大数据环境配置分布式安装hdoop

 Hadoop完全分布式安装教程

 

 

目录

一、软件版本....................................................... 2

二、安装教程....................................................... 2

1、VMWare安装教程.............................................. 2

2、Ubuntu安装教程.............................................. 2

3、安装VMWare-Tools............................................ 5

4、用户创建..................................................... 8

5、主机配置..................................................... 8

6、SSH无密码验证配置........................................... 9

7、Java环境配置................................................ 9

8、hadoop集群安装............................................. 10

三、运行wordcount程序............................................ 20

 

 

 

 

 

 

 

 

 

 

 

 


 

一、软件版本

Hadoop版本号:hadoop-2.6.0.tar;

VMWare版本号:VMware-workstation-full-11.0.0-2305329

Ubuntu版本号:ubuntu-14.04.1-desktop-i386其他版本也可

Jdk版本号:jdk-6u45-linux-i586.bin

后三项对版本要求不严格,如果使用Hbase1.0.0版本,需要JDK1.8以上版本。

 

二、安装教程

1、VMWare安装教程

       VMWare虚拟机是个软件,安装后可用来创建虚拟机,在虚拟机上再安装系统,在这个虚拟系统上再安装应用软件,所有应用就像操作一台真正的电脑,

请直接到VMWare官方网站下载相关软件

http://www.vmware.com/cn/products/workstation/workstation-evaluation

       以上链接如果因为官方网站变动发生变化,可以直接在搜索引擎中搜索VMWare来查找其下载地址,建议不要在非官方网站下载。

       安装试用版后有30天的试用期。

 

2、Ubuntu安装教程

打开VMWare点击创建新的虚拟机

选择典型

点击浏览


选择ubuntu

 

  暂时只建两个虚拟机,注意分别给两个虚拟机起名为Ubuntu1和Ubuntu2;也可以按照自己的习惯取名,但是后续的许多配置文件要相应更改,会带来一些麻烦。

  密码也请记牢,后面会经常使用。

 

3、安装VMWare-Tools

  

Ubuntu中会显示有光盘插入了光驱

 

双击打开光盘将光盘中VMwareTools-9.6.1-1378637.tar.gz复制到桌面,复制方法类似windows系统操作。

 

点击Extract Here

 

从菜单打开Ubuntu的控制终端

 

cdDesktop/vmware-tools-distrib/

sudo./vmware-install.pl

输入root密码,一路回车,重启系统

原理

编辑

sudo1980年前后被写出之前,一般用户管理系统的方式是利用su切换为超级用户。但是使用su的缺点之一在于必须要先告知超级用户的密码。

sudo使一般用户不需要知道超级用户的密码即可获得权限。首先超级用户将普通用户的名字、可以执行的特定命令、按照哪种用户或用户组的身份执行等信息,登记在特殊的文件中(通常是/etc/sudoers),即完成对该用户的授权(此时该用户称为“sudoer”);在一般用户需要取得特殊权限时,其可在命令前加上“sudo”,此时sudo将会询问该用户自己的密码(以确认终端机前的是该用户本人),回答后系统即会将该命令的进程以超级用户的权限运行。之后的一段时间内(默认为5分钟,可在/etc/sudoers自定义),使用sudo不需要再次输入密码。

由于不需要超级用户的密码,部分Unix系统甚至利用sudo使一般用户取代超级用户作为管理帐号,例如UbuntuMac OS X等。

 

 

注意: ubuntu安装后,root 用户默认是被锁定了的,不允许登录,也不允许“ su” 到root 。

允许 su 到root

 

非常简单,下面是设置的方法:

 

注意:ubuntu安装后要更新软件源:

cd /etc/apt

sudo apt-get update

安装各种软件比较方便

 

4、用户创建

创建hadoop用户组:sudo addgroup hadoop 

   创建hduser用户:sudoadduser -ingroup hadoop hduser

   注意这里为hduser用户设置同主用户相同的密码

   为hadoop用户添加权限:sudogedit /etc/sudoers,在root ALL=(ALL) ALL下添加

hduser ALL=(ALL) ALL。

执行命令报错切换到目录编辑

设置好后重启机器:sudo reboot

 

