RHadoop安装与使用

标签: 统计之都 hadoop MapReduce R rhadoop | 发表时间:2013-03-24 21:59 | 作者:张 丹
出处:http://cos.name

Author:张丹(Conan)
Date: 2013-03-07

Weibo: @Conan_Z
Email: [email protected]
Blog: http://www.fens.me/blog

APPs:
@晒粉丝 http://www.fens.me
@每日中国天气 http://apps.weibo.com/chinaweatherapp

RHadoop实践系列文章

RHadoop实践系列文章,包含了R语言与Hadoop结合进行海量数据分析。Hadoop主要用来存储海量数据,R语言完成MapReduce 算法,用来替代Java的MapReduce实现。有了RHadoop可以让广大的R语言爱好者,有更强大的工具处理大数据。1G, 10G, 100G, TB,PB 由于大数据所带来的单机性能问题,可能会一去联复返了。

RHadoop实践是一套系列文章,主要包括”Hadoop环境搭建”,”RHadoop安装与使用”,”R实现MapReduce的算法案 例”,”HBase和rhbase的安装与使用”。对于单独的R语言爱好者,Java爱好者,或者Hadoop爱好者来说,同时具备三种语言知识并不容 易。此文虽为入门文章,但R,Java,Hadoop基础知识还是需要大家提前掌握。

第二篇 RHadoop安装与使用部分,分为3个章节。

  1. 环境准备
2. RHadoop安装
3. RHadoop程序用例

每一章节,都会分为”文字说明部分”和”代码部分”,保持文字说明与代码的连贯性。

注:Hadoop环境搭建的详细记录,请查看 同系列上一篇文章 “RHadoop实践系列文章之Hadoop环境搭建”。
由于两篇文章并非同一时间所写,hadoop版本及操作系统,分步式环境都略有不同。
两篇文章相互独立,请大家在理解的基础上动手实验,不要完成依赖两篇文章中的运行命令。

环境准备

文字说明部分:

首先环境准备,这里我选择了Linux Ubuntu操作系统12.04的64位版本,大家可以根据自己的使用习惯选择顺手的Linux。

但JDK一定要用Oracle SUN官方的版本,请从官网下载,操作系统的自带的OpenJDK会有各种不兼容。JDK请选择1.6.x的版本,JDK1.7版本也会有各种的不兼容情况。
http://www.oracle.com/technetwork/java/javase/downloads/index.html

Hadoop的环境安装,请参考RHadoop实践系统”Hadoop环境搭建”的一文。

R语言请安装2.15以后的版本,2.14是不能够支持RHadoop的。
如果你也使用Linux Ubuntu操作系统12.04,请先更新软件包源,否则只能下载到2.14版本的R。

代码部分:

1. 操作系统Ubuntu 12.04 x64

  ~ uname -a
Linux domU-00-16-3e-00-00-85 3.2.0-23-generic #36-Ubuntu SMP Tue Apr 10 20:39:51 UTC 2012 x86_64 x86_64 x86_64 GNU/Linux

2 JAVA环境

  ~ java -version

java version "1.6.0_29"
Java(TM) SE Runtime Environment (build 1.6.0_29-b11)
Java HotSpot(TM) 64-Bit Server VM (build 20.4-b02, mixed mode)

3 HADOOP环境(这里只需要hadoop)

  hadoop-1.0.3  hbase-0.94.2  hive-0.9.0  pig-0.10.0  sqoop-1.4.2  thrift-0.8.0  zookeeper-3.4.4

4 R的环境

  R version 2.15.3 (2013-03-01) -- "Security Blanket"
Copyright (C) 2013 The R Foundation for Statistical Computing
ISBN 3-900051-07-0
Platform: x86_64-pc-linux-gnu (64-bit)

4.1 如果是Ubuntu 12.04,请更新源再下载R2.15.3版本

  sh -c "echo deb http://mirror.bjtu.edu.cn/cran/bin/linux/ubuntu precise/ >>/etc/apt/sources.list"
apt-get update
apt-get install r-base

