1.hadoop的伪分布安装（CentOS）

1.hadoop的伪分布安装

1.1 设置ip地址

    执行命令 service network restart

    验证: ifconfig

1.2 关闭防火墙

    执行命令 service iptables stop

    验证: service iptables status

1.3 关闭防火墙的自动运行

    执行命令 chkconfig iptables off

    验证: chkconfig --list | grep iptables

1.4 设置主机名

    执行命令

        (1)hostname hadoop

        (2)vi /etc/sysconfig/network

1.5 ip与hostname绑定

    执行命令 vi /etc/hosts

    验证: ping hadoop

1.6 设置ssh免密码登陆

    执行命令 

        (1)ssh-keygen -t rsa

        (2)cp ~/.ssh/id_rsa.pub ~/.ssh/authorized_keys

    验证： ssh hadoop

1.7 安装jdk

    执行命令

        (1)cd /usr/local

        (2)chmod u+x jdk-6u24-linux-i586.bin                

        (3)./jdk-6u24-linux-i586.bin                

        (4)mv jdk-1.6.0_24  jdk                

        (5)vi /etc/profile 增加内容如下:                

            export JAVA_HOME=/usr/local/jdk

            export PATH=.:$JAVA_HOME/bin:$PATH                

        (6)source /etc/profile                

    验证: java -version

1.8 安装hadoop

    执行命令

        (1)tar -zxvf hadoop-1.1.2.tar.gz

        (2)mv hadoop-1.1.2 hadoop

        (3)vi /etc/profile 增加内容如下:

            export JAVA_HOME=/usr/local/jdk

            export HADOOP_HOME=/usr/local/hadoop

           export PATH=.:$HADOOP_HOME/bin:$JAVA_HOME/bin:$PATH

     (4)source /etc/profile

     (5)修改conf目录下的配置文件hadoop-env.sh、core-site.xml、hdfs-site.xml、mapred-site.xml

            1.hadoop-env.sh  export JAVA_HOME=/usr/local/jdk/

            2.core-site.xml 

    <configuration>
        <property>
            <name>fs.default.name</name>
            <value>hdfs://hadoop:9000</value>
        </property>
        <property>
            <name>hadoop.tmp.dir</name>
            <value>/usr/local/hadoop/tmp</value>
        </property>
    </configuration>

            3.hdfs-site.xml

    <configuration>
        <property>
            <name>dfs.replication</name>
            <value>1</value>
        </property>
        <property>
             <name>dfs.permissions</name>
             <value>false</value>
        </property>
    </configuration>

        4.mapred-site.xml

    <configuration>    
        <property>
            <name>mapred.job.tracker</name>
            <value>hadoop:9001</value>
        </property>
    </configuration>

    (6)hadoop namenode -format

    (7)start-all.sh

        验证:

            (1)执行命令jps 如果看到5个新的java进程，分别是NameNode、SecondaryNameNode、DataNode、JobTracker、TaskTracker

            (2)在浏览器查看，http://hadoop:50070  http://hadoop:50030  

1.9 启动时没有NameNode的可能原因:

     (1)没有格式化

     (2)环境变量设置错误

     (3)ip与hostname绑定失败

转载于:https://my.oschina.net/iduanyingjie/blog/204967