切换到hduser用户登录;

 

5、主机配置

Hadoop集群中包括2个节点:1个Master,2个Salve,其中虚拟机Ubuntu1既做Master,也做Slave;虚拟机Ubuntu2只做Slave。

   配置hostname:Ubuntu下修改机器名称:sudo gedit /etc/hostname ,改为Ubuntu1;修改成功后用重启命令:hostname,查看当前主机名是否设置成功;

 

此时可以用虚拟机克隆的方式再复制一个。(先关机 vmware 菜单--虚拟机-管理--克隆)

注意:修改克隆的主机名为Ubuntu2。

  

   配置hosts文件:查看Ubuntu1和Ubuntu2的ip:ifconfig;

   打开hosts文件:sudogedit /etc/hosts

,添加如下内容:

  192.168.xxx.xxx Ubuntu1

  192.168.xxx.xxx Ubuntu2

 注意这里的ip地址需要学员根据自己的电脑的ip设置。

 在Ubuntu1上执行命令:pingUbuntu2,若能ping通,则说明执行正确。

$$$$$$$$$$$$配置ssh连接linux 速度快

sudo vi /etc/ssh/sshd_config

2.sshd_config配置文件末尾中添加:

[java] view plain copy

1.     Ciphers aes128-cbc,aes192-cbc,aes256-cbc,aes128-ctr,aes192-ctr,aes256-ctr,3des-cbc,arcfour128,arcfour256,arcfour,blowfish-cbc,cast128-cbc  

2.     MACs hmac-md5,hmac-sha1,umac-64@openssh.com,hmac-ripemd160,hmac-sha1-96,hmac-md5-96  

3.     KexAlgorithms diffie-hellman-group1-sha1,diffie-hellman-group14-sha1,diffie-hellman-group-exchange-sha1,diffie-hellman-group-exchange-sha256,ecdh-sha2-nistp256,ecdh-sha2-nistp384,ecdh-sha2-nistp521,diffie-hellman-group1-sha1,curve25519-sha256@libssh.org  

Ciphersaes128-cbc,aes192-cbc,aes256-cbc,aes128-ctr,aes192-ctr,aes256-ctr,3des-cbc,arcfour128,arcfour256,arcfour,blowfish-cbc,cast128-cbc
MACshmac-md5,hmac-sha1,[email protected],hmac-ripemd160,hmac-sha1-96,hmac-md5-96
KexAlgorithms diffie-hellman-group1-sha1,diffie-hellman-group14-sha1,diffie-hellman-group-exchange-sha1,diffie-hellman-group-exchange-sha256,ecdh-sha2-nistp256,ecdh-sha2-nistp384,ecdh-sha2-nistp521,diffie-hellman-group1-sha1,[email protected]

3.重启sshd服务后,即可正常连接:

[java] view plain copy

1.     sudo /etc/init.d/ssh restart  

sudo /etc/init.d/ssh restart

$$$$$$$$$$$$配置ssh连接linux 速度快

 

6、SSH无密码验证配置

   安装ssh服务器,默认安装了ssh客户端:sudoapt-get install openssh-server;

   在Ubuntu1上生成公钥和秘钥:ssh-keygen-t rsa -P "" ;

   查看路径/home/hduser/.ssh文件里是否有id_rsa和id_rsa.pub;
   将公钥赋给authorized_keys:cat$HOME/hduser/.ssh/id_rsa.pub >> $HOME/hduser/.ssh/authorized_keys;

cat id_rsa.pub >> authorized_keys

   无密码登录:sshlocalhost;

   无密码登陆到Ubuntu2,在Ubuntu1上执行:ssh-copy-idUbuntu2,查看Ubuntu2的/home/hduser/.ssh文件里是否有authorized_keys;

   在Ubuntu1上执行命令:sshUbuntu2,首次登陆需要输入密码,再次登陆则无需密码;

   若要使Ubuntu2无密码登录Ubuntu1,则在Ubutu2上执行上述相同操作即可。

注:若无密码登录设置不成功,则很有可能是文件夹/文件权限问题,修改文件夹/文件权限即可。sudochmod 777 “文件夹” 即可。

root@Ubuntu1:/home/hduser#

root@Ubuntu1:/home/hduser# ssh Ubuntu2--不能用root用户连接ubuntu2无权限

root@ubuntu2's password:

Permission denied, please try again.

