本篇文章主要对完全分布式Hadoop集群环境的安装与配置步骤进行介绍。
集群的节点规划信息如下：

Host Name	IP Address	Node Type	User Name
hadoop34	172.16.250.234	DataNode / NodeManager / NameNode	hadoop / root
hadoop39	172.16.250.239	DataNode / NodeManager / SecondaryNameNode	hadoop / root
hadoop40	172.16.250.240	DataNode / NodeManager /ResourceManager	hadoop / root

Hadoop官方地址：http://hadoop.apache.org

环境配置

All nodes are disabled SELinux and firewalld
All nodes can ping with each other
All nodes have same hadoop directory structure and a same user account
Create a hadoop user, home directory is /home/hadoop, add into root group
hadoop directory is /usr/local/hadoop, directory owner is hadoop
Master node and slave node can SSH with no password publick key authentication
All nodes have same /etc/hosts, add master node and slave node record line

修改主机名（root权限）

由于 Hadoop 集群内部有时需要通过主机名来进行相互通信，因此我们需要保证每一台机器的主机名都不相同。
下面给出不同CentOS版本下修改主机名的操作命令：
Centos6:

[root@localhost ~]# hostname           # 查看当前的 hostname
localhost
[root@localhost ~]# vim /etc/sysconfig/network # 编辑 network 文件修改hostname行（重启生效）
[root@localhost ~]# cat /etc/sysconfig/network # 检查修改
NETWORKING=yes
HOSTNAME=hadoop34
[root@localhost ~]# hostname hadoop34      # 设置当前的hostname(立即生效）

Centos7:

[root@localhost ~]# hostname                      # 查看当前的 hostnmae
localhost
[root@localhost ~]# hostnamectl set-hostname hadoop34  # 永久修改hostname(立即生效）
[root@localhost ~]# hostname               # 检查修改
hadoop34

修改HOSTS（root权限）

修改HOSTS的原因主要有两点：
1.Hadoop 内部机制需要通过主机名对主机进行访问。
2.使用主机名对 Hadoop 集群进行配置，看起来更加一目了然。
在CentOS中修改HOSTS 操作命令如下：

# 修改每台机器的 /etc/hosts 文件
[root@localhost ~]# vi /etc/hosts
# 在文件中添加集群所有主机的 IP 和主机名的对应关系，IP与主机名之间使用一个 TAB 键分隔
172.16.250.234  hadoop34
172.16.250.239  hadoop39
172.16.250.240  hadoop40

如果想要多个主机名路由到同一个 IP，只需要在 IP 后边添加多个主机名即可，多个主机名之间同样使用 TAB 键进行分隔，例如：

172.16.250.234  hadoop34  namenode34  resourcemanager34

关闭SELinux（root权限）

因为CentOS的所有访问权限都是有SELinux来管理的，为了避免我们安装中由于权限关系而导致的失败，需要先将其关闭，以后根据需要再进行重新管理。
在CentOS中关闭SELinux使用如下操作命令：

[root@localhost ~]# getenforce                     # 查看当前的 SELinux 状态
Enforcing
# setenforce 1 可以设置 SELinux 为 enforcing 模式
[root@localhost ~]# setenforce 0                    # 将 SELinux 的状态临时设置为 Permissive 模式（立即生效）
[root@localhost ~]# getenforce                    # 检查修改
Permissive
[root@localhost ~]# vim /etc/selinux/config              
# 编辑 config 文件将 SELINUX=enforcing 修改为 SELINUX=disabled（重启生效）
[root@localhost ~]# cat /etc/selinux/config              # 检查修改

# This file controls the state of SELinux on the system.
# SELINUX= can take one of these three values:
#   enforcing - SELinux security policy is enforced.
#   permissive - SELinux prints warnings instead of enforcing.
#   disabled - No SELinux policy is loaded.
SELINUX=disabled
# SELINUXTYPE= can take one of these two values:
#   targeted - Targeted processes are protected,
#   mls - Multi Level Security protection.
SELINUXTYPE=targeted