RHadoop安装

文字说明部分:

RHadoop是RevolutionAnalytics的工程的项目,开源实现代码在GitHub社区可以找到。RHadoop包含三个R包 (rmr,rhdfs,rhbase),分别是对应Hadoop系统架构中的,MapReduce, HDFS, HBase 三个部分。由于这三个库不能在CRAN中找到,所以需要自己下载。
https://github.com/RevolutionAnalytics/RHadoop/wiki

接下我们需要先安装这三个库的依赖库。
首先是rJava,上个章节我们已经配置好了JDK1.6的环境,运行R CMD javareconf命令,R的程序从系统变量中会读取Java配置。然后打开R程序,通过install.packages的方式,安装rJava。

然后,我还要安装其他的几个依赖库,reshape2,Rcpp,iterators,itertools,digest,RJSONIO,functional,通过install.packages都可以直接安装。

接下安装rhdfs库,在环境变量中增加 HADOOP_CMD 和 HADOOP_STREAMING 两个变量,可以用export在当前命令窗口中增加。但为下次方便使用,最好把变量增加到系统环境变更/etc/environment文件中。再用 R CMD INSTALL安装rhdfs包,就可以顺利完成了。

安装rmr库,使用R CMD INSTALL也可以顺利完成了。

安装rhbase库,后面”HBase和rhbase的安装与使用”文章中会继续介绍,这里暂时跳过。

最后,我们可以查看一下,RHADOOP都安装了哪些库。
由于我的硬盘是外接的,使用mount和软连接(ln -s)挂载了R类库的目录,所以是R的类库在/disk1/system下面
/disk1/system/usr/local/lib/R/site-library/
一般R的类库目录是/usr/lib/R/site-library或者/usr/local/lib/R/site-library,用户也可以使用whereis R的命令查询,自己电脑上R类库的安装位置

代码部分:

1. 下载RHadoop相关的3个程序包

https://github.com/RevolutionAnalytics/RHadoop/wiki/Downloads

  rmr-2.1.0
rhdfs-1.0.5
rhbase-1.1

2. 复制到/root/R目录

  ~/R# pwd
/root/R

~/R# ls
rhbase_1.1.tar.gz  rhdfs_1.0.5.tar.gz  rmr2_2.1.0.tar.gz

3. 安装依赖库

  命令行执行
~ R CMD javareconf 
~ R

启动R程序
install.packages("rJava")
install.packages("reshape2")
install.packages("Rcpp")
install.packages("iterators")
install.packages("itertools")
install.packages("digest")
install.packages("RJSONIO")
install.packages("functional")

4. 安装rhdfs库

  ~ export HADOOP_CMD=/root/hadoop/hadoop-1.0.3/bin/hadoop
~ export HADOOP_STREAMING=/root/hadoop/hadoop-1.0.3/contrib/streaming/hadoop-streaming-1.0.3.jar (rmr2会用到)
~ R CMD INSTALL /root/R/rhdfs_1.0.5.tar.gz 

4.1 最好把HADOOP_CMD设置到环境变量

  ~ vi /etc/environment

    HADOOP_CMD=/root/hadoop/hadoop-1.0.3/bin/hadoop
    HADOOP_STREAMING=/root/hadoop/hadoop-1.0.3/contrib/streaming/hadoop-streaming-1.0.3.jar

. /etc/environment

5. 安装rmr库

  ~  R CMD INSTALL rmr2_2.1.0.tar.gz 

6. 安装rhbase库 (暂时跳过)

7. 所有的安装包

  ~ ls /disk1/system/usr/local/lib/R/site-library/
digest  functional  iterators  itertools  plyr  Rcpp  reshape2  rhdfs  rJava  RJSONIO  rmr2  stringr

RHadoop程序用例

文字说明部分:

安装好rhdfs和rmr两个包后,我们就可以使用R尝试一些hadoop的操作了。

首先,是基本的hdfs的文件操作。

查看hdfs文件目录
hadoop的命令:hadoop fs -ls /user
R语言函数:hdfs.ls(”/user/“)