 

 

 

7、Java环境配置

获取opt文件夹权限:sudo chmod 777 /opt

将java压缩包放在/opt/,root模式执行sudo./jdk-6u45-linux-i586.bin

配置jdk的环境变量:sudo gedit /etc/profile,将一下内容复制进去并保存

   # java

   exportJAVA_HOME=/opt/jdk1.6.0_45

   exportJRE_HOME=$JAVA_HOME/jre

   exportCLASSPATH=$JAVA_HOME/lib:$JRE_HOME/lib:$CLASSPATH

   exportPATH=$JAVA_HOME/bin:$JRE_HOME/bin:$PATH

  

   执行命令,使配置生效:source/etc/profile;

   执行命令:java-version,若出现java版本号,则说明安装成功。

sudo apt-get install openjdk-7-jdk-headless

 

http://openjdk.java.net/install/

8、hadoop集群安装

8.1 安装

将hadoop压缩包hadoop-2.6.0.tar.gz放在/home/hduser目录下,并解压缩到本地,重命名为hadoop;配置hadoop环境变量,执行:sudogedit /etc/profile,将以下复制到profile内:

    #hadoop

exportHADOOP_HOME=/home/hduser/hadoop   

exportPATH=$HADOOP_HOME/bin:$PATH

 

执行:source /etc/profile

 

注意:Ubuntu1、ubuntu2都要配置以上步骤;

8.2 配置

主要涉及的配置文件有7个:都在/hadoop/etc/hadoop文件夹下,可以用gedit命令对其进行编辑。

(1)进去hadoop配置文件目录

cd /home/hduser/hadoop/etc/hadoop/



(2)配置 hadoop-env.sh文件-->修改JAVA_HOME

gedit hadoop-env.sh

添加如下内容

# The java implementation to use.

export JAVA_HOME=/usr/lib/jvm/java-7-openjdk-i386/

#JAVA_HOME=/usr/lib/jvm/java-7-openjdk-i386

(3)配置 yarn-env.sh 文件-->>修改JAVA_HOME

添加如下内容

# some Java parameters

exportJAVA_HOME=/opt/jdk1.6.0_45

(4)配置slaves文件-->>增加slave节点 

(删除原来的localhost)

添加如下内容

Ubuntu1

Ubuntu2

(5)配置 core-site.xml文件-->>增加hadoop核心配置

(hdfs文件端口是9000、file:/home/hduser/hadoop/tmp)

添加如下内容


 
  fs.defaultFS
  hdfs://Ubuntu1:9000
 

 
  io.file.buffer.size
  131072
 

 
  hadoop.tmp.dir
  file:/home/hduser/hadoop/tmp
  Abasefor other temporarydirectories.
 

 hadoop.native.lib
  true
  Should native hadoop libraries, if present, beused.

(6)配置  hdfs-site.xml 文件-->>增加hdfs配置信息

(namenode、datanode端口和目录位置)


 
  dfs.namenode.secondary.http-address
  Ubuntu1:9001
 


 
   dfs.namenode.name.dir
   file:/home/hduser/hadoop/dfs/name
 


 
  dfs.datanode.data.dir
  file:/home/hduser/hadoop/dfs/data
 


 
  dfs.replication
  2
 


 
  dfs.webhdfs.enabled
  true
 

(7)配置  mapred-site.xml 文件-->>增加mapreduce配置

(使用yarn框架、jobhistory使用地址以及web地址)