[root@localhost ~]# sestatus                 # 查看 SELinux 当前的详细状态
SELinux status:         enabled
SELinuxfs mount:        /selinux
Current mode:          permissive
Mode from config file:     disabled
Policy version:         24
Policy from config file:    targeted

注意:使用getenforce 命令获取当前 SELinux 的运行状态为 permissive 或者 disabled 时均表示关闭。

关闭防火墙（root权限）

为避免由于防火墙策略导致安装失败问题，需要先关闭防火墙，下面给出不同CentOS版本下关闭防火墙的操作命令：
Centos6:

[root@localhost ~]# service iptables start         # 开启防火墙
iptables: Applying firewall rules:             [ OK ]
[root@localhost ~]# service iptables status        # 查看防火墙状态
Table: filter
Chain INPUT (policy ACCEPT)
num target   prot opt source        destination    
1  ACCEPT   all -- 0.0.0.0/0      0.0.0.0/0      state RELATED,ESTABLISHED
2  ACCEPT   icmp -- 0.0.0.0/0      0.0.0.0/0     
3  ACCEPT   all -- 0.0.0.0/0      0.0.0.0/0     
4  ACCEPT   tcp -- 0.0.0.0/0      0.0.0.0/0      state NEW tcp dpt:22
5  REJECT   all -- 0.0.0.0/0      0.0.0.0/0      reject-with icmp-host-prohibited

Chain FORWARD (policy ACCEPT)
num target   prot opt source        destination    
1  REJECT   all -- 0.0.0.0/0      0.0.0.0/0      reject-with icmp-host-prohibited

Chain OUTPUT (policy ACCEPT)
num target   prot opt source        destination    
[root@localhost ~]# chkconfig iptables off         # 永久关闭防火墙（重启生效）
[root@localhost ~]# service iptables stop          # 临时关闭防火墙（立即生效）
iptables: Setting chains to policy ACCEPT: filter     [ OK ]
iptables: Flushing firewall rules:             [ OK ]
iptables: Unloading modules:                [ OK ]
[root@localhost ~]# service iptables status         # 检查修改
iptables: Firewall is not running.

Centos7:

# Centos7 中使用 systemctl 命令来管理服务，命令格式如下
# systemctl [start 开启]|[stop 停止]|[restart 重启]|[status 状态][enable 开机启动]| [disable 禁止开机启动] 服务名称
[root@localhost ~]# systemctl start firewalld       # 开启防火墙
[root@localhost ~]# systemctl status firewalld       # 查看防火墙状态
● firewalld.service - firewalld - dynamic firewall daemon
 Loaded: loaded (/usr/lib/systemd/system/firewalld.service; disabled; vendor preset: enabled)
 Active: active (running) since Wed 2018-08-08 09:18:09 CST; 24s ago      
         # active (running) 表示防火墙开启
  Docs: man:firewalld(1)
Main PID: 21501 (firewalld)
 CGroup: /system.slice/firewalld.service
     └─21501 /usr/bin/python -Es /usr/sbin/firewalld --nofork --nopid

Aug 08 09:18:07 localhost.localdomain systemd[1]: Starting firewalld - dynamic firewall daemon...
Aug 08 09:18:09 localhost.localdomain systemd[1]: Started firewalld - dynamic firewall daemon.
[root@localhost ~]# systemctl disable firewalld      # 永久关闭防火墙（重启生效）
[root@localhost ~]# systemctl stop firewalld        # 临时关闭防火墙（立即生效）
[root@localhost ~]# systemctl status firewalld       # 检查修改
● firewalld.service - firewalld - dynamic firewall daemon
 Loaded: loaded (/usr/lib/systemd/system/firewalld.service; disabled; vendor preset: enabled)
 Active: inactive (dead)
         # inactive (dead) 表示防火墙关闭
  Docs: man:firewalld(1)

Aug 08 09:18:07 localhost.localdomain systemd[1]: Starting firewalld - dynamic firewall daemon...
Aug 08 09:18:09 localhost.localdomain systemd[1]: Started firewalld - dynamic firewall daemon.
Aug 08 09:19:24 localhost.localdomain systemd[1]: Stopping firewalld - dynamic firewall daemon...
Aug 08 09:19:24 localhost.localdomain systemd[1]: Stopped firewalld - dynamic firewall daemon.