查看hadoop数据文件
hadoop的命令:hadoop fs -cat /user/hdfs/o_same_school/part-m-00000
R语言函数:hdfs.cat(”/user/hdfs/o_same_school/part-m-00000″)

接下来,我们执行一个rmr算法的任务

普通的R语言程序:

  > small.ints = 1:10
> sapply(small.ints, function(x) x^2)

MapReduce的R语言程序:

  > small.ints = to.dfs(1:10)
> mapreduce(input = small.ints, map = function(k, v) cbind(v, v^2))
> from.dfs("/tmp/RtmpWnzxl4/file5deb791fcbd5")

因为MapReduce只能访问HDFS文件系统,先要用to.dfs把数据存储到HDFS文件系统里。MapReduce的运算结果再用from.dfs函数从HDFS文件系统中取出。

第二个,rmr的例子是wordcount,对文件中的单词计数

  > input<- '/user/hdfs/o_same_school/part-m-00000'
> wordcount = function(input, output = NULL, pattern = " "){

  wc.map = function(., lines) {
            keyval(unlist( strsplit( x = lines,split = pattern)),1)
    }

    wc.reduce =function(word, counts ) {
            keyval(word, sum(counts))
    }         

    mapreduce(input = input ,output = output, input.format = "text",
        map = wc.map, reduce = wc.reduce,combine = T)
}

> wordcount(input)
> from.dfs("/tmp/RtmpfZUFEa/file6cac626aa4a7")

我在HDFS上提前放置了数据文件/user/hdfs/o_same_school/part-m-00000。写wordcount的MapReduce函数,执行wordcount函数,最后用from.dfs从HDFS中取得结果。

代码部分:

1. rhdfs包的使用

  启动R程序
> library(rhdfs)

Loading required package: rJava
HADOOP_CMD=/root/hadoop/hadoop-1.0.3/bin/hadoop
Be sure to run hdfs.init()

> hdfs.init()

1.1 命令查看hadoop目录

  ~ hadoop fs -ls /user

Found 4 items
drwxr-xr-x   - root supergroup          0 2013-02-01 12:15 /user/conan
drwxr-xr-x   - root supergroup          0 2013-03-06 17:24 /user/hdfs
drwxr-xr-x   - root supergroup          0 2013-02-26 16:51 /user/hive
drwxr-xr-x   - root supergroup          0 2013-03-06 17:21 /user/root

1.2 rhdfs查看hadoop目录

  > hdfs.ls("/user/")

  permission owner      group size          modtime        file
1 drwxr-xr-x  root supergroup    0 2013-02-01 12:15 /user/conan
2 drwxr-xr-x  root supergroup    0 2013-03-06 17:24  /user/hdfs
3 drwxr-xr-x  root supergroup    0 2013-02-26 16:51  /user/hive
4 drwxr-xr-x  root supergroup    0 2013-03-06 17:21  /user/root

1.3 命令查看hadoop数据文件

  ~ hadoop fs -cat /user/hdfs/o_same_school/part-m-00000

10,3,tsinghua university,2004-05-26 15:21:00.0
23,4007,北京第一七一中学,2004-05-31 06:51:53.0
51,4016,大连理工大学,2004-05-27 09:38:31.0
89,4017,Amherst College,2004-06-01 16:18:56.0
92,4017,斯坦福大学,2012-11-28 10:33:25.0
99,4017,Stanford University Graduate School of Business,2013-02-19 12:17:15.0
113,4017,Stanford University,2013-02-19 12:17:15.0
123,4019,St Paul's Co-educational College - Hong Kong,2004-05-27 18:04:17.0
138,4019,香港苏浙小学,2004-05-27 18:59:58.0
172,4020,University,2004-05-27 19:14:34.0
182,4026,ff,2004-05-28 04:42:37.0
183,4026,ff,2004-05-28 04:42:37.0
189,4033,tsinghua,2011-09-14 12:00:38.0
195,4035,ba,2004-05-31 07:10:24.0
196,4035,ma,2004-05-31 07:10:24.0
197,4035,southampton university,2013-01-07 15:35:18.0
246,4067,美国史丹佛大学,2004-06-12 10:42:10.0
254,4067,美国史丹佛大学,2004-06-12 10:42:10.0
255,4067,美国休士顿大学,2004-06-12 10:42:10.0
257,4068,清华大学,2004-06-12 10:42:10.0
258,4068,北京八中,2004-06-12 17:34:02.0
262,4068,香港中文大学,2004-06-12 17:34:02.0
310,4070,首都师范大学初等教育学院,2004-06-14 15:35:52.0
312,4070,北京师范大学经济学院,2004-06-14 15:35:52.0

