【费元星原创】一键安装Hadoo2.7.6 集群完全分布式脚本-完美解决

有Bug 欢迎反馈，我不烦：feiyuanxing@gmail.com

  1 #!/bin/bash
 #@author：feiyuanxing 【既然笨到家，就要努力到家】
 #@date：2017-01-05
 #@E-Mail：feiyuanxing@gmail.com
 #@TARGET:一键安装hadoop 2.7.6 centos 64位
 #@CopyRight:本脚本遵守 未来星开源协议（http://feiyuanxing.com/kaiyuanxieyi/kaiyuanxieyi.html）

 #讲解：
 #0.前提条件：
 #  0.1 已经配置好JAVA环境变量
 #  0.2 已经配置免密登录环境
 #1.请使用root用户执行，软件默认安装在work用户，通常会在linux的/home 挂载一个大型磁盘
 #2.软件安装在/home/work/local/hadoop/hadoop  有人问为什么这么深
 #   2.1 work目录下local包含所有软件 hadoop安装大数据生态圈其他软件
 #   2.2 hadoop的数据安装在 /home/work/data ，此目录由于数量较大，可以单独挂载一个磁盘
 #
 ####################################################################################
 # 个性化设置
 # Hadoop 生态目录
 INSTALL_HOME=/home/work/local/hadoop

 #data 目录
 DATA_HOME=/home/work/data
 # Hadoop目录
 ROOT_HOME=${INSTALL_HOME}/hadoop_tmp

 #name 通常为主机（本机），缺省配置为master
 NAMENODE=master
 #name 通常为第一台从机，缺省配置为slave01
 SECONDARYNAMENODE=slave01
 #hadoop 版本
 HADOOP_VERSION=hadoop-2.7.6.tar.gz
 ####################################################################################
 # Hadoop 生态数据路径
 hadoop_logs=${DATA_HOME}/hadoop_logs
 hadoop_tmp=${DATA_HOME}/hadoop_tmp
 hadoop_data=${DATA_HOME}/hadoop_data
 hadoop_name=${DATA_HOME}/hadoop_name
 #当前脚本的位置
 SCRIPT_HOME=$(cd `dirname $0`; pwd)

 mkdir -p ${INSTALL_HOME} && cd ${INSTALL_HOME}
 mkdir -p ${ROOT_HOME} && mkdir -p ${hadoop_logs} && mkdir -p ${hadoop_tmp}&& mkdir -p ${hadoop_data}&& mkdir -p ${hadoop_name}

 #chown -R work:work /home/work/data/hadoop

 #集群机器信息
 INSTALL_IP_ARRAY=()
 INSTALL_HOSTNAME_ARRAY=()

 # 需要判断用户是否存在
 function add_work_user(){
     adduer work -d /home/work
     passwd work

 }

 #需要判断jdk是否存在
 function install_jdk(){
     # 有jdk
     echo $JAVA_HOME

 }

 #下载hadoop
 function download_hodoop(){
     echo "(2-1/6)正在获取安装包：${INSTALL_HOME}/hadoop-2.7.6.tar.gz"
     if [ ! -f ${INSTALL_HOME}/hadoop-2.7.6.tar.gz ] ;then
         echo "开始下载"
         wget https://mirrors.tuna.tsinghua.edu.cn/apache/hadoop/common/hadoop-2.7.6/${HADOOP_VERSION}
     fi
     echo "(2-2/6)获取安装包完成：${INSTALL_HOME}/hadoop-2.7.6.tar.gz"