JDK的安装与配置

由于Hadoop 的编译及MapReduce的运行都需要使用JDK，所以需要在集群中的每一台机器上提前安装好能够满足Hadoop 最低版本要求的JDK。
如何在 Centos 中安装JDK，可以参考文章：https://blog.csdn.net/pengjunlee/article/details/53932094

SSH免密码登录配置

在 Hadoop 集群中的各个节点之间需要使用 SSH 频繁地进行通信，为了避免每次的通信都要求输入密码，需要对各个节点进行 SSH 免密码登录配置。

开启sshd秘钥认证

在进行SSH免密码登录配置之前，需要先开启 sshd 秘钥认证：编辑每一台机器的 /etc/ssh/sshd_config 文件，去掉下面这3行前的 “#” 注释。

 # RSAAuthentication yes
 # PubkeyAuthentication yes
 # AuthorizedKeysFile   .ssh/authorized_keys

修改完成后保存，并执行以下命令重启 sshd 服务使修改生效。

# Centos6:
service sshd restart         # 重启 sshd 服务
# Centos7:
systemctl restart sshd       # 重启 sshd 服务

创建免密码登录账户

由于Hadoop 集群中的各节点默认会使用当前的账号SSH免密码登录其它节点，所以需要在每个节点中创建一个相同的供 Hadoop 集群专用的账户，本例中使用的账户为 hadoop 。
在各台机器中执行下面几行命令，创建免密码登录的 hadoop 账户。

[root@localhost ~]# useradd hadoop
[root@localhost ~]# passwd hadoop
Changing password for user hadoop.
New password:
Retype new password:
passwd: all authentication tokens updated successfully.

生成公钥和私钥

在每一台机器中都执行 su hadoop 命令，从 root 用户切换到要免密码登录的 hadoop 账户，然后执行 ssh-keygen -t rsa 命令，生成用来SSH免密码登录的公钥和私钥。

[root@localhost ~]# su hadoop
[hadoop@hadoop34 root]$ ssh-keygen -t rsa
Generating public/private rsa key pair.
Enter file in which to save the key (/home/hadoop/.ssh/id_rsa):
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in /home/hadoop/.ssh/id_rsa.
Your public key has been saved in /home/hadoop/.ssh/id_rsa.pub.
The key fingerprint is:
SHA256:b7+xzQ8SlkIH4amLnxQuaSphVFKPSH04x0VkM2tTAPU hadoop@hadoop34
The key's randomart image is:
+---[RSA 2048]----+
| .o.o.*X.+.   |
| ...=o+. B o   |
| .o.+. + E .  |
| .  . + . .  |
| .   S . +  |
| o  + + o .  |
| . . = + o o . |
| . o + o . * . |
|  ..  o  +.o..|
+----[SHA256]-----+

上述秘钥生成过程中，无需指定秘钥存放目录和口令密码，直接回车，命令执行完毕后会在 hadoop 账户的家目录中（/home/hadoop/.ssh）生成两个文件：

