如何搭建Hadoop集群

今天决定重新搭建Hadoop集群，如果是新手看这篇文章一定不要觉得很难，你只需要耐心和细心就ok了

准备工作

1.判断机器上是否有ssh服务，

[linuxidc @ www.linuxidc.com Desktop]$ ssh -verison
OpenSSH_5.3p1, OpenSSL 1.0.0-fips 29 Mar 2010
Bad escape character 'rison'.

我的系统自带的，所以不用装了。

2.判断机器上是否有JDK

[linuxidc @ www.linuxidc.com Desktop]$ java -version
java version "1.6.0_24"
OpenJDK Runtime Environment (IcedTea6 1.11.1) (rhel-1.45.1.11.1.el6-i386)
OpenJDK Server VM (build 20.0-b12, mixed mode)
[linuxidc @ www.linuxidc.com Desktop]$ javac -version
javac 1.6.0_24

如果是系统自带的JDK，最好重装一下。 http://www.linuxidc.com/Linux/2012-08/67185.htm

进入主题

1.下载和安装hadoop，我下载的是hadoop-0.20.2.tar.gz

解压文件：[root@ www.linuxidc.com Downloads]# tar -zxvf hadoop-0.20.2.tar.gz

移动文件：[root@ www.linuxidc.com Downloads]# mv hadoop-0.20.2  /usr/local/

安装文件：[root@ www.linuxidc.com Downloads]# ln -s hadoop-0.20.2   hadoop

2.修改环境变量

[root@ www.linuxidc.com local]#vi /etc/profile

在文件的下面添加，不能直接在文件的上面添加

export HADOOP_HOME=/usr/local/hadoop
export PATH=$PATH:$HADOOP_HOME/bin

[root@ www.linuxidc.com local]#.  /etc/profile

[root@ www.linuxidc.com local]# vi /usr/local/hadoop/conf/hadoop.env.sh(配置JAVA_HOME)

[root@ www.linuxidc.com Desktop]# hadoop version
Hadoop 0.20.2
Subversion https://svn.apache.org/repos/asf/hadoop/common/branches/branch-0.20 -r 911707
Compiled by chrisdo on Fri Feb 19 08:07:34 UTC 2010

压轴大戏

1.NameNode配置

[hadoop@hadoop1 ~]# vi /etc/hosts
192.168.127.145  hadoop1
192.168.127.146  hadoop2
192.168.127.147  hadoop3
192.168.127.148  hadoop4

[root@ www.linuxidc.com conf]# vi core-site.xml

1. <configuration>  
2.    <property>  
3.      <name>fs.default.name</name>  
4.      <value>hdfs://hadoop1:9000</value>  
5.    </property>  
6. </configuration>  

[root@ www.linuxidc.com conf]# vi hdfs-site.xml 

1. <configuration>  
2.     <property>  
3.        <name>dfs.replication</name>  
4.        <value>3</value>  
5.     </property>  
6.   
7.     <property>  
8.        <name>dfs.name.dir</name>  
9.        <value>/usr/local/hadoop/namenode/</value>  
10.     </property>  
11.   
12.     <property>  
13.        <name>hadoop.tmp.dir</name>  
14.        <value>/usr/local/hadoop/tmp/</value>  
15.     </property>  
16. </configuration>  

[root@ www.linuxidc.com conf]# vi mapred-site.xml

1. <configuration>  
2.     <property>  
3.         <name>mapred.job.tracker</name>  
4.         <value>hadoop1:9001</value>  
5.     </property>  
6.   
7.     <property>  
8.        <name>mapred.tasktracker.map.tasks.maximum</name>  
9.        <value>4</value>  
10.     </property>  
11.   
12.     <property>  
13.        <name>mapred.tasktracker.reduce.tasks.maximum</name>  
14.        <value>4</value>  
15.     </property>  
16. </configuration>

datanode配置 （只需修改hdfs-site.xml，mapred-site.xml 和core-site.xml跟NameNode一样 )
[hadoop@hadoop2 ~]$ vi hdfs-site.xml

1.  <configuration>  
2. <property>  
3.        <name>dfs.replication</name>  
4.        <value>3</value>  
5.     </property>  
6.   
7.     <property>  
8.        <name>dfs.data.dir</name>  
9.        <value>/home/hadoop/data</value>  
10.     </property>  
11.   
12.     <property>  
13.        <name>hadoop.tmp.dir</name>  
14.        <value>/usr/local/hadoop/tmp/</value>  
15.     </property>  
16. </configuration>  

[hadoop@hadoop1 conf]$ vi masters
hadoop1
[hadoop@hadoop1 conf]$ vi slaves
hadoop2
hadoop3
hadoop4
[hadoop@hadoop1 ~]$ start-all.sh
[hadoop@hadoop1 ~]$ stop-all.sh