hadoop fully distributed mode of operation

First, prepare the virtual machine environment

1. clone a virtual machine

2. Modify the static IP cloned virtual machine

vim /etc/sysconfig/network-scripts/ifcfg-eth33

3. Modify the host name

View hostname hostname
modify the hostnamevi /etc/sysconfig/network

NETWORKING=yes
NETWORKING_IPV6=no
HOSTNAME= hadoop101

Linux modify host mapping file (hosts file) vim /etc/hosts
add

8.8.8.101 hadoop101
8.8.8.102 hadoop102
8.8.8.103 hadoop103

After editing *重启设备*

4. Close the firewall (centos 7)

Check firewall status firewall-cmd --state
stop firewall systemctl stop firewalld.service
prohibit firewall bootsystemctl stop firewalld.service

5. Create user banana

Adding a user useradd banana
to set the user password passwd banana
to see if the user exists id banana
to see which users have created cat /etc/passwd
delete a user deletes a user but to save the user's home directory userdel bananato delete the user and the user's home directory userdel -r banana
's own user name displaywhoami

6.sudo configure banana root user

Modify the configuration file vi /etc/sudoers
find the following line (line 91), add the following line to the root

## Allow root to run any commands anywhere
root    ALL=(ALL)     ALL
banana   ALL=(ALL)     ALL

Or when configured to use sudo command, no password

## Allow root to run any commands anywhere
root      ALL=(ALL)     ALL
atguigu   ALL=(ALL)     NOPASSWD:ALL

Modification, can now use banana account login, and then use the command sudo, you can get root privileges to operate ( sudo useradd zhangsan)

Second, the distribution of cluster write scripts xsync

1.scp (secure copy) copy security [push, pull, third parties]

Definitions: SCP data copy may be implemented between the server and the server
Syntax: scp -r $pdir/$fname $user@hadoop$host:$pdir/$fname
i.e. scp -r banana@hadoop101:/soft/software root@hadoop103:/soft/software(at 102 copies the file 101 to 103)

Note: Do not forget to copy over the configuration file at the source / etc / profile
Note: copied the / opt / module directory, do not forget to modify all the files in hadoop102 on, hadoop103, owner and owner group.sudo chown banana:banana -R /soft/software

2.rsync remote synchronization tool

用rsync做文件的复制要比scp的速度快，rsync只对差异文件做更新。scp是把所有文件都复制过去
语法:rsync -rvl $pdir/$fname $user@hadoop$host:$pdir/$fname

3.xsync集群分发脚本

定义：循环复制文件到所有节点的相同目录下
脚本实现
1.在/home/banana目录下创建bin目录，并在bin目录下xsync创建文件
2.文件内容

#!/bin/bash
#1 获取输入参数个数，如果没有参数，直接退出
pcount=$#
if((pcount==0)); then
echo no args;
exit;
fi

#2 获取文件名称
p1=$1
fname=`basename $p1`
echo fname=$fname

#3 获取上级目录到绝对路径
pdir=`cd -P $(dirname $p1); pwd`
echo pdir=$pdir

#4 获取当前用户名称
user=`whoami`

#5 循环
for((host=103; host<105; host++)); do
        echo ------------------- hadoop$host --------------
        rsync -rvl $pdir/$fname $user@hadoop$host:$pdir
done

3.修改脚本 xsync 具有执行权限 chmod 777 xsync
4.调用脚本形式 xsync /home/banana/bin

注意：如果将xsync放到/home/atguigu/bin目录下仍然不能实现全局使用，可以将xsync移动到/usr/local/bin目录下。

三、完全分布式运行模式

1）准备3台客户机（关闭防火墙、静态ip、主机名称）

2）安装JDK并配置环境变量

解压 tar -zxvf jdk-8u161-linux-x64.tar.gz -C /soft/software (/soft/software是自己新建的目录)
配置环境变量 vim /etc/profile

文件尾部加入：
export JAVA_HOME=/soft/software/jdk1.8.0_161/    #安装jdk的路径
export PATH=$PATH:$JAVA_HOME/bin
export CLASSPATH=.:$JAVA_HOME/lib/tools.jar:$JAVA_HOME/lib/dt.jar

