HADOOP配置文件、伪分布式集群、完全分布式、Mapreduce任务

(1) 添加HADOOP命令以及JDK到环境变量中

	    ~]# vim /etc/profile
			export JAVA_HOME=/app/jdk1.8.0_60
			export PATH=$JAVA_HOME/bin:$PATH
		~]# vim /etc/profile.d/hadoop.sh
			export HADOOP_HOME=/app/hadoop-2.7.2    //根据实际安装位置修改
			export PATH=$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$PATH

(2) 修改hadoop中的core-site.xml以方便使用hdfs

	    <property>
			<name>fs.defaultFS</name>
			<value>hdfs://node3:8020</value>   //hdfs安装节点URI
		</property>

(3) 修改hadoop中的hadoop-env.sh文件中的JAVA_HOME;

	JAVA_HOME=/app/jdk1.8.0_60             //根据实际JDK安装情况修改

(4) 启动Namenode以及Datanode并格式化Namenode

	    ~]# hdfs namenode -format              //格式化namenode,切记,只有第一次需要格式化
		~]# hadoop-daemon.sh start namenode    //启动namenode
		~]# jps					               //查看是否有"namenode",如果有则说明启动成功
		~]# hadoop-daemon.sh start datanode    //启动datanode

(5) 配置伪分布式集群模式

		~]# vim $HADOOP_HOME/etc/hadoop/hdfs-site.xml		
				<property>
					<name>dfs.repication</name>
					<value>1</value>
				</property>

(6) 跑测试MapReduce程序

		~]# mkdir $HADOOP_HOME/input
		~]# cp $HADOOP_HOME/etc/hadoop/*.xml $HADOOP_HOME/input
		~]# hadoop jar $HADOOP_HOME/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.7.2.jar grep input output 'hdf[a-z]+'
											// input是我们创建的目录,指定mapreduce从哪里读数据,grep是筛选的意思,output是输出目录,不存在的话会自动创建
		~]# ls $HADOOP_HOME/part*          // 会查看到处理后的数据

(7) 在HDFS上跑MapReduce程序

 		~]# hdfs dfs -mkdir -p /home/fang/input
		~]# vim test.txt                    // 编写测试文件
			fang
			Google
			learn
			Hadoop
			Mapreduce
			fang
		~]# hdfs dfs -put test.txt /home/fang/input  // 将本地创建的test.txt传送到hdfs上的指定目录下
		~]# hadoop jar $HADOOP_HOME/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.7.2.jar wordcount /home/fang/input /home/fang/output
			// wordcount是统计文件中的单词个数,从hdfs上的"/home/fang/input"中读取文件,处理后传到hdfs的"/home/fang/output"
		http://node3:50070 --> "Utilities" --> Browse the file system 可查看到hdfs上存储的数据(会看到在指定目录下生成的output文件夹)

(8) 在Yarn上跑MapReduce程序

		~]# vim $HADOOP_HOME/etc/hadoop/yarn-env.sh
			JAVA_HOME=/app/jdk1.8.0_60
		~]# vim $HADOOP_HOME/etc/hadoop/mapred-env.sh
			JAVA_HOME=/app/jdk1.8.0_60
		~]# cp $HADOOP_HOME/mapred-site.xml.template mapred-site.xml
		~]# vim mapred-site.xml
			<property>
				<name>mapreduce.framework.name</name>
				<value>yarn</value>
			</property>
		~]# vim $HADOOP_HOME/etc/hadoop/yarn-site.xml
			<property>
				<name>yarn.nodemanager.aux-services</name>
				<value>mapreduce_shuffle</value>
			</property>
			<property>
				<name>yarn.resourcemanager.hostname</name>
				<value>node3</value>
			</property>
		~]# yarn-daemon.sh start resourcemanager
		~]# yarn-daemon.sh start nodemanager
		http://node3:8088 --> application 可查看到MapReduce跑的任务

(9) 修改本地文件临时存储目录

		~]# mkdir -p $HADOOP_HOME/data/tmp
		~]# vim $HADOOP_HOME/etc/hadoop/core-site.xml
			<proprety>
				<name>hadoop.tmp.dir</name>
				<value>/app/hadoop-2.7.2/data/tmp</value>
			<proprety>
		~]# yarn-daemon.sh stop nodemanager
		~]# yarn-daemon.sh stop resourcemanager
		~]# hadoop-daemon.sh stop datanode
		~]# hadoop-daemon.sh stop namenode
		~]# rm -rf /tmp/hadoop-root        //删除之前生成的临时目录
		~]# hdfs namenode -format          // 重新格式化namenode
			// 再全部启动

(10) Hadoop相关配置文件(core-site.xml)

	~]# cd $HADOOP_HOME/share/hadoop/common/
	~]# unzip hadoop-common-2.7.2.jar -d /app
	~]# vim /app/core-site.xml       // 此文件中是配置文件的模板,可修改完后将内容复制到$HADOOP_HOME/etc/hadoop/core-site.xml

(11) Hadoop相关配置文件(hdfs-default.xml)

	    ~]# cd $HADOOP_HOME/share/hadoop/hdfs
		~]# unzip hadoop-hdfs-2.7.2.jar -d /app

(12) Hadoop相关配置文件(yarn-default.xml)

	    ~]# cd $HADOOP_HOME/share/hadoop/yarn
		~]# unzip hadoop-yarn-common-2.7.2.jar -d /app

