hadoop 2.3.0 搭建指南

一. 准备工作(搭建环境：Ubuntu 12.04)

假设我们有三台机器。一台做master，另外两台做slave。所有的操作都先只在master上操作，然后scp同步到slaves。

1、创建hadoop用户,建立master到两台slaves的信任关系（如何建立，可以百度一下）

[plain] view plain copy print ?

adduser hadoop
usermod hadoop -G sudo -a # 添加到超级管理员用户组

adduser hadoop
usermod hadoop -G sudo -a       # 添加到超级管理员用户组

2、下载&解压 hadoop 2.3.0 压缩包

到这里找一个镜像，然后下载 hadoop 2.3.0 压缩包：

http://www.apache.org/dyn/closer.cgi/hadoop/common/

这里我们解压到 /home/hadoop/hadoop

目录结构如下：

[plain] view plain copy print ?

hadoop@master:~/hadoop$ pwd
/home/hadoop/hadoop
hadoop@master:~/hadoop$ ls
bin etc include lib libexec LICENSE.txt logs NOTICE.txt README.txt sbin share

hadoop@master:~/hadoop$ pwd
/home/hadoop/hadoop
hadoop@master:~/hadoop$ ls
bin  etc  include  lib  libexec  LICENSE.txt  logs  NOTICE.txt  README.txt  sbin  share

3、下载jdk

直接到这里下载 jdk：

http://www.oracle.com/technetwork/java/javase/downloads/jdk7-downloads-1880260.html

挑选对应自己操作系统的版本，只需要解压到任意目录，不需要执行任何操作。

这里我们解压到/usr/local/jdk

[plain] view plain copy print ?

hadoop@master:/usr/local/jdk$ pwd
/usr/local/jdk
hadoop@master:/usr/local/jdk$ ls
bin db jre LICENSE README.html src.zip THIRDPARTYLICENSEREADME.txt
COPYRIGHT include lib man release THIRDPARTYLICENSEREADME-JAVAFX.txt

hadoop@master:/usr/local/jdk$ pwd
/usr/local/jdk
hadoop@master:/usr/local/jdk$ ls
bin        db       jre  LICENSE  README.html  src.zip                             THIRDPARTYLICENSEREADME.txt
COPYRIGHT  include  lib  man      release      THIRDPARTYLICENSEREADME-JAVAFX.txt

4、修改各种配置文件

/etc/hostname:

master 上的这个文件就写master，两个slaves上的这个文件分别写 node1、 node2

/etc/hosts(ip地址根据自己的情况修改)

[plain] view plain copy print ?

127.0.0.1 localhost
192.168.204.128 master
192.168.204.129 node1
192.168.204.130 node2

127.0.0.1       localhost

192.168.204.128 master
192.168.204.129 node1
192.168.204.130 node2

/etc/profile(在最后面添加):

[plain] view plain copy print ?

#hadoop
export JAVA_HOME=/usr/local/jdk
export PATH=$JAVA_HOME/bin:$JAVA_HOME/jre/bin:$PATH
export CLASSPATH=.:$JAVA_HOME/lib:$JAVA_HOME/jre/lib
export HADOOP_ROOT=/home/hadoop
export HADOOP_HOME=$HADOOP_ROOT/hadoop
export PATH=$HADOOP_ROOT/hadoop/bin:$HADOOP_ROOT/hadoop/sbin:$PATH

#hadoop
export JAVA_HOME=/usr/local/jdk
export PATH=$JAVA_HOME/bin:$JAVA_HOME/jre/bin:$PATH
export CLASSPATH=.:$JAVA_HOME/lib:$JAVA_HOME/jre/lib
export HADOOP_ROOT=/home/hadoop
export HADOOP_HOME=$HADOOP_ROOT/hadoop
export PATH=$HADOOP_ROOT/hadoop/bin:$HADOOP_ROOT/hadoop/sbin:$PATH

打开hadoop配置目录（/home/hadoop/hadoop/etc/hadoop)

core-site.xml:

[plain] view plain copy print ?

<configuration>
<property>
<name>fs.defaultFS</name>
<value>hdfs://master</value>
</property>
</configuration>

<configuration>
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://master</value>
    </property>
</configuration>

hdfs-site.xml:

[plain] view plain copy print ?

<configuration>
<property>
<name>dfs.namenode.name.dir</name>
<value>/data1/hadoop-nn</value>
</property>
<property>
<name>fs.defaultFS</name>
<value>hdfs://master/</value>
</property>
<property>
<name>dfs.replication</name>
<value>2</value>
</property>
<property>
<name>dfs.hosts.exclude</name>
<value>/home/hadoop/hadoop/etc/hadoop/dfs.exclude</value>
</property>
<property>
<name>dfs.datanode.data.dir</name>
<value>/data1/hadoop-dn,/data2/hadoop-dn</value>
</property>
</configuration>

<configuration>
<property>
    <name>dfs.namenode.name.dir</name>
    <value>/data1/hadoop-nn</value>
</property>
<property>
  <name>fs.defaultFS</name>
  <value>hdfs://master/</value>
</property>
<property>
  <name>dfs.replication</name>
  <value>2</value>
</property>
<property>
   <name>dfs.hosts.exclude</name>
   <value>/home/hadoop/hadoop/etc/hadoop/dfs.exclude</value>
</property>
<property>
    <name>dfs.datanode.data.dir</name>
    <value>/data1/hadoop-dn,/data2/hadoop-dn</value>
</property>

</configuration>

从上面的配置文件可以看出，我们把namenode的数据存储目录定位 /data1/hadoop-nn

datanode的存储目录定位为 /data1/hadoop-dn, /data2/hadoop-dn

所以要保证这两个目录正常存在，并且有空闲硬盘挂在上面。

slaves:

[plain] view plain copy print ?

node1
node2

node1
node2

5、同步各种配置文件到slaves上

二. 启动hdfs

1、在namenode上执行：

[plain] view plain copy print ?

hadoop-daemon.sh --script hdfs start namenode

hadoop-daemon.sh  --script hdfs start namenode

2、在两个datanode上分别执行：

[plain] view plain copy print ?

hadoop-daemon.sh --script hdfs start datanode

hadoop-daemon.sh  --script hdfs start datanode

三.测试hdfs

在任意一台server上执行

[plain] view plain copy print ?

hadoop fs -ls /

hadoop fs -ls /

[plain] view plain copy print ?

hadoop fs -put ./test /

hadoop fs -put ./test /

hadoop新老版本下载URL：http://archive.apache.org/dist/hadoop/core/