1.4 rhdfs查看hadoop数据文件

  >  hdfs.cat("/user/hdfs/o_same_school/part-m-00000")

 [1] "10,3,tsinghua university,2004-05-26 15:21:00.0"
 [2] "23,4007,北京第一七一中学,2004-05-31 06:51:53.0"
 [3] "51,4016,大连理工大学,2004-05-27 09:38:31.0"
 [4] "89,4017,Amherst College,2004-06-01 16:18:56.0"
 [5] "92,4017,斯坦福大学,2012-11-28 10:33:25.0"
 [6] "99,4017,Stanford University Graduate School of Business,2013-02-19 12:17:15.0"
 [7] "113,4017,Stanford University,2013-02-19 12:17:15.0"
 [8] "123,4019,St Paul's Co-educational College - Hong Kong,2004-05-27 18:04:17.0"
 [9] "138,4019,香港苏浙小学,2004-05-27 18:59:58.0"
[10] "172,4020,University,2004-05-27 19:14:34.0"
[11] "182,4026,ff,2004-05-28 04:42:37.0"
[12] "183,4026,ff,2004-05-28 04:42:37.0"
[13] "189,4033,tsinghua,2011-09-14 12:00:38.0"
[14] "195,4035,ba,2004-05-31 07:10:24.0"
[15] "196,4035,ma,2004-05-31 07:10:24.0"
[16] "197,4035,southampton university,2013-01-07 15:35:18.0"
[17] "246,4067,美国史丹佛大学,2004-06-12 10:42:10.0"
[18] "254,4067,美国史丹佛大学,2004-06-12 10:42:10.0"
[19] "255,4067,美国休士顿大学,2004-06-12 10:42:10.0"
[20] "257,4068,清华大学,2004-06-12 10:42:10.0"
[21] "258,4068,北京八中,2004-06-12 17:34:02.0"
[22] "262,4068,香港中文大学,2004-06-12 17:34:02.0"
[23] "310,4070,首都师范大学初等教育学院,2004-06-14 15:35:52.0"
[24] "312,4070,北京师范大学经济学院,2004-06-14 15:35:52.0"

2. rmr2包的使用

  启动R程序
> library(rmr2)

Loading required package: Rcpp
Loading required package: RJSONIO
Loading required package: digest
Loading required package: functional
Loading required package: stringr
Loading required package: plyr
Loading required package: reshape2

2.1 执行r任务

  > small.ints = 1:10
> sapply(small.ints, function(x) x^2)

[1]   1   4   9  16  25  36  49  64  81 100

2.2 执行rmr2任务

  > small.ints = to.dfs(1:10)

13/03/07 12:12:55 INFO util.NativeCodeLoader: Loaded the native-hadoop library
13/03/07 12:12:55 INFO zlib.ZlibFactory: Successfully loaded & initialized native-zlib library
13/03/07 12:12:55 INFO compress.CodecPool: Got brand-new compressor

