSink, and use FileChannel as the channel.
It should be noted that HDFS Sink needs to write the data of different mysql business tables to different paths, and the path should contain a layer of date to distinguish the data of each day. The key configuration is as follows:

the specific data example is as follows:

(2) Flume configuration practice
(a) Create a Flume configuration file
Create kafka_to_hdfs_db.conf in the job directory of Flume on the hadoop104 node.
The configuration content is as follows

a1.sources = r1
a1.channels = c1
a1.sinks = k1

a1.sources.r1.type = org.apache.flume.source.kafka.KafkaSource
a1.sources.r1.batchSize = 5000
a1.sources.r1.batchDurationMillis = 2000
a1.sources.r1.kafka.bootstrap.servers = hadoop102:9092,hadoop103:9092
a1.sources.r1.kafka.topics = topic_db
a1.sources.r1.kafka.consumer.group.id = flume
a1.sources.r1.setTopicHeader = true
a1.sources.r1.topicHeader = topic
a1.sources.r1.interceptors = i1
a1.sources.r1.interceptors.i1.type = com.zhm.gmall.flume.interceptor.TimestampAndTableNameInterceptor$Builder

a1.channels.c1.type = file
a1.channels.c1.checkpointDir = /opt/module/flume/checkpoint/behavior2
a1.channels.c1.dataDirs = /opt/module/flume/data/behavior2/
a1.channels.c1.maxFileSize = 2146435071
a1.channels.c1.capacity = 1000000
a1.channels.c1.keep-alive = 6

## sink1
a1.sinks.k1.type = hdfs
a1.sinks.k1.hdfs.path = /origin_data/gmall/db/%{
    
    tableName}_inc/%Y-%m-%d
a1.sinks.k1.hdfs.filePrefix = db
a1.sinks.k1.hdfs.round = false


a1.sinks.k1.hdfs.rollInterval = 10
a1.sinks.k1.hdfs.rollSize = 134217728
a1.sinks.k1.hdfs.rollCount = 0


a1.sinks.k1.hdfs.fileType = CompressedStream
a1.sinks.k1.hdfs.codeC = gzip

## 拼装
a1.sources.r1.channels = c1
a1.sinks.k1.channel= c1

(b) Write an interceptor
Create a new Maven project and add the following configuration to the pom.xml file

<dependencies>
    <dependency>
        <groupId>org.apache.flume</groupId>
        <artifactId>flume-ng-core</artifactId>
        <version>1.9.0</version>
        <scope>provided</scope>
    </dependency>

    <dependency>
        <groupId>com.alibaba</groupId>
        <artifactId>fastjson</artifactId>
        <version>1.2.62</version>
    </dependency>
</dependencies>

<build>
    <plugins>
        <plugin>
            <artifactId>maven-compiler-plugin</artifactId>
            <version>2.3.2</version>
            <configuration>
                <source>1.8</source>
                <target>1.8</target>
            </configuration>
        </plugin>
        <plugin>
            <artifactId>maven-assembly-plugin</artifactId>
            <configuration>
                <descriptorRefs>
                    <descriptorRef>jar-with-dependencies</descriptorRef>
                </descriptorRefs>
            </configuration>
            <executions>
                <execution>
                    <id>make-assembly</id>
                    <phase>package</phase>
                    <goals>
                        <goal>single</goal>
                    </goals>
                </execution>
            </executions>
        </plugin>
    </plugins>
</build>

Create the TimestampAndTableNameInterceptor class under the com.zhm.gmall.flume.interceptor package

package com.zhm.gmall.flume.interceptor;

import com.alibaba.fastjson.JSONObject;
import org.apache.flume.Context;
import org.apache.flume.Event;
import org.apache.flume.interceptor.Interceptor;

import java.nio.charset.StandardCharsets;
import java.util.List;
import java.util.Map;

public class TimestampAndTableNameInterceptor implements Interceptor {
    
    
    @Override
    public void initialize() {
    
    

    }

    @Override
    public Event intercept(Event event) {
    
    

        Map<String, String> headers = event.getHeaders();
String log = new String(event.getBody(), StandardCharsets.UTF_8);

 		JSONObject jsonObject = JSONObject.parseObject(log);

 		Long ts = jsonObject.getLong("ts");
 		//Maxwell输出的数据中的ts字段时间戳单位为秒，Flume HDFSSink要求单位为毫秒
 		String timeMills = String.valueOf(ts * 1000);

 		String tableName = jsonObject.getString("table");

 		headers.put("timestamp", timeMills);
 		headers.put("tableName", tableName);
		return event;

    }

    @Override
    public List<Event> intercept(List<Event> events) {
    
    

        for (Event event : events) {
    
    
            intercept(event);
        }

        return events;
    }

    @Override
    public void close() {
    
    

    }

    public static class Builder implements Interceptor.Builder {
    
    


        @Override
        public Interceptor build() {
    
    
            return new TimestampAndTableNameInterceptor ();
        }

        @Override
        public void configure(Context context) {
    
    

