Ubuntu 14.04 install hadoop 2.4.1 note

sudo adduser hadoop #该命令会自动创建用户hadoop,用户组hadoop,以及用户目录/home/hadoop   在Master和Slave1上分别编辑文件/etc/hosts:(添加以下两行)

10.211.55.1  Master  
10.211.55.10 Slave1    完成hosts修改后,保证Master和Slave1之间能够ping通.   ssh无密码登录配置   Master节点终端上操作: 
 
cd ~/.ssh
ssh-keygen -t rsa   Master节点需要无密码登录ssh本机,Master节点上操作:

cat ~/.ssh/id_rsa.pub>>~/.ssh/authorized_keys  
ssh Master #测试   将Master的公钥传输到Slave1节点  

ssh-copy-id -i ~/.ssh/id_rsa.pub hadoop@Slave1  #ssh-copy-id 把Master的公钥追加到Slave1的.ssh/authorized_keys上
ssh hadoop@Slave1  #or ssh Slave1 配置完网络交互环境后,下面进行hadoop配置文件的设置:   Master节点上操作   1.将hadoop-2.4.1的安装目录设置到环境变量HADOOP_HOME,并将$HADOOP_HOME/sbin添加到环境变量PATH,$HADOOP_HOME/share/hadoop/tools/lib/hadoop-streaming-2.4.1.jar设置到HADOOP_STREAMING环境变量中,在/etc/profile(全局用户)或者~/.bashrc(当前用户)中进行相应的相关修改.   2.进入到$HADOOP_HOME/etc/hadoop目录下,对slave,core-site.xml,hdfs-site.xml,mapred-site.xml(该文件若不存在,需要从模板中复制一份)   slave文件中记录Slave的名字Slave1(如果有多个Slave记录多个对应的名字).   core-site.xml:  

<property>
    <name>fs.defaultFS</name>
    <value>hdfs://Master:9000</value>
</property>
<property>
    <name>hadoop.tmp.dir</name>
    <value>file:/usr/local/hadoop/tmp</value>
    <description>Abase for other temporary directories.</description>
</property> hdfs-site.xml:

<property>
    <name>dfs.namenode.secondary.http-address</name>
    <value>Master:50090</value>
</property>
<property>
    <name>dfs.namenode.name.dir</name>
    <value>file:/usr/local/hadoop/tmp/dfs/name</value>
</property>
<property>
    <name>dfs.datanode.data.dir</name>
    <value>file:/usr/local/hadoop/tmp/dfs/data</value>
</property>
<property>
    <name>dfs.replication</name>
    <value>1</value>
</property> mapred-site.xml:

<property>
<name>mapreduce.framework.name</name>
<value>yarn</value>
</property> yarn-site.xml:

<property>
    <name>yarn.resourcemanager.hostname</name>
    <value>Master</value>
</property>
<property>
    <name>yarn.nodemanager.aux-services</name>
    <value>mapreduce_shuffle</value>
</property>

scp -r $HADOOP_HOME Slave1:~/ 
ssh Slave
sudo chown -R hadoop.hadoop $HADOOP_HOME #$HADOOP_HOME为Slave节点上的Hadoop安装目录  

hdfs namenode -format
start-dfs.sh
start-yarn.sh

jps  #检查是否启动了Nameode,SecondaryNameNode,ResourceManager,(Jps)进程
ssh Slave1 #登录到Slave1 
jps #检查是否启动了DataNode,NodeManager,(Jps)进程.

hdfs dfsadmin -report #Master节点上操作