安装:hadoop版本为1.2.1
实验环境为三台虚拟机,一台master,两台slave
1、关闭防火墙:{
centos为chkconfig iptables off(需要重启),不需要重启的是service iptables stop
}
2、修改hosts文件:vi /etc/hosts:{
第一列为ip地址
第二例为主机名
第三列为主机别名
}
注意:hadoop中主机名不能有下划线“_”否则后果很严重
3、修改主机名:vi /etc/sysconfig/network,将master改为hadoop-m,slave修改为hadoop-s1;hadoop-s2
4、添加用户:useradd grid
5、修改用户密码:passwd grid
6、配置ssh:{
ssh-keygen -t rsa -f ~/.ssh/id_rsa
cp id_rsa.pub authorized_keys
scp authorized_keys grid@hadoop-s1:~/.ssh
cat id_rsa.pub authorized_keys
scp authorized_keys grid@hadoop-m:~/.ssh
ping grid@hadoop_m
}
7、检查ssh是否配置成功:
8、将文件解压到家目录中:[grid@hadoop-m hadoop]$ tar -zxvf hadoop-1.2.1-bin.tar.gz
9、修改hadoop-1.2.1文件夹的名字:[grid@hadoop-m ~]$ mv hadoop-1.2.1 hadoop
10、配置hadoop-env.sh,将java环境变量配置到Java_HOME上
[grid@hadoop-m conf]$ vi hadoop-env.sh
11、修改core-site.xml,添加名称节点真实的ip:vi core-site.xml
如果是伪分布式,hadoop-m改为localhost
完全分布式模式:{
}
12、配置hdfs-site.xml 配置数据块存放位置和复制的个数:[grid@hadoop-m conf]$ vi hdfs-site.xml
![](http://www.dataguru.cn/kindeditor/attached/image/20130828/20130828124531_89145.png)
13、配置mapred-site.xml 作业跟踪器:[grid@hadoop-m conf]$ vi mapred-site.xml
![](http://www.dataguru.cn/kindeditor/attached/image/20130828/20130828124553_16603.png)
14、配置master:[grid@hadoop-m conf]$ vi masters
15、配置slave:[grid@hadoop-m conf]$ vi slaves
16、复制节点到slave中:scp -r hadoop grid@hadoop-s2:~/
17、格式化:[grid@hadoop-m hadoop]$ bin/hadoop namenode -format
18、启动hadoop:bin/start-all.sh