(13) Hadoop相关配置文件(mapred-default.xml)

		~]# cd $HADOOP_HOME/share/hadoop/mapreduce
		~]# unzip hadoop-mapreduce-client-core-2.7.2.jar -d /app

Hadoop分布式集群的配置
一、步骤

(1) 配置免密
(2) 编写xsync同步文件脚本
(3) 编写xcall可在所有节点执行命令脚本
(4) 集群的部署规划

二、编写xsync脚本

~]# vim /usr/bin/xsync
			#!/bin/bash
			#判断是否有传入参数
			pcount=$#
			if (($pcount==0));then
				echo "Please input arg!" && exit 2;
			fi
			#取参数的文件名
			f1=$1
			fname=`basename $fname`
			echo fname=$fname
			#取参数的绝对路径
			pdir=`cd -P $(dirname $f1)`
			echo pdir=$pdir
			#获取当前用户名称
			user=`whoami`
			#循环,这里host根据自己的节点数和主机名设置
			for((host=1;host<=3;host++))
			do
				echo --------------- hadoop$host ----------------
				rsync -rvl $pdir/$fname $user@node$host:$pdir
			done
		~]# chmod +x /usr/bin/xsync
		~]# xsync /app/te
			fname=te
			pdir=/app
			/app/te root@node1:/app
			==================slave1==================
			sending incremental file list
	
			sent 25 bytes  received 12 bytes  24.67 bytes/sec
			total size is 0  speedup is 0.00
			/app/te root@node2:/app
			==================slave2==================
			sending incremental file list
			te
	
			sent 64 bytes  received 31 bytes  190.00 bytes/sec
			total size is 0  speedup is 0.00
			/app/te root@node3:/app
			==================slave3==================
			sending incremental file list
			te
	
			sent 64 bytes  received 31 bytes  190.00 bytes/sec
			total size is 0  speedup is 0.00

三、编写xcall脚本

	~]# vim /usr/bin/xcall
		#!/bin/bash
		pcount=$#
		if (($pcount==0));then
			echo "Please input arg!" && exit 2;
		fi
		user=`whoami`
		echo -------------localhost----------
		$@
		for ((host=1;host<=3;host++))
		do
			echo ----------node$host---------
			ssh $user@node$host $@
		done
	~]# chmod +x /usr/bin/xcall
	~]# xcall ls /app                // 即会打印node{1..3}中的文件

四、集群的部署规划

	node1 DataNode、Nodemanager
	node2 DataNode、Nodemanager、SecondaryNameNode
	node3 HDFS、Namenode、DataNode、ResourceManager

五、配置集群

操作Node3
	~]# vim $HADOOP_HOME/etc/hadoop/core-site.xml
		<property>
			<name>fs.defaultFS</name>
			<value>hdfs://node3:8020</value>   //hdfs安装节点URI
		</property>
		 <property>
		  <name>hadoop.tmp.dir</name>
		  <value>/app/hadoop-2.7.2/data/tmp</value>
		 </property>
		</configuration>
	~]# vim $HADOOP_HOME/etc/hadoop/hadoop-env.sh
		export JAVA_HOME=/app/jdk1.8.0_60
	~]# vim $HADOOP_HOME/etc/hadoop/hdfs-site.xml
		<property>
			<name>dfs.replication</name>
			<value>3</value>
		</property>
		<property>
			<name>dfs.namenode.secondary.http-address</name>
			<value>node2:50090</value>
		</property>
	~]# vim $HADOOP_HOME/etc/hadoop/slave     // DataNode所在节点
		node1
		node2
		node3
	~]# vim $HADOOP_HOME/etc/hadoop/yarn-env.sh
		JAVA_HOME=/app/jdk1.8.0_60/
	~]# vim $HADOOP_HOME/etc/hadoop/yarn-site.xml
		<property>
			<name>yarn.resourcemanager.hostname</name>
			<value>node3</value>
		</property>
		<property>
			<name>yarn.nodemanager.aux-services</name>
			<value>mapreduce_shuffle</value>
		</property>
	~]# vim $HADOOP_HOME/etc/hadoop/mapred-env.sh
		export JAVA_HOME=/app/jdk1.8.0_60/
	~]# vim $HADOOP_HOME/etc/hadoop/mapred-site.xml
		<property>
			<name>mapreduce.framework</name>
			<value>yarn</value>
		</property>

六、在集群上分发所有配置文件

~]# xsync $HADOOP_HOME/etc/hadoop

七、启动集群环境

~]# hdfs namenode -format
~]# start-dfs.sh
~]# start-yarn.sh              // 如果yarn与namenode不在同一台机器上就不能启动yarn,应该在Resourcemanager所在的机器上启动yarn

八、上传文件

~]# touch /app/test.txt
~]# hdfs dfs -put /app/test.txt /home/fang

九、下载文件

~]# hdfs dfs -get /home/fang/test /root

十、Hadoop启动停止方式

~]# hadoop-daemon.sh start|stop namenode|datanode|secondarynamenode // 逐一启动
~]# yarn-daemon.sh start|stop resourcemanager|nodemanager   //逐一启动
~]# start-dfs.sh  // 整体启动hdfs
~]# stop-dfs.sh // 整体停止hdfs
~]# start-yarn.sh // 整体启动yarn
~]# stop-yarn.sh // 整体停止yarn