 
   mapreduce.framework.name
   yarn
 

 
  mapreduce.jobhistory.address
  Ubuntu1:10020
 

 
  mapreduce.jobhistory.webapp.address
  Ubuntu1:19888
 

(8)配置   yarn-site.xml  文件-->>增加yarn功能


 
   yarn.nodemanager.aux-services
   mapreduce_shuffle
 

 
  yarn.nodemanager.aux-services.mapreduce.shuffle.class
   org.apache.hadoop.mapred.ShuffleHandler
 

 
   yarn.resourcemanager.address
   Ubuntu1:8032
 

 
   yarn.resourcemanager.scheduler.address
   Ubuntu1:8030
 

 
  yarn.resourcemanager.resource-tracker.address
   Ubuntu1:8035
 

 
   yarn.resourcemanager.admin.address
   Ubuntu1:8033
 

 
   yarn.resourcemanager.webapp.address
   Ubuntu1:8088
 


(9)将配置好的Ubuntu1中/hadoop/etc/hadoop文件夹复制到到Ubuntu2对应位置(删除Ubuntu2原来的文件夹/hadoop/etc/hadoop)

scp-r /home/hduser/hadoop/etc/hadoop/ hduser@Ubuntu2:/home/hduser/hadoop/etc/

8.3 验证

下面验证Hadoop配置是否正确:

(1)格式化namenode:

hduser@Ubuntu1:~$ cd hadoop

hduser@Ubuntu1:~/hadoop$ ./bin/hdfs namenode -format

hduser@Ubuntu2:~$ cd hadoop

hduser@Ubuntu2:~/hadoop$ ./bin/hdfs namenode -format

(2)启动hdfs:

hduser@Ubuntu1:~/hadoop$ ./sbin/start-dfs.sh

15/04/2704:18:45 WARN util.NativeCodeLoader: Unable to load native-hadoop library foryour platform... using builtin-java classes where applicable

Startingnamenodes on [Ubuntu1]

Ubuntu1:starting namenode, logging to/home/hduser/hadoop/logs/hadoop-hduser-namenode-Ubuntu1.out

Ubuntu1:starting datanode, logging to /home/hduser/hadoop/logs/hadoop-hduser-datanode-Ubuntu1.out

Ubuntu2:starting datanode, logging to/home/hduser/hadoop/logs/hadoop-hduser-datanode-Ubuntu2.out

Startingsecondary namenodes [Ubuntu1]

Ubuntu1:starting secondarynamenode, logging to /home/hduser/hadoop/logs/hadoop-hduser-secondarynamenode-Ubuntu1.out

15/04/2704:19:07 WARN util.NativeCodeLoader: Unable to load native-hadoop library foryour platform... using builtin-java classes where applicable

 

查看java进程(Java Virtual Machine Process Status Tool)

hduser@Ubuntu1:~/hadoop$ jps

8008 NameNode

8443 Jps

8158 DataNode

8314SecondaryNameNode

使用 jps发现NameNode进程没有正确运行,

停止服务,

重新格式化namenode,hadoop namenode  -format

start-all.sh

NameNode进程已运行

 

 

(3)停止hdfs:

hduser@Ubuntu1:~/hadoop$ ./sbin/stop-dfs.sh

Stoppingnamenodes on [Ubuntu1]

Ubuntu1:stopping namenode

Ubuntu1:stopping datanode

Ubuntu2:stopping datanode

Stoppingsecondary namenodes [Ubuntu1]

Ubuntu1:stopping secondarynamenode

查看java进程

hduser@Ubuntu1:~/hadoop$ jps

8850 Jps

(4)启动yarn:

hduser@Ubuntu1:~/hadoop$ ./sbin/start-yarn.sh

starting yarndaemons

startingresourcemanager, logging to/home/hduser/hadoop/logs/yarn-hduser-resourcemanager-Ubuntu1.out

Ubuntu2:starting nodemanager, logging to/home/hduser/hadoop/logs/yarn-hduser-nodemanager-Ubuntu2.out

Ubuntu1:starting nodemanager, logging to/home/hduser/hadoop/logs/yarn-hduser-nodemanager-Ubuntu1.out

 