> mapreduce(input = small.ints, map = function(k, v) cbind(v, v^2))

packageJobJar: [/tmp/RtmpWnzxl4/rmr-local-env5deb2b300d03, /tmp/RtmpWnzxl4/rmr-global-env5deb398a522b, /tmp/RtmpWnzxl4/rmr-streaming-map5deb1552172d, /root/hadoop/tmp/hadoop-unjar7838617732558795635/] [] /tmp/streamjob4380275136001813619.jar tmpDir=null
13/03/07 12:12:59 INFO mapred.FileInputFormat: Total input paths to process : 1
13/03/07 12:12:59 INFO streaming.StreamJob: getLocalDirs(): [/root/hadoop/tmp/mapred/local]
13/03/07 12:12:59 INFO streaming.StreamJob: Running job: job_201302261738_0293
13/03/07 12:12:59 INFO streaming.StreamJob: To kill this job, run:
13/03/07 12:12:59 INFO streaming.StreamJob: /disk1/hadoop/hadoop-1.0.3/libexec/../bin/hadoop job  -Dmapred.job.tracker=hdfs://r.qa.tianji.com:9001 -kill job_201302261738_0293
13/03/07 12:12:59 INFO streaming.StreamJob: Tracking URL: http://192.168.1.243:50030/jobdetails.jsp?jobid=job_201302261738_0293
13/03/07 12:13:00 INFO streaming.StreamJob:  map 0%  reduce 0%
13/03/07 12:13:15 INFO streaming.StreamJob:  map 100%  reduce 0%
13/03/07 12:13:21 INFO streaming.StreamJob:  map 100%  reduce 100%
13/03/07 12:13:21 INFO streaming.StreamJob: Job complete: job_201302261738_0293
13/03/07 12:13:21 INFO streaming.StreamJob: Output: /tmp/RtmpWnzxl4/file5deb791fcbd5

> from.dfs("/tmp/RtmpWnzxl4/file5deb791fcbd5")

$key
NULL

$val
       v
 [1,]  1   1
 [2,]  2   4
 [3,]  3   9
 [4,]  4  16
 [5,]  5  25
 [6,]  6  36
 [7,]  7  49
 [8,]  8  64
 [9,]  9  81
[10,] 10 100

2.3 wordcount执行rmr2任务

  > input<- '/user/hdfs/o_same_school/part-m-00000'
> wordcount = function(input, output = NULL, pattern = " "){

    wc.map = function(., lines) {
            keyval(unlist( strsplit( x = lines,split = pattern)),1)
    }

    wc.reduce =function(word, counts ) {
            keyval(word, sum(counts))
    }         

    mapreduce(input = input ,output = output, input.format = "text",
        map = wc.map, reduce = wc.reduce,combine = T)
}

> wordcount(input)

packageJobJar: [/tmp/RtmpfZUFEa/rmr-local-env6cac64020a8f, /tmp/RtmpfZUFEa/rmr-global-env6cac73016df3, /tmp/RtmpfZUFEa/rmr-streaming-map6cac7f145e02, /tmp/RtmpfZUFEa/rmr-streaming-reduce6cac238dbcf, /tmp/RtmpfZUFEa/rmr-streaming-combine6cac2b9098d4, /root/hadoop/tmp/hadoop-unjar6584585621285839347/] [] /tmp/streamjob9195921761644130661.jar tmpDir=null
13/03/07 12:34:41 INFO util.NativeCodeLoader: Loaded the native-hadoop library
13/03/07 12:34:41 WARN snappy.LoadSnappy: Snappy native library not loaded
13/03/07 12:34:41 INFO mapred.FileInputFormat: Total input paths to process : 1
13/03/07 12:34:41 INFO streaming.StreamJob: getLocalDirs(): [/root/hadoop/tmp/mapred/local]
13/03/07 12:34:41 INFO streaming.StreamJob: Running job: job_201302261738_0296
13/03/07 12:34:41 INFO streaming.StreamJob: To kill this job, run:
13/03/07 12:34:41 INFO streaming.StreamJob: /disk1/hadoop/hadoop-1.0.3/libexec/../bin/hadoop job  -Dmapred.job.tracker=hdfs://r.qa.tianji.com:9001 -kill job_201302261738_0296
13/03/07 12:34:41 INFO streaming.StreamJob: Tracking URL: http://192.168.1.243:50030/jobdetails.jsp?jobid=job_201302261738_0296
13/03/07 12:34:42 INFO streaming.StreamJob:  map 0%  reduce 0%
13/03/07 12:34:59 INFO streaming.StreamJob:  map 100%  reduce 0%
13/03/07 12:35:08 INFO streaming.StreamJob:  map 100%  reduce 17%
13/03/07 12:35:14 INFO streaming.StreamJob:  map 100%  reduce 100%
13/03/07 12:35:20 INFO streaming.StreamJob: Job complete: job_201302261738_0296
13/03/07 12:35:20 INFO streaming.StreamJob: Output: /tmp/RtmpfZUFEa/file6cac626aa4a7