 }

 function configuration_ssh(){
     #设置本机免密码登录
     (echo -e "\n"
     sleep 1
     echo -e "\n"
     sleep 1
     echo -e "\n")|ssh-keygen -t rsa
     cat  ~/.ssh/id_rsa.pub >> ~/.ssh/authorized_keys
     chmod 700  ~/.ssh
     chmod 600  ~/.ssh/authorized_keys
 }

 function readFileToArra(){
     echo "(1/6)正在读取配置文件：${SCRIPT_HOME}/hostip_hadoop.txt"
     if [ ! -f "${SCRIPT_HOME}"/hostip_hadoop.txt ];then
         echo "请在本目录下的【hostip_hadoop.txt】文件中添加Hadoop集群机器信息"
         echo "#USENAME@IP:PORT,PASSWD（第一个必须配置为主，通常为本机）
 root@191.168.1.1:22,123456
 root@191.168.1.2:22,123456" > ${SCRIPT_HOME}/hostip_hadoop.txt
         exit;
     else
         INSTALL_IP_ARRAY=($(cat "${SCRIPT_HOME}"/hostip_hadoop.txt| grep -v "^#"))
         INSTALL_HOSTNAME_ARRAY=($(cat "${SCRIPT_HOME}"/hostip_hadoop.txt| grep -v "^#"|awk -F '@' '{print $2}'|awk -F ':' '{print $1}'))
     fi

     #for(( i=0 ; i<${#INSTALL_IP_ARRAY[@]} ; i++)) do
     #    echo "打印数组：${INSTALL_IP_ARRAY[i]}"
     #done;
 }

 #readFileToArra

 function configure_hadoop(){

     echo "(3-1/6)正在解压安装包："
     #最终生成的文件名为install_hadoop.bin而我们的all.tar.gz被>>到该文件后面
     #tail -c $size  install_hadoop.bin >all.tar.gz
     rm -rf hadoop hadoop_tmp
     tar -zxf ${HADOOP_VERSION}
     #pwd
     mv hadoop-2.7.6 hadoop_tmp && cd hadoop_tmp

     echo "(3-2/6)正在配置Hadoop环境变量："
     #设置环境变量
     hadoop_home_before=`cat /etc/profile| grep "HADOOP_HOME"`

     if [ -z "${hadoop_home_before}" ] ;then
         echo "hadoop 环境变量已存在..."
         sed -i '/HADOOP_HOME/d' /etc/profile
         echo "export HADOOP_HOME=${INSTALL_HOME}" >> /etc/profile
         echo 'export PATH=$PATH:${HADOOP_HOME}/bin:${HADOOP_HOME}/sbin' >> /etc/profile
         #使得环境变量生效
         source /etc/profile
         sleep 1;
         source /etc/profile

     fi

     echo "(4/6)正在配置Hadoop基本配置："
     #通过强大的sed命令来修改hadoop的配置文件
     jdk_home=`echo $JAVA_HOME`
     if [ -z ${jdk_home} ] ;then
         #安装JDK
         echo "程序已退出，请先配置JDK"
         exit
     fi

