淘先锋技术网

首页 1 2 3 4 5 6 7

提示:文章内容代码部分地方需要根据自身的环境路径进行修改!

目录

前言

一.Flume汇入数据到Hive

方法一:汇入到Hive指定的HDFS路径中:

1.在hive中创建数据库和外部表

2.在/root中创建hive.log文件 

 3.在flume的conf路径中编写配置文件

4.运行flume

 5.查询hdfs中的数据

 6.在hive表中加载数据

7.查询hive表中的数据

 方法二:利用HiveSink汇入数据

1.从hive/lib和和hive/hcatalog/share/hcatalog/中找寻下列JAR包,放入到flume/lib中。如果flume中有重名的则先删除flume中的再进行复制。

 2.编写flume的配置文件

 3.在hive中创建表

 4.在hive中设置权限

 5.启动metastore服务

6.运行flume

 7.查询数据

二.Flume 汇入数据到HBase

一、Flume 的HBaseSinks 详细介绍

二.HBaseSink

 三.HBaseSinks的三种序列化模式使用

1.SimpleHbaseEventSerializer

2.SimpleAsyncHbaseEventSerializer

​编辑 3.RegexHbaseEventSerializer

总结


前言

提示:这里可以添加本文要记录的大概内容:

主要讲述Flume汇入数据到Hive&Hbase的方法及操作。


提示:以下是本篇文章正文内容,下面案例可供参考

一.Flume汇入数据到Hive

方法一:汇入到Hive指定的HDFS路径中:

1.在hive中创建数据库和外部

代码:

create database flume;

代码:

create external table flume_into_hive(name string,age int) partitioned by (dt string) row format delimited fields terminated by ',' location '/user/hive/warehouse/flume.db/flume_into_hive';

2.在/root中创建hive.log文件 

代码:

mkdir flume-hive
cd flume-hive/
vi hive.log

 

 3.在flume的conf路径中编写配置文件

代码:

cd /opt/software/flume/conf/
vi flume-into-hive-1.conf

代码:

 

agent.sources=r1
agent.channels=c1
agent.sinks=s1

agent.sources.r1.type=exec
agent.sources.r1.command=tail -F /root/flume-hive/hive.log

agent.channels.c1.type=memory
agent.channels.capacity=1000
agent.channels.c1.transactionCapacity=100

agent.sinks.s1.type=hdfs
agent.sinks.s1.hdfs.path = hdfs://node01:9000/user/hive/warehouse/flume.db/flume_into_hive/dt=%Y%m%d
agent.sinks.s1.hdfs.filePrefix = upload-
agent.sinks.s1.hdfs.fileSuffix=.txt
#是否按照时间滚动文件夹
agent.sinks.s1.hdfs.round = true
#多少时间单位创建一个新的文件夹
agent.sinks.s1.hdfs.roundValue = 1
#重新定义时间单位
agent.sinks.s1.hdfs.roundUnit = hour
#是否使用本地时间戳
agent.sinks.s1.hdfs.useLocalTimeStamp = true
#积攒多少个 Event 才 flush 到 HDFS 一次
agent.sinks.s1.hdfs.batchSize = 100
#设置文件类型,可支持压缩
agent.sinks.s1.hdfs.fileType = DataStream
agent.sinks.s1.hdfs.writeFormat=Text
#多久生成一个新的文件
agent.sinks.s1.hdfs.rollInterval = 60
#设置每个文件的滚动大小大概是 128M
agent.sinks.s1.hdfs.rollSize = 134217700
#文件的滚动与 Event 数量无关
agent.sinks.s1.hdfs.rollCount = 0

agent.sources.r1.channels=c1
agent.sinks.s1.channel=c1

4.运行flume

代码:

bin/flume-ng agent -c conf -f conf/hive/flume-into-hive-1.conf -n agent

 5.查询hdfs中的数据

 6.在hive表中加载数据

代码:

load data inpath '/user/hive/warehouse/flume.db/flume_into_hive/dt=20221109' into table flume_into_hive partition(dt=20221109);

7.查询hive表中的数据

select * from flume_into_hive;

 方法二:利用HiveSink汇入数据

1.hive/lib和hive/hcatalog/share/hcatalog/中找寻下列JAR包,放入到flume/lib中。如果flume中有重名的则先删除flume中的再进行复制。

 1.检查flume中的JAR包:

 删除这两个重复jar包:

 2.hive/libhive/hcatalog/share/hcatalog/中找寻JAR包,放入到flume/lib中

 

在/root中创建hive.log文件

代码:

cd flume-hive/
vi hive.log

 

 

 

 2.编写flume的配置文件

 代码:

vi flume-into-hive-2.conf
a1.sources = s1
a1.channels = c1
a1.sinks = k1

a1.sources.s1.type=exec
a1.sources.s1.command=tail -F /root/flume-hive/hive.log

a1.sinks.k1.type = hive
a1.sinks.k1.channel=c1
a1.sinks.k1.hive.metastore = thrift://node01:9083
a1.sinks.k1.hive.database = flume
a1.sinks.k1.hive.table = flume_into_hive_1
a1.sinks.k1.useLocalTimeStamp = true
a1.sinks.k1.round = false
a1.sinks.k1.roundValue = 10
a1.sinks.k1.roundUnit = minute
a1.sinks.k1.serializer = DELIMITED
a1.sinks.k1.serializer.fieldnames =name,age

a1.channels.c1.type=memory
a1.channels.c1.capacity=1000
a1.channels.c1.transactionCapacity=100

a1.sinks.k1.channel = c1
a1.sources.s1.channels = c1

 3.在hive中创建表

代码:

create table flume_into_hive_1(name string,age int) clustered by (age) into 2 buckets stored as orc tblproperties("transactional"='true');

