Hadoop-2.0.2-alpha安装 CentOs6.2 jdk1.7

首先，下载jdk1.7，安装jdk1.7,在此不详细说明；下载Hadoop-2.0.2（与先前版本安装方法有区别）；

然后，配置SSH登录不用输入密码：

      #ssh-keygen -t rsa -P ''

      #cat ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys   /*然后按enter键即可*/

      #ssh localhost

最后安装和配置Hadoop：

     1.解压Hadoop到/usr/local/hadoop目录中

     2.添加环境变量：

      #vim /etc/profile   末行添加如下

           export HADOOP_PREFIX=/usr/local/haddop

           export  PATH=$PATH:$HADOOP_PREFIX/bin:$HADOOP_PREFIX/sbin

          export HADOOP_MAPRED_HOME=${HADOOP_PREFIX}

          export HADOOP_COMMON_HOME=${HADOOP_PREFIX}

           export HADOOP_HDFS_HOME=${HADOOP_PREFIX}

           export YARN_HOME=${HADOOP_PREFIX}

      3.修改Hadoop的配置文件：

          hadoop-env.sh：

          #vim  /usr/local/hadoop/etc/hadoop/hadoop-env.sh

          修改JAVA_HOME，这里JAVA_HOME的路径必须指定为真实的路径，不能引用${JAVA_HOME}，否则运行的时候会有错误JAVA_HOME is not setd

         export JAVA_HOME=/usr/java/jdk1.7              //此处填写本地jdk的安装目录

         -----------------------------------------------------------------------------------------------------------------------------

        core-site.xml      

             <configuration>

               <property>

                  <name>fs.default.name</name>

                   <value>hdfs://localhost:8020</value>

                  <description>The name of the default file system. Either the

                                    literal string "local" or a host:port for NDFS.

                 </description>

                  <final>true</final>

               </property>

             </configuration>

        ----------------------------------------------------------------------------------------------------------------------------------

           hdfs-site.xml
             其中，/home/hduser/dfs/name，/home/hduser/dfs/data都是文件系统中的目录

             <configuration>

                      <property>

                        <name>dfs.namenode.name.dir</name>

                       <value>file:/home/hduser/dfs/name</value>

                        <description>Determines where on the local filesystem the DFS name node

                                           should store the name table. If this is a comma-delimited list

                                          of directories then the name table is replicated in all of the

                                         directories, for redundancy. </description>

                          <final>true</final>

                    </property>

                     <property>

                             <name>dfs.datanode.data.dir</name>

                             <value>file:/home/hduser/dfs/data</value>

                              <description>Determines where on the local filesystem an DFS data node

                                                should store its blocks. If this is a comma-delimited

                                             list of directories, then data will be stored in all named

                                                 directories, typically on different devices.

                                                 Directories that do not exist are ignored.

                                  </description>

                              <final>true</final>

                         </property>

                     <property>

                       <name>dfs.replication</name>

                     <value>1</value>

                        </property>

                              <property>

                                     <name>dfs.permissions</name>

                                      <value>false</value>

                              </property>

               </configuration>

          -----------------------------------------------------------------------------------------------------------------------------------------------------------
                 mapred-site.xml
                  其中，/home/hduser/mapred/system，/home/hduser/mapred/local都是文件系统中的目录
                <configuration>

                        <property>

                           <name>mapreduce.framework.name</name>

                          <value>yarn</value>

                      </property>

                      <property>

                                <name>mapred.system.dir</name>

                                <value>file:/home/hduser/mapred/system</value>

                                <final>true</final>

                      </property>

                       <property>

                             <name>mapred.local.dir</name>

                                   <value>file:/home/hduser/mapred/local</value>

                                  <final>true</final>

                           </property>

                 </configuration>

           ------------------------------------------------------------------------------------------------------------------------------------

         yarn-site.xml

        <configuration>

                  <property>

                     <name>yarn.nodemanager.aux-services</name>

                      <value>mapreduce.shuffle</value>

                 </property>

                <property>

                        <name>yarn.nodemanager.aux-services.mapreduce.shuffle.class</name>

                         <value>org.apache.hadoop.mapred.ShuffleHandler</value>

                   </property>

          </configuration>

          -----------------------------------------------------------------------------------------------------------------------------------------------------------         

             以上配置好后启动HDFS：

                  # hdfs namenode -format

          运行成功以后可以使用一下命令启动NameNode和DataNode

                  # hadoop-daemon.sh start namenode

                  # hadoop-daemon.sh start datanode

                 或者

                 # start-dfs.sh                      --一起启动

         启动 Yarn Daemons ：

                 #yarn-daemon.sh start resourcemanager

                 # yarn-daemon.sh start nodemanager

                或者

                # start-yarn.sh

 

                检查一下Demo是否启动
              #jps
                在我的机器上输出是这样的： 
              2048 NameNode
              2322 SecondaryNameNode
             3024 Jps
             2439 ResourceManager

             2441NodeManager
 
              ----------------------------------------------------end------------------------------------------------------------