hadoop2.2.0的伪分布安装

软件的安装目录：
     /opt/modules/
   
安装:
     0）说明
          1、系统：CentOS 6.4 64位
        
          2、关闭防火墙和SELinux
               service iptables status
               service iptables stop
               chkconfig iptables off
        
               vi /etc/sysconfig/selinux        
               设置 SELINUX=disabled

          3、设置静态IP地址
               vi /etc/sysconfig/network-scripts/ifcfg-eth0
             
          4、修改HostName
               hostname hadoop-yarn.dragon.org     
               vi /etc/sysconfig/network

          5、IP与HostName绑定
               vi /etc/hosts
               内容：
                    192.168.48.128     hadoop-yarn.dragon.org     hadoop-yarn
          6、安装JDK
               安装目录：/opt/modules/jdk1.6.0_45
               版本：jdk-6u45-linux-x64.bin
               命令：
                    chmod u+x jdk-6u45-linux-x64.bin
                    ./jdk-6u45-linux-x64.bin
               设置环境变量：
                    vi /etc/profile
                    添加内容：
                         export JAVA_HOME=/opt/modules/jdk1.6.0_45
                         export PATH=$PATH:$JAVA_HOME/bin
                    生效：
                         # source /etc/profile
        
     1）解压tar包
          tar -zxvf hadoop-2.2.0.tar.gz -C /opt/modules/
        
     2）安装目录
          HADOOP_HOME=/opt/modules/hadoop-2.2.0
   
     3）配置
          1、hadoop-env.sh
               export JAVA_HOME=/opt/modules/jdk1.6.0_45
        
          2、yarn-env.sh
               export JAVA_HOME=/opt/modules/jdk1.6.0_45
   
          3、mapred-env.sh
               export JAVA_HOME=/opt/modules/jdk1.6.0_45
             
          4、core-site.xml
               <property>        
                    <name>fs.default.name</name>
                    <value>hdfs://hadoop-yarn.dragon.org:8020</value>
               </property>

               <property>
                    <name>hadoop.tmp.dir</name>
                    <value>/opt/modules/hadoop-2.2.0/data/tmp</value>
               </property>
             
          5、hdfs-site.xml
               <property>        
                    <name>dfs.replication</name>
                    <value>1</value>
               </property>
        
          6、yarn-site.xml
               <property>
                    <name>yarn.nodemanager.aux-services</name>
                    <value>mapreduce_shuffle</value>
               </property>
             
          7、mapred-site.xml
               <property>                          
                    <name>mapreduce.framework.name</name>
                    <value>yarn</value>
               </property>
     4）启动
          1、启动HDFS
          NameNode、DataNode、SecondaryNameNode
               * NameNode 格式化
                    bin/hdfs namenode -format          -- 产生一个Cluster ID
                  
                    拓展：
                         * 指定Cluster ID
                         bin/hdfs namenode -format -clusterid yarn-cluster
                       
                         * Block Pool ID
                              数据块池ID
                            
                         * NameNode Fedaration
               * 启动NameNode
                    sbin/hadoop-daemon.sh start namenode
               * 启动DataNode
                    sbin/hadoop-daemon.sh start datanode
               * 启动SecondaryNameNode
                    sbin/hadoop-daemon.sh start secondarynamenode
        
          2、启动YARN
          ResourceManager、NodeManager
               * 启动ResourceManger
                    sbin/yarn-daemon.sh start resourcemanager
               * 启动NodeManager
                    sbin/yarn-daemon.sh start nodemanager
                  
        
     5）日志
          【hadoop】-【hadoop】-【namenode】-【hadoop-yarn.dragon.org】.【log】
        
          框架名称     用户名          服务名称          主机名                       日志格式
   
     6）HDFS 操作错误
          信息：
               [root@hadoop-yarn hadoop-2.2.0]# bin/hdfs dfs -mkdir /test
               mkdir: Permission denied: user=root, access=WRITE, inode="/":hadoop:supergroup:drwxr-xr-x
   
     7）启动历史服务器
          用于查看MapReduce 应用程序的历史运行情况
               sbin/mr-jobhistory-daemon.sh start historyserver