> from.dfs("/tmp/RtmpfZUFEa/file6cac626aa4a7")

$key
 [1] "-"
 [2] "04:42:37.0"
 [3] "06:51:53.0"
 [4] "07:10:24.0"
 [5] "09:38:31.0"
 [6] "10:33:25.0"
 [7] "10,3,tsinghua"
 [8] "10:42:10.0"
 [9] "113,4017,Stanford"
[10] "12:00:38.0"
[11] "12:17:15.0"
[12] "123,4019,St"
[13] "138,4019,香港苏浙小学,2004-05-27"
[14] "15:21:00.0"
[15] "15:35:18.0"
[16] "15:35:52.0"
[17] "16:18:56.0"
[18] "172,4020,University,2004-05-27"
[19] "17:34:02.0"
[20] "18:04:17.0"
[21] "182,4026,ff,2004-05-28"
[22] "183,4026,ff,2004-05-28"
[23] "18:59:58.0"
[24] "189,4033,tsinghua,2011-09-14"
[25] "19:14:34.0"
[26] "195,4035,ba,2004-05-31"
[27] "196,4035,ma,2004-05-31"
[28] "197,4035,southampton"
[29] "23,4007,北京第一七一中学,2004-05-31"
[30] "246,4067,美国史丹佛大学,2004-06-12"
[31] "254,4067,美国史丹佛大学,2004-06-12"
[32] "255,4067,美国休士顿大学,2004-06-12"
[33] "257,4068,清华大学,2004-06-12"
[34] "258,4068,北京八中,2004-06-12"
[35] "262,4068,香港中文大学,2004-06-12"
[36] "312,4070,北京师范大学经济学院,2004-06-14"
[37] "51,4016,大连理工大学,2004-05-27"
[38] "89,4017,Amherst"
[39] "92,4017,斯坦福大学,2012-11-28"
[40] "99,4017,Stanford"
[41] "Business,2013-02-19"
[42] "Co-educational"
[43] "College"
[44] "College,2004-06-01"
[45] "Graduate"
[46] "Hong"
[47] "Kong,2004-05-27"
[48] "of"
[49] "Paul's"
[50] "School"
[51] "University"
[52] "university,2004-05-26"
[53] "university,2013-01-07"
[54] "University,2013-02-19"
[55] "310,4070,首都师范大学初等教育学院,2004-06-14"

$val
 [1] 1 2 1 2 1 1 1 4 1 1 2 1 1 1 1 2 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
[39] 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1

相关 [rhadoop] 推荐:

RHadoop安装与使用

- - 统计之都
Author:张丹(Conan). @晒粉丝 http://www.fens.me. @每日中国天气 http://apps.weibo.com/chinaweatherapp. RHadoop实践系列文章. RHadoop实践系列文章,包含了R语言与Hadoop结合进行海量数据分析. Hadoop主要用来存储海量数据,R语言完成MapReduce 算法,用来替代Java的MapReduce实现.

RHadoop实践系列之三 R实现MapReduce的协同过滤算法

- - 统计之都
Author:张丹(Conan). @晒粉丝 http://www.fens.me. @每日中国天气 http://apps.weibo.com/chinaweatherapp. RHadoop实践系列文章. RHadoop实践系列文章,包含了R语言与Hadoop结合进行海量数据分析. Hadoop主要用来存储海量数据,R语言完成MapReduce 算法,用来替代Java的MapReduce实现.