查看java进程

hduser@Ubuntu1:~/hadoop$jps

8911ResourceManager

9247 Jps

9034NodeManager

(5)停止yarn:

hduser@Ubuntu1:~/hadoop$  ./sbin/stop-yarn.sh

stopping yarndaemons

stoppingresourcemanager

Ubuntu1:stopping nodemanager

Ubuntu2:stopping nodemanager

no proxyserverto stop

查看java进程

hduser@Ubuntu1:~/hadoop$jps

9542 Jps

(6)查看集群状态:

首先启动集群:./sbin/start-dfs.sh

hduser@Ubuntu1:~/hadoop$./bin/hdfs dfsadmin -report

ConfiguredCapacity: 39891361792 (37.15 GB)

Present Capacity:28707627008 (26.74 GB)

DFS Remaining: 28707569664(26.74 GB)

DFS Used: 57344(56 KB)

DFS Used%: 0.00%

Under replicatedblocks: 0

Blocks withcorrupt replicas: 0

Missing blocks: 0

 

-------------------------------------------------

Live datanodes(2):

 

Name:192.168.159.132:50010 (Ubuntu2)

Hostname: Ubuntu2

DecommissionStatus : Normal

ConfiguredCapacity: 19945680896 (18.58 GB)

DFS Used: 28672(28 KB)

Non DFS Used:5575745536 (5.19 GB)

DFS Remaining:14369906688 (13.38 GB)

DFS Used%: 0.00%

DFS Remaining%:72.05%

Configured CacheCapacity: 0 (0 B)

Cache Used: 0 (0B)

Cache Remaining: 0(0 B)

Cache Used%:100.00%

Cache Remaining%:0.00%

Xceivers: 1

Last contact: MonApr 27 04:26:09 PDT 2015

 

Name:192.168.159.131:50010 (Ubuntu1)

Hostname: Ubuntu1

DecommissionStatus : Normal

ConfiguredCapacity: 19945680896 (18.58 GB)

DFS Used: 28672(28 KB)

Non DFS Used:5607989248 (5.22 GB)

DFS Remaining:14337662976 (13.35 GB)

DFS Used%: 0.00%

DFS Remaining%:71.88%

Configured CacheCapacity: 0 (0 B)

Cache Used: 0 (0B)

Cache Remaining: 0(0 B)

Cache Used%:100.00%

Cache Remaining%:0.00%

Xceivers: 1

Last contact: MonApr 27 04:26:08 PDT 2015

(7)查看hdfs:http://Ubuntu1:50070/

 

三、运行wordcount程序

(1)创建 file目录

hduser@Ubuntu1:~$ mkdir file

(2)在file创建file1.txt、file2.txt并写内容(在图形界面)

分别填写如下内容

file1.txt输入内容:Hello world hiHADOOP

file2.txt输入内容:Hello hadoop hiCHINA

创建后查看:

hduser@Ubuntu1:~ /hadoop $ cat file/file1.txt

Hello world hiHADOOP

hduser@Ubuntu1:~ /hadoop $ cat file/file2.txt

Hello hadoop hiCHINA

(3)在hdfs创建/input2目录

hduser@Ubuntu1:~/hadoop$ ./bin/hadoop fs -mkdir/input2

 

bin/hdfs dfs-mkdir -p /input

(4)将file1.txt、file2.txt文件copy到hdfs /input2目录

hduser@Ubuntu1:~/hadoop$ ./bin/hadoop fs -putfile/file*.txt /input2

(5)查看hdfs上是否有file1.txt、file2.txt文件

hduser@Ubuntu1:~/hadoop$ bin/hadoop fs -ls /input2/

 

 bin/hdfs dfs-put ../file/file*.txt /input2

 

Found 2 items

-rw-r--r--   2 hduser supergroup         21 2015-04-27 05:54 /input2/file1.txt

-rw-r--r--   2 hduser supergroup         24 2015-04-27 05:54 /input2/file2.txt

(6)执行wordcount程序

先启动hdfs和yarn(注意jar包名)

 

hduser@Ubuntu1:~/hadoop$ ./bin/hadoop jarshare/hadoop/mapreduce/hadoop-mapreduce-examples-2.6.5.jar wordcount /input2//output2/wordcount1