     #echo "开始配置hadoop"
     # 1.配置hadoop-env.sh
     sed -i "s!\${JAVA_HOME}!$(echo ${jdk_home})!g" ${ROOT_HOME}/etc/hadoop/hadoop-env.sh
     # 2.配置core-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>fs.defaultFS\<\/name\>\n\\t\\t\<value\>hdfs://${NAMENODE}:9000\<\/value\>\n\\t\<\/property\>\n\\t\<property\>\n\\t\\t\<name\>hadoop.tmp.dir\<\/name\>\n\\t\\t\<value\>file:$(echo ${hadoop_tmp})\<\/value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/core-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>io.file.buffer.size\</name\>\n\\t\\t\<value\>131072\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/core-site.xml
     # 3.配置hdfs-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>dfs.webhdfs.enabled\</name\>\n\\t\\t\<value\>true\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/hdfs-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>dfs.replication\</name\>\n\\t\\t\<value\>2\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/hdfs-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>dfs.datanode.data.dir\</name\>\n\\t\\t\<value\>file:${hadoop_name}\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/hdfs-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>dfs.namenode.name.dir\</name\>\n\\t\\t\<value\>file:${hadoop_data}\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/hdfs-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>dfs.namenode.secondary.http-address\</name\>\n\\t\\t\<value\>master:9001\</value\>\n\\t\</property\>"   ${ROOT_HOME}/etc/hadoop/hdfs-site.xml
     # 4.配置mapred-site.xml
     cp ${ROOT_HOME}/etc/hadoop/mapred-site.xml.template  ${ROOT_HOME}/etc/hadoop/mapred-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>mapreduce.jobhistory.webapp.address\</name\>\n\\t\\t\<value\>master:19888\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/mapred-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>mapreduce.jobhistory.address\</name\>\n\\t\\t\<value\>master:10020\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/mapred-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>mapreduce.framework.name\</name\>\n\\t\\t\<value\>yarn\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/mapred-site.xml
     # 5.配置yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.resourcemanager.webapp.address\</name\>\n\\t\\t\<value\>master:8088\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.resourcemanager.admin.address\</name\>\n\\t\\t\<value\>master:8033\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.resourcemanager.resource-tracker.address\</name\>\n\\t\\t\<value\>master:8031\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.resourcemanager.scheduler.address\</name\>\n\\t\\t\<value\>master:8030\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.resourcemanager.address\</name\>\n\\t\\t\<value\>master:8032\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.nodemanager.aux-services.mapreduce.shuffle.class\</name\>\n\\t\\t\<value\>org.apache.hadoop.mapred.ShuffleHandler\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     sed -i "/<configuration>/a\\\t\<property\>\n\\t\\t\<name\>yarn.nodemanager.aux-services\</name\>\n\\t\\t\<value\>mapreduce_shuffle\</value\>\n\\t\</property\>"  ${ROOT_HOME}/etc/hadoop/yarn-site.xml
     # 6.配置slaves文件
     echo "" > ${ROOT_HOME}/etc/hadoop/slaves
     for(( i=0 ; i<${#INSTALL_HOSTNAME_ARRAY[@]} ; i++)) do
         echo "${INSTALL_HOSTNAME_ARRAY[i]}" >> ${ROOT_HOME}/etc/hadoop/slaves
     done;

     echo "(5/6)正在发送hadoop到其他机器"
     for line in ${INSTALL_IP_ARRAY[@]}
     do
         user=`echo ${line}| awk -F "@" '{print $1}'`
         ip=`echo ${line}| cut -d "@" -f 2| cut -d ":" -f 1`
         port=`echo ${line}| cut -d "@" -f 2| cut -d ":" -f 2|cut -d "," -f 1`
         echo ${user} - ${ip} - ${port} -  ${passwd}
         ssh -p ${port} -o StrictHostKeyChecking=no "${user}@${ip}" "rm -rf ${DATA_HOME}/hadoop_*;rm -rf ${INSTALL_HOME}/hadoop;mkdir -p ${INSTALL_HOME}/hadoop;mkdir -p ${hadoop_logs} && mkdir -p ${hadoop_tmp}&& mkdir -p ${hadoop_data}&& mkdir -p ${hadoop_name}"
         #echo "scp -P ${port} -o StrictHostKeyChecking=no -rpq ${ROOT_HOME}  \"${user}\"@\"${ip}\":${INSTALL_HOME}"
         scp -P ${port} -o StrictHostKeyChecking=no -rpq ${ROOT_HOME}/*  "${user}"@"${ip}":${INSTALL_HOME}/hadoop
         ssh -p ${port} -o StrictHostKeyChecking=no "${user}@${ip}" "chown -R work:work ${INSTALL_HOME}"
     done

     echo "(6/6)Hadoop配置成功"

 }

 #配置Hadoop启动服务
 function start_hadoop(){
     #格式化hadoop
     hdfs  namenode -format
     echo "格式化完成..开始运行"
     #先启动HDFS
     start-dfs.sh
     #再启动YARN
     start-yarn.sh

     #检查是否启动成功
     jps

 }
 #动态读取配置文件
 readFileToArra
 #add_work_user
 download_hodoop;
 #configuration_ssh
 configure_hadoop
 #start_hadoop