hadoop单机安装

操作系统:centos7.4
cdh的hadoop发行版:hadoop-2.6-chd5.15.1

ssh免密登录配置

• ssh安装(已安装跳过):yum install -y openssh-server openssh-clients

• 生成密钥:ssh-keygen -t rsa

• 拷贝密钥:ssh-copy-id -i ~/.ssh/id_rsa.pub [ip/hosts]

注意事项: 1) .ssh目录的权限必须是700
2) .ssh/authorized_keys文件权限必须是600

hadoop环境变量配置

• hadoop-env.sh配置:${HADOOP_HOME}/etc/hadoop/hadoop-env.sh

 
   
   
 

  
    
    
  
export JAVA_HOME=/bigdata/jdk/jdk1.8.0_231
  
    
    
  
export HADOOP_HOME=/bigdata/hadoop/hadoop-2.6.0-cdh5.15.1
  
    
    
  
export HADOOP_CONF_DIR=/bigdata/hadoop/hadoop-2.6.0-cdh5.15.1/etc/hadoop
 
   
   
 

HDFS启动

• core-site.xml配置:${HADOOP_HOME}/etc/hadoop/core-site.xml

 
   
   
 

  
    
    
  
<configuration>
  
    
    
  
 <property>
  
    
    
  
 <name>fs.defaultFS</name>
  
    
    
  
 <value>hdfs://主机名:9000</value>
  
    
    
  
 </property>
  
    
    
  
</configuration>
 
   
   
 

• hdfs-site.xml配置:${HADOOOP_HOME}/etc/hadoop/hdfs-site.xml

 
   
   
 

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

• slave文件配置:${HADOOP_HOME}/etc/hadoop/slaves

 
   
   
 

  
    
    
  
#根据自己的主机名填写
  
    
    
  
${主机名}
 
   
   
 

• 
  

  HDFS文件系统格式化:${HADOOP_HOME}/bin/hdfs namenode -format

  
  

• 
  

  启动HDFS:${HADOOP_HOME}/sbin/start-dfs.sh

  
  

YARN启动

• mapred-site.xml配置:${HADOOP_HOME}/etc/hadoop/mapred-site.xml

 
   
   
 

  
    
    
  
<configuration>
  
    
    
  
 <property>
  
    
    
  
 <name>mapreduce.framework.name</name>
  
    
    
  
 <value>yarn</value>
  
    
    
  
 </property>
  
    
    
  
</configuration>
 
   
   
 

• yarn-site.xml配置：${HADOOP_HOME}/etc/hadoop/yarn-site.xml

 
   
   
 

  
    
    
  
<configuration>
  
    
    
  
 <property>
  
    
    
  
 <name>yarn.nodemanager.aux-services</name>
  
    
    
  
 <value>mapreduce_shuffle</value>
  
    
    
  
 </property>
  
    
    
  
</configuration>
 
   
   
 

• 启动yarn：${HADOOP_HOME}/sbin/start-yarn.sh