        }
    }
}

重新打包。
将打好的包放入到hadoop104的/opt/module/flume/lib文件夹下
（3）通道测试
（a）启动Zookeeper、Kafka集群
（b）启动hadoop104的Flume
（c）生成模拟数据
（d）观察HDFS上的目标路径是否有数据出现
若HDFS上的目标路径已有增量表的数据出现了，就证明数据通道已经打通。
（e）数据目标路径的日期说明
仔细观察，会发现目标路径中的日期，并非模拟数据的业务日期，而是当前日期。这是由于Maxwell输出的JSON字符串中的ts字段的值，是数据的变动日期。而真实场景下，数据的业务日期与变动日期应当是一致的。
（4）编写Flume启停脚本
在hadoop102节点的/home/atguigu/bin目录下创建脚本f3.sh
填写以下内容

#!/bin/bash

case $1 in
"start")
        echo " --------启动 hadoop104 业务数据flume-------"
        ssh hadoop104 "nohup /opt/module/flume/bin/flume-ng agent -n a1 -c /opt/module/flume/conf -f /opt/module/flume/job/kafka_to_hdfs_db.conf >/dev/null 2>&1 &"
;;
"stop")

        echo " --------停止 hadoop104 业务数据flume-------"
        ssh hadoop104 "ps -ef | grep kafka_to_hdfs_db | grep -v grep |awk '{print \$2}' | xargs -n1 kill"
;;
esac

增加脚本执行权限

chmod 777 f3.sh

3、MaxWell配置
1、Maxwell时间戳问题

修改Maxwell配置文件config.properties，增加mock_date参数，如下

log_level=info

producer=kafka
kafka.bootstrap.servers=hadoop102:9092,hadoop103:9092

#kafka topic配置
kafka_topic=topic_db

#注：该参数仅在maxwell教学版中存在，修改该参数后重启Maxwell才可生效
mock_date=2020-06-14

# mysql login info
host=hadoop102
user=maxwell
password=maxwell
jdbc_options=useSSL=false&serverTimezone=Asia/Shanghai

重启Maxwell
重新生成模拟数据

4、增量表首日全量同步
通常情况下，增量表需要在首日进行一次全量同步，后续每日再进行增量同步，首日全量同步可以使用Maxwell的bootstrap功能，方便起见，下面编写一个增量表首日全量同步脚本。
（1）在~/bin目录创建mysql_to_kafka_inc_init.sh
脚本内容如下

#!/bin/bash

# 该脚本的作用是初始化所有的增量表，只需执行一次

MAXWELL_HOME=/opt/module/maxwell

import_data() {
    
    
    $MAXWELL_HOME/bin/maxwell-bootstrap --database gmall --table $1 --config $MAXWELL_HOME/config.properties
}

case $1 in
"cart_info")
  import_data cart_info
  ;;
"comment_info")
  import_data comment_info
  ;;
"coupon_use")
  import_data coupon_use
  ;;
"favor_info")
  import_data favor_info
  ;;
"order_detail")
  import_data order_detail
  ;;
"order_detail_activity")
  import_data order_detail_activity
  ;;
"order_detail_coupon")
  import_data order_detail_coupon
  ;;
"order_info")
  import_data order_info
  ;;
"order_refund_info")
  import_data order_refund_info
  ;;
"order_status_log")
  import_data order_status_log
  ;;
"payment_info")
  import_data payment_info
  ;;
"refund_payment")
  import_data refund_payment
  ;;
"user_info")
  import_data user_info
  ;;
"all")
  import_data cart_info
  import_data comment_info
  import_data coupon_use
  import_data favor_info
  import_data order_detail
  import_data order_detail_activity
  import_data order_detail_coupon
  import_data order_info
  import_data order_refund_info
  import_data order_status_log
  import_data payment_info
  import_data refund_payment
  import_data user_info
  ;;
esac

(2) Add execution permission to mysql_to_kafka_inc_init.sh

chmod 777 ~/bin/mysql_to_kafka_inc_init.sh

(3) Test synchronization script
(a) Clean up historical data

hadoop fs -ls /origin_data/gmall/db | grep _inc | awk '{print $8}' | xargs hadoop fs -rm -r -f

(b) Execute the synchronization script

mysql_to_kafka_inc_init.sh all 

(4) Check the synchronization result
and observe whether the incremental table data reappears on HDFS.

2.3 Acquisition channel start/stop script

1. Create a script cluster.sh in the /home/atguigu/bin directory
and fill in the following content in the script

#!/bin/bash

case $1 in
"start"){
    
    
        echo ================== 启动 集群 ==================

        #启动 Zookeeper集群
        zk.sh start

        #启动 Hadoop集群
        hdp.sh start

        #启动 Kafka采集集群
        kf.sh start

        #启动采集 Flume
        f1.sh start

#启动日志消费 Flume
        f2.sh start

#启动业务消费 Flume
        f3.sh start

#启动 maxwell
        mxw.sh start

        };;
"stop"){
    
    
        echo ================== 停止 集群 ==================

#停止 Maxwell
        mxw.sh stop

#停止 业务消费Flume
        f3.sh stop

#停止 日志消费Flume
        f2.sh stop

#停止 日志采集Flume
        f1.sh stop

        #停止 Kafka采集集群
        kf.sh stop

        #停止 Hadoop集群
        hdp.sh stop

        #停止 Zookeeper集群
        zk.sh stop

};;
esac

2. Add script execution permission

chmod 777 cluster.sh

3. Data warehouse environment preparation

Hive installation and deployment

Just install and configure Hive