15/04/27 05:57:17 WARN util.NativeCodeLoader: Unable to load native-hadooplibrary for your platform... using builtin-java classes where applicable

15/04/27 05:57:17 INFO client.RMProxy: Connecting to ResourceManager atUbuntu1/192.168.159.131:8032

15/04/27 05:57:19 INFO input.FileInputFormat: Total input paths to process: 2

15/04/27 05:57:19 INFO mapreduce.JobSubmitter: number of splits:2

15/04/27 05:57:19 INFO mapreduce.JobSubmitter: Submitting tokens for job:job_1430138907536_0001

15/04/27 05:57:20 INFO impl.YarnClientImpl: Submitted applicationapplication_1430138907536_0001

15/04/27 05:57:20 INFO mapreduce.Job: The url to track the job:http://Ubuntu1:8088/proxy/application_1430138907536_0001/

15/04/27 05:57:20 INFO mapreduce.Job: Running job: job_1430138907536_0001

15/04/27 05:57:32 INFO mapreduce.Job: Job job_1430138907536_0001 runningin uber mode : false

15/04/27 05:57:32 INFO mapreduce.Job: map 0% reduce 0%

15/04/27 05:57:43 INFO mapreduce.Job: map 100% reduce 0%

15/04/27 05:57:58 INFO mapreduce.Job: map 100% reduce 100%

15/04/27 05:57:59 INFO mapreduce.Job: Job job_1430138907536_0001 completedsuccessfully

15/04/27 05:57:59 INFO mapreduce.Job: Counters: 49

       File System Counters

              FILE: Number of bytesread=84

              FILE: Number of byteswritten=317849

              FILE: Number of readoperations=0

              FILE: Number of largeread operations=0

              FILE: Number of writeoperations=0

              HDFS: Number of bytesread=247

              HDFS: Number of byteswritten=37

              HDFS: Number of readoperations=9

              HDFS: Number of largeread operations=0

              HDFS: Number of writeoperations=2

       Job Counters

              Launched map tasks=2

              Launched reduce tasks=1

              Data-local map tasks=2

              Total time spent by allmaps in occupied slots (ms)=16813

              Total time spent by allreduces in occupied slots (ms)=12443

              Total time spent by allmap tasks (ms)=16813

              Total time spent by allreduce tasks (ms)=12443

              Total vcore-secondstaken by all map tasks=16813

              Total vcore-secondstaken by all reduce tasks=12443

              Total megabyte-secondstaken by all map tasks=17216512

              Total megabyte-secondstaken by all reduce tasks=12741632

       Map-Reduce Framework

              Map input records=2

              Map output records=8

              Map output bytes=75

              Map output materializedbytes=90

              Input split bytes=202

              Combine input records=8

              Combine outputrecords=7

              Reduce input groups=5

              Reduce shuffle bytes=90

              Reduce input records=7

              Reduce output records=5

              Spilled Records=14

              Shuffled Maps =2

              Failed Shuffles=0

              Merged Map outputs=2

              GC time elapsed(ms)=622

              CPU time spent(ms)=2000

              Physical memory (bytes)snapshot=390164480

              Virtual memory (bytes)snapshot=1179254784

              Total committed heapusage (bytes)=257892352

       Shuffle Errors

              BAD_ID=0

              CONNECTION=0

              IO_ERROR=0

              WRONG_LENGTH=0

              WRONG_MAP=0

              WRONG_REDUCE=0

       File Input Format Counters

              Bytes Read=45

       File Output Format Counters

              Bytes Written=37

 

(7)查看运行结果

hduser@Ubuntu1:~/hadoop$ ./bin/hdfs dfs -cat /output2/wordcount1/*

CHINA   1

Hello      2

hadoop    2

hi         2

world      1

 

 

 

 

——————————————

显示出以上结果,表明您已经成功安装了Hadoop!