 4.在hive中设置权限

代码:

set hive.support.concurrency=true;
set hive.txn.manager=org.apache.hadoop.hive.ql.lockmgr.DbTxnManager;

 5.启动metastore服务

 代码:

hive --service metastore -p 9083

6.运行flume

 代码:

bin/flume-ng agent -c conf -f conf/flume-into-hive-2.conf -n a1

 

 7.查询数据

 

二.Flume 汇入数据到HBase

一、Flume 的HBaseSinks 详细介绍

Flume 有两大类 HBasesinks: HBaseSink (org.apache.flume.sink.hbase.HBaseSink) 和 AsyncHBaseSink (org.apache.flume.sink.hbase.AsyncHBaseSink) 。

二.HBaseSink

HBaseSink提供两种序列化模式:SimpleHbaseEventSerializer和RegexHbaseEventSerializer。

 三.HBaseSinks的三种序列化模式使用

1.SimpleHbaseEventSerializer

首先在HBase里面建立一个表flume-hbase-table,拥有colfamily1colfamily2两个列族

代码:

create 'flume-hbase-table','colfamily1','colfamily2'

 然后写一个flume的配置文件flume-into-hbase.conf

代码:

 

agent.sources = r1
agent.channels = c1
agent.sinks = s1

agent.sources.r1.type = exec
agent.sources.r1.command = tail -F /root/flume-hbase/test.log
agent.sources.r1.checkperiodic = 50

agent.channels.c1.type = memory
agent.channels.c1.capacity = 1000
agent.channels.c1.transactionCapacity = 100


agent.sinks.s1.type = org.apache.flume.sink.hbase.HBaseSink
agent.sinks.s1.zookeeperQuorum=node01:2181

agent.sinks.s1.table=flume-hbase-table
#HBase表的列族名称
agent.sinks.s1.columnFamily=colfamily1
agent.sinks.s1.serializer = org.apache.flume.sink.hbase.SimpleHbaseEventSerializer
#HBase表的列族下的某个列名称
agent.sinks.s1.serializer.payloadColumn=column-1

agent.sources.r1.channels = c1
agent.sinks.s1.channel=c1

 运行Flume:

bin/flume-ng agent -c conf -f conf/hbase/flume-into-hbase.conf -n agent  -Dflume.root.logger=INFO,console

 数据:

 

 代码:

scan 'flume-hbase-table'

2.SimpleAsyncHbaseEventSerializer

 编写flume-into-hbase-1.conf配置文件:

 代码:

agent.sources = r1
agent.channels = c1
agent.sinks = s1

agent.sources.r1.type = exec
agent.sources.r1.command = tail -F /root/flume-hbase/test.log
agent.sources.r1.checkperiodic = 50

agent.channels.c1.type = memory
agent.channels.c1.capacity = 1000
agent.channels.c1.transactionCapacity = 100


agent.sinks.s1.type = org.apache.flume.sink.hbase.AsyncHBaseSink
agent.sinks.s1.zookeeperQuorum=node01:2181

agent.sinks.s1.table=flume-hbase-table
#HBase表的列族名称
agent.sinks.s1.columnFamily=colfamily2
agent.sinks.s1.serializer = org.apache.flume.sink.hbase.SimpleAsyncHbaseEventSerializer
#HBase表的列族下的某个列名称
agent.sinks.s1.serializer.payloadColumn=column-2

agent.sources.r1.channels = c1
agent.sinks.s1.channel=c1

 运行flume:

bin/flume-ng agent -c conf -f conf/hbase/flume-into-hbase-1.conf -n agent  -Dflume.root.logger=INFO,console

在hbase中查看:

 3.RegexHbaseEventSerializer

编写flume-into-hbase-2.conf配置文件:

 代码;

agent.sources = r1
agent.channels = c1
agent.sinks = s1

agent.sources.r1.type = exec
agent.sources.r1.command = tail -F /root/flume-hbase/test.log
agent.sources.r1.checkperiodic = 50

agent.channels.c1.type = memory
agent.channels.c1.capacity = 1000
agent.channels.c1.transactionCapacity = 100

agent.sinks.s1.type = org.apache.flume.sink.hbase.HBaseSink
agent.sinks.s1.zookeeperQuorum=node01:2181

agent.sinks.s1.table=flume-hbase-table
#HBase表的列族名称
agent.sinks.s1.columnFamily=colfamily1
agent.sinks.s1.serializer = org.apache.flume.sink.hbase.RegexHbaseEventSerializer
agent.sinks.s1.serializer.regex=\\[(.*?)\\]\\ \\[(.*?)\\]\\ \\[(.*?)\\]
agent.sinks.s1.serializer.colNames=time,url,number

agent.sources.r1.channels = c1
agent.sinks.s1.channel=c1

 运行Flume:

bin/flume-ng agent -c conf -f conf/hbase/flume-into-hbase-2.conf -n agent  -Dflume.root.logger=INFO,console

在/root/flume-hbase/test.log中添加如下数据:

 

[2022-05-17] [http://www.baidu.com] [20]
[2022-05-17] [http://www.bilibili.com] [25]
[2022-05-17] [http://www.qq.com] [26]

 

 查看hbase的flume-hbase-table:


总结

提示:这里对文章进行总结:
    以上就是今天要讲的内容,本文主要讲述Flume汇入数据到Hive&Hbase的方法及操作.