使配置生效 source /etc/profile

3）安装Hadoop并配置环境变量

解压 tar -zxvf hadoop-2.9.2.tar.gz -C /soft/software
配置环境变量 vim /etc/profile

文件尾部加入：
export HADOOP_HOME=/soft/software/hadoop-2.9.2                #自己安装的Hadoop路径
export PATH=$PATH:$HADOOP_HOME/bin:$HADOOP_HOME/sbin

使配置生效 source /etc/profile

4）配置集群

• 核心配置文件 vi core-site.xml

<!-- 指定HDFS中NameNode的地址 -->
<property>
        <name>fs.defaultFS</name>
      <value>hdfs://hadoop101:9000</value>
</property>

<!-- 指定Hadoop运行时产生文件的存储目录 -->
<property>
        <name>hadoop.tmp.dir</name>
        <value>/soft/module/hadoop-2.9.2/data/tmp</value>
</property>

• HDFS配置文件

配置hadoop-env.sh vi hadoop-env.sh

export JAVA_HOME=/soft/module/jdk1.8.0_161

配置hdfs-site.xml vi hdfs-site.xml

<property>
        <name>dfs.replication</name>
        <value>3</value>
</property>

<!-- 指定Hadoop辅助名称节点主机配置 -->
<property>
      <name>dfs.namenode.secondary.http-address</name>
      <value>hadoop103:50090</value>
</property>

• YARN配置文件

vi yarn-env.sh

export JAVA_HOME=/soft/module/jdk1.8.0_161

vi yarn-site.xml

<!-- Reducer获取数据的方式 -->
<property>
        <name>yarn.nodemanager.aux-services</name>
        <value>mapreduce_shuffle</value>
</property>

<!-- 指定YARN的ResourceManager的地址 -->
<property>
        <name>yarn.resourcemanager.hostname</name>
        <value>hadoop102</value>
</property>

• MapReduce配置文件

配置mapred-env.sh vi mapred-env.sh

export JAVA_HOME=/soft/module/jdk1.8.0_161

配置mapred-site.xml cp mapred-site.xml.template mapred-site.xml vi mapred-site.xml

vi mapred-site.xml
在该文件中增加如下配置
<!-- 指定MR运行在Yarn上 -->
<property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
</property>

• 分发配置好的hadoop配置文件 xsync /opt/module/hadoop-2.9.2/
• 在另外的节点上查看文件分发情况 cat /opt/module/hadoop-2.9.2/etc/hadoop/core-site.xml

5）单点启动

• 如果集群是第一次启动，需要格式化NameNode hadoop namenode -format
• 在hadoop101上启动NameNode hadoop-daemon.sh start namenode
• 在hadoop101、hadoop102以及hadoop103上分别启动DataNode hadoop-daemon.sh start datanode jps查看进程

6）配置ssh、群起集群

• 各节点间无密登录

生成公钥和私钥(然后敲（三个回车），就会生成两个文件id_rsa（私钥）、id_rsa.pub（公钥）)
ssh-keygen -t rsa

将公钥拷贝到要免密登录的目标机器上
ssh-copy-id hadoop101
ssh-copy-id hadoop102
ssh-copy-id hadoop103

X3(在所有节点上执行)
========================================
配置slaves
vim /soft/module/hadoop-2.9.2/etc/hadoop/slaves
加上
hadoop101
hadoop102
hadoop103

分发文件
xsync.sh /soft/module/hadoop-2.9.2/etc/hadoop/slaves

• 启动集群
  • 如果集群是第一次启动，需要格式化NameNode（注意格式化之前，一定要先停止上次启动的所有namenode和datanode进程，然后再删除data和log数据） hdfs namenode -format
  • 启动HDFS 在namenode(hadoop101)节点上执行 sbin/start-dfs.sh
  • 启动YARN 在ResouceManager(hadoop102)节点上执行 sbin/start-yarn.sh
• 在网页上查看集群信息

7）测试集群

• 上传文件到集群
• 查看文件存放在什么位置 (HDFS文件存储路径)(HDFS在磁盘存储文件内容)
• 拼接文件
• 下载文件