 

 

 3、环境变量的添加:

echo "export PATH=:./:$PATH:/usr/local/eclipse" >> /etc/profile

echo "export PATH=:./:$PATH:/usr/local/eclipse" >> ~/.bashrc

 

source /et/profile

source ~/.bashrc

Eclipse开发环境的建立

1,  需要下载eclipse

2,  需要插件,插件的终极解决方案是

https://github.com/winghc/hadoop2x-eclipse-plugin下载并编译。

也可用提供好的插件。

3,  复制编译好的jar到eclipse插件目录,重启eclipse

4,  配置hadoop 安装目录

window ->preference -> hadoopMap/Reduce -> Hadoop installation directory

 

5,      配置Map/Reduce 视图

window ->Open Perspective ->other->Map/Reduce -> 点击“OK”

windows → show view →other->Map/Reduce Locations-> 点击“OK”

 

6,在“Map/Reduce Locations”Tab页点击图标<大象+>或者在空白的地方右键,选择“New Hadoop location…”,弹出对话框“New hadoop location…”,

进行相应配置

MR Master和DFS Master配置必须和mapred-site.xml和core-site.xml等配置文件一致

7,打开Project Explorer,查看HDFS文件系统。

8,新建Map/Reduce任务

需要先启动Hadoop服务

File->New->project->Map/ReduceProject->Next

编写WordCount类:

import java.io.IOException;

import java.util.StringTokenizer;

 

import org.apache.hadoop.conf.Configuration;

import org.apache.hadoop.fs.Path;

import org.apache.hadoop.io.IntWritable;

import org.apache.hadoop.io.Text;

import org.apache.hadoop.mapreduce.Job;

import org.apache.hadoop.mapreduce.Mapper;

import org.apache.hadoop.mapreduce.Reducer;

import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;

import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

 

public class WordCount {

 

  public static class TokenizerMapper

       extends Mapper{

 

    private final static IntWritableone = new IntWritable(1);

    private Text word = new Text();

public void map(Object key, Text value, Contextcontext) throws IOException, InterruptedException {

// Object key,Text value就是输入的key和value, Context记录输入的key和value

      StringTokenizer itr = newStringTokenizer(value.toString());

      while (itr.hasMoreTokens()) {

        word.set(itr.nextToken());

        context.write(word, one);

      }

    }

  }

 

  public static class IntSumReducer

       extendsReducer {

    private IntWritable result = newIntWritable();

 

    public void reduce(Text key,Iterable values,

                       Context context

                       ) throwsIOException, InterruptedException {

//reduce函数与map函数基本相同,但value是一个迭代器的形式Iterablevalues,也就是说reduce的输入是一个key对应一组的值的value

      int sum = 0;

      for (IntWritable val : values){

        sum += val.get();

      }

      result.set(sum);

      context.write(key, result);//结果例如World, 2

    }

  }

 

  public static void main(String[]args) throws Exception {

    Configuration conf = newConfiguration();

    Job job = Job.getInstance(conf,"word count");//指定job名称,及运行对象 

   job.setJarByClass(WordCount.class);       job.setMapperClass(TokenizerMapper.class); //指定map函数

   job.setCombinerClass(IntSumReducer.class); // combiner整合

   job.setReducerClass(IntSumReducer.class);//设定reduce函数

    job.setOutputKeyClass(Text.class);//设定输出key数据类型

   job.setOutputValueClass(IntWritable.class);//设定输出value数据类型

    FileInputFormat.addInputPath(job,new Path(args[0]));//设定输入目录

   FileOutputFormat.setOutputPath(job, new Path(args[1]));

   System.exit(job.waitForCompletion(true) ? 0 : 1);

  }


音乐记录倒排索引

MapReduce程序开发

1、  我们的任务要求是:

有一批音乐播放记录清单,包含歌曲被播放的用户

tom                            LittleApple              

jack                              YesterdayOnceMore  

Rose                            MyHeartWillGoOn     

jack                              LittleApple            

John                            MyHeartWillGoOn     

kissinger                     LittleApple            

kissinger                     YesterdayOnceMore

2、  我们的任务输出结果是:

完成一个倒排索引形成的文本文件如下

LittleApple                         tom| jack| kissinger

YesterdayOnceMore          jack|kissinger

MyHeartWillGoOn             Rose|John

 