  id_rsa: 私钥
  id_rsa.pub:公钥

将公钥导入到认证文件

秘钥生成之后，执行以下命令将每台机器的公钥都拷贝到认证文件 authorized_keys 中。

[hadoop@hadoop34 bin]$ cat /home/hadoop/.ssh/id_rsa.pub >> /home/hadoop/.ssh/authorized_keys
[hadoop@hadoop34 bin]$ ssh hadoop@hadoop39 cat /home/hadoop/.ssh/id_rsa.pub >> /home/hadoop/.ssh/authorized_keys
The authenticity of host 'hadoop39 (172.16.250.239)' can't be established.
ECDSA key fingerprint is SHA256:QWIpDi4MJ3VkzHwuSOE/3z1+QxiMfc9j5/VPH9cCnKY.
ECDSA key fingerprint is MD5:cb:77:1c:a7:25:56:49:61:93:14:ab:8a:0a:3b:8e:f3.
Are you sure you want to continue connecting (yes/no) yes
Warning: Permanently added 'hadoop39,172.16.250.239' (ECDSA) to the list of known hosts.
hadoop@hadoop39's password:                      # 输入刚刚创建的 hadoop 账户的密码
[hadoop@hadoop34 bin]$ ssh hadoop@hadoop40 cat /home/hadoop/.ssh/id_rsa.pub >> /home/hadoop/.ssh/authorized_keys
The authenticity of host 'hadoop40 (172.16.250.240)' can't be established.
ECDSA key fingerprint is SHA256:QWIpDi4MJ3VkzHwuSOE/3z1+QxiMfc9j5/VPH9cCnKY.
ECDSA key fingerprint is MD5:cb:77:1c:a7:25:56:49:61:93:14:ab:8a:0a:3b:8e:f3.
Are you sure you want to continue connecting (yes/no) yes
Warning: Permanently added 'hadoop40,172.16.250.240' (ECDSA) to the list of known hosts.
hadoop@hadoop40's password:                      # 输入刚刚创建的 hadoop 账户的密码

查看 authorized_keys 文件内容如下：

[hadoop@hadoop34 bin]$ cat /home/hadoop/.ssh/authorized_keys
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQC/psoBkOHlulUweuF2bQrRY1IjyR2d19Byu8GnoAlpPSabdtCfRD7m2XPymK8z1xe3lfPg1VHTlPnCEJg5WbcAxEp+R70tgjgCoUKuhrfdHGp+XvXSQ9SYq+iplSFwGj3DLrHeycDIc1l+jtCHaUb1+XdId3M2jRxD+RYZO6MOqnjflcRyAax01Sv4T3amp1h0nmAd4Bylk3BZdV98ClQ3ZPZnZbK62W88I617JpBoR1XjnbNxAVq6SeDSmL1lucQtf3bJTfZMVKUVsG0uewtUwQozV9AYmIHS6rKFvpXqvlTGuogoOSF1otVlt0pweEJ7bEgzS8qjg1MV6kFJMFeD hadoop@hadoop34
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDsH9nQF7ixOmhAUzilxHzgpT3wIHf+eibp/49RLbT3DFH3M9huWCQ613k2rcye8pSrNMKuirLoBjPWXrvEUEYrtLZTzSuJbyi6l0fcvPVvhUkYnOjOoEywc+CEPQilBL463wThRt6+4+gMqdPslSuZ8uRWe+5WGRuLUsBitEkEybUdkou+aeuSjukSqbaJvB6R8vVND0LocLAxLu6IzSrr3lOUu/l8WlPCKppDbbdzgPy0JOCocLB6cEcxOSaBpBK6Qf5YXfClI/lCj8F8yZVTHM5cYmheeLv0Z7icJdDbD+6j5QI3t2ad9Qwqit/mQQ1ILO/JCN7TF+rvAv3cltYD hadoop@hadoop39
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQC9XQsRhN2VmYMPMZ1gngpRBoUBvUJXJADleDPGqk4stmo7TZoy9fMUB++UQsdDTM0C4+zEY7CZb6QRMRhNR7fw9v1F/bDijdYeOSt/SDcOAuShSAzoT0KI6AT+hWGq4JHDWbSydL/x1unLFbS16n9m1EXnvpZL2ksxbCGRMi4EN91+5uy2QmNujQsIkHVe6v04C+p1Q1N36SVurzOAbT6bDTGO++NvViN9EAEBMH0UUHWbbtxplNo7ADwoXKsY17Cu5EjoqVadF6stTfMbW5NFO2FFDUbylWB2ijf0DUfGQ2ayXKYCGrqjufnYs4vv59GJufn6iAh4SqHtt2okxA6L hadoop@hadoop40

