多节点分布式集群

准备n台（3台为例）服务器
配置好网络，主机名，ip-主机名的映射，关闭防火墙，安装java

---

• hadoop-server-00:

ssh-copy-id hadoop-server-01
ssh-copy-id hadoop-server-02
scp -r /usr/local/apps/hadoop-2.5.2/ hadoop-server-01:/usr/local/apps
scp -r /usr/local/apps/hadoop-2.5.2/ hadoop-server-02:/usr/local/apps
cd /usr/local/apps/hadoop-2.5.2/etc/
vi slaves
修改为
hadoop-server-00
hadoop-server-01
hadoop-server-02
vi /etc/profile
增加
export HADOOP_HOME=/usr/local/apps/hadoop-2.5.2
修改
export PATH=$PATH:$JAVA_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
source /etc/profile

• hadoop-server-01/02

cd /usr/local/apps/hadoop-2.5.2/
rm -rf tmp/

---

在hadoop-server-00任意目录即可启动hadoop服务

start-dfs.sh
start-yarn.sh