 

3、  我们的算法思路是:

将源文件按照每行进行分割,在mapper 过程中以歌曲名(LittleApple)作为key,以用户名(Tom)作为value,在reducer过程中是相同个歌曲码汇总,输出为倒排索引。

tom                            LittleApple              

jack                              YesterdayOnceMore  

Rose                            MyHeartWillGoOn

Map函数对应的

< YesterdayOnceMore, Jack >

< MyHeartWillGoOn, Rose>

Reduce函数将歌曲汇总

输出是

LittleApple      tom

                            Jack

Kissinger

最终输出到HDFS为结果

LittleApple                         tom| jack| kissinger

YesterdayOnceMore          jack|kissinger

MyHeartWillGoOn             Rose|John

 

 

4、  倒排索引源程序的注释:

importjava.io.IOException;

importorg.apache.hadoop.conf.Configuration;

importorg.apache.hadoop.conf.Configured;

importorg.apache.hadoop.fs.Path;

importorg.apache.hadoop.io.*;

importorg.apache.hadoop.mapreduce.*;

importorg.apache.hadoop.mapreduce.lib.input.FileInputFormat;

importorg.apache.hadoop.mapreduce.lib.output.FileOutputFormat;

importorg.apache.hadoop.mapreduce.lib.output.TextOutputFormat;

importorg.apache.hadoop.util.Tool;

importorg.apache.hadoop.util.ToolRunner;

 

 

publicclass Test_1 extends Configured implements Tool

{

 

 

  enum Counter

  {

    LINESKIP, // 出错的行

  }

 

 

  public static class Map extendsMapper

  {

    public void map(LongWritable key, Textvalue, Context context) throws IOException, InterruptedException

    {

    Stringline = value.toString(); // 读取源数据,将其字符串化

         try

         {                  

       // 数据处理

           String[] lineSplit = line.split("");

//将数据用空格进行分割,例如Tom  LittleApple 

           String anum = lineSplit[0]; //此处anum为Tom

           String bnum = lineSplit[1]; //此处bnum为 LittleApple

           context.write(new Text(bnum), newText(anum));

// 输出到context的键值对为

          }

         catch(java.lang.ArrayIndexOutOfBoundsException e)  //出错保障

         {

          context.getCounter(Counter.LINESKIP).increment(1);

           return;

         }

     }

   }

 

 

   public static class Reduce extendsReducer

   {

      public void reduce(Text key,Iterable values, Context context) throws IOException,InterruptedException

      {

        String valueString;

        String out = "";

       

        for (Text value : values)

        {

          valueString = value.toString();

          out += valueString +"|";  //将听同一歌曲用|分隔符隔开累加

         //System.out.println("Ruduce:key="+key+"  value="+value);

        }

        context.write(key, new Text(out));

      }

   }

 

 

  @Override

   public int run(String[] args) throwsException

   {

     Configuration conf = this.getConf();

    

     Job job = new Job(conf,"Test_1"); // 任务名

     job.setJarByClass(Test_1.class); // 指定Class

 

     FileInputFormat.addInputPath(job, new Path(args[0]));// 输入路径

     FileOutputFormat.setOutputPath(job, newPath(args[1])); // 输出路径

 

     job.setMapperClass(Map.class); // 调用上面Map类作为Map任务代码

     job.setReducerClass(Reduce.class); // 调用上面Reduce类作为Reduce任务代码

     job.setOutputFormatClass(TextOutputFormat.class);

     job.setOutputKeyClass(Text.class); // 指定输出的KEY的格式

     job.setOutputValueClass(Text.class); // 指定输出的VALUE的格式

 

     job.waitForCompletion(true);

 

     return job.isSuccessful()?0:1;

    }

 

 

    public static void main(String[] args) throwsException

    {

      // 运行任务

      int res = ToolRunner.run(newConfiguration(), new Test_1(), args);

      System.exit(res);

    }

}

 

5、  注意设置输入输出的路径:

可以在eclipse上直接运行,也可打成jar包后运行。

 

你可能感兴趣的:(大数据)