安装服务
分布式mr1:jobtracker、tasktracker
环境准备:
主节点:192.168.58.129(hostname:master)
子节点:192.168.58.130(slave1),192.168.58.131(slave2),192.168.58.132(slave3)
节点已经安装完成HDFS服务,参见
hadoop分布式部署系列1:HDFS
安装包:
mr1-2.0.0-mr1-cdh4.2.1.tar.gz
部署步骤:
1. 上传安装包并解压
/home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1
注:四个节点都安装
2. 修改以下文件
mr1文件1:
cp ~/app/hadoop-2.0.0-cdh4.2.1/etc/hadoop/core-site.xml ~/app/hadoop-2.0.0-mr1-cdh4.2.1/conf/
mr1文件2:/home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/conf/mapred-site.xml:
<property>
<name>mapred.job.tracker</name>
<value>master:9001</value>
</property>
mr1文件2:/home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/conf/mapred-site.xml:
<property>
<name>mapred.job.tracker</name>
<value>master:9001</value>
</property>
<property>
<name>mapred.temp.dir</name>
<value>${hadoop.tmp.dir}/mapred/temp</value>
<description>A shared directory for temporary</description>
</property>
mr1文件3:/home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/conf/slaves:
slave1
slave2
slave3
mr1文件4:/home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/conf/hadoop-env.sh:
export JAVA_HOME=/home/liulu/app/jdk1.6.0_31
注:上述修改文件在四个节点都修改
3. 启动mr1集群
cd /home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/bin
./start-mapred.sh
4. 检查mr1集群
看进程(JobTracker、TaskTracker):
[liulu@master conf]$ jps -m
2749 NameNode
2932 SecondaryNameNode
3263 JobTracker
[liulu@slave1 ~]$ jps -m
5068 DataNode
5210 TaskTracker
看JobTracker监控页面:
http://192.168.58.129:50030/jobtracker.jsp
5. mr1操作
执行wordcount例子
cd /home/liulu/app/hadoop-2.0.0-cdh4.2.1/bin
./hdfs dfs -mkdir /in
./hdfs dfs -put ~/testfile /in/(准备源数据)
cd /home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/bin
./hadoop jar ../hadoop-examples-2.0.0-mr1-cdh4.2.1.jar wordcount /in /out(执行wordcount)
./hadoop fs -ls /out(查看输出结果)
6. 关闭mr1
cd /home/liulu/app/hadoop-2.0.0-mr1-cdh4.2.1/bin
./stop-mapred.sh