设置认证文件访问权限（hadoop34主机）

在 hadoop34 主机上执行如下命令，对认证文件的操作权限进行设置：

[hadoop@hadoop34 bin]$ chmod 700 /home/hadoop/.ssh
[hadoop@hadoop34 bin]$ chmod 600 /home/hadoop/.ssh/authorized_keys

补全known_hosts中主机列表

[hadoop@hadoop34 ~]$ vim /home/hadoop/.ssh/known_hosts
hadoop39,172.16.250.239 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBETbnFS5S1bJ2r8j9YxsoQLZJd8hG82ey7aeyMb09fVYjK6npXAyPOPmxhRROle3PEKM2yVN3mooPqUVfDk/NpU=
hadoop40,172.16.250.240 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBETbnFS5S1bJ2r8j9YxsoQLZJd8hG82ey7aeyMb09fVYjK6npXAyPOPmxhRROle3PEKM2yVN3mooPqUVfDk/NpU=

发现此时hadoop34主机的 known_hosts 中仅有hadoop39和hadoop40两台主机的信息，缺少hadoop34主机自身的信息，所以需要ssh免密登录一次将自身的主机信息添加到known_hosts列表中。

[hadoop@hadoop34 bin]$ ssh hadoop34
The authenticity of host 'hadoop34 (172.16.250.234)' can't be established.
ECDSA key fingerprint is SHA256:9/GfAEOJoVF5C6WoIPDYBqH1twQV4bpmxtRj0c4YF90.
ECDSA key fingerprint is MD5:da:6f:4b:87:2e:81:ab:3b:e1:6c:76:3a:a3:e0:1a:7c.
Are you sure you want to continue connecting (yes/no) yes
Warning: Permanently added 'hadoop34,172.16.250.234' (ECDSA) to the list of known hosts.
Last login: Sat Aug 11 17:41:56 2018

再次查看known_hosts中的主机列表。

[hadoop@hadoop34 ~]$ vim /home/hadoop/.ssh/known_hosts
hadoop39,172.16.250.239 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBETbnFS5S1bJ2r8j9YxsoQLZJd8hG82ey7aeyMb09fVYjK6npXAyPOPmxhRROle3PEKM2yVN3mooPqUVfDk/NpU=
hadoop40,172.16.250.240 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBETbnFS5S1bJ2r8j9YxsoQLZJd8hG82ey7aeyMb09fVYjK6npXAyPOPmxhRROle3PEKM2yVN3mooPqUVfDk/NpU=
hadoop34,172.16.250.234 ecdsa-sha2-nistp256 AAAAE2VjZHNhLXNoYTItbmlzdHAyNTYAAAAIbmlzdHAyNTYAAABBBEU7TSuLX7wyJMhr8GEc/Fw75tIcQDXRRVS20kAvp6zL+bZSbTE9kF+gkYBtiyEq0SEKWYPHjUUMRL4ZKhPmOWA=

将认证文件复制到其他主机

在 hadoop34 主机上执行以下命令将生成的 authorized_keys、known_hosts两个文件从 master 复制到 hadoop39 和 hadoop40 。

# scp [要传输的本地文件] [远程主机用户名]@远程主机ip或主机名:[文件要传输到的目标位置]
[hadoop@hadoop34 ~]$ scp /home/hadoop/.ssh/authorized_keys hadoop@hadoop39:/home/hadoop/.ssh/authorized_keys
hadoop@hadoop39's password:
authorized_keys                                  100% 1191   1.4MB/s  00:00  
[hadoop@hadoop34 ~]$ scp /home/hadoop/.ssh/authorized_keys hadoop@hadoop40:/home/hadoop/.ssh/authorized_keys
hadoop@hadoop40's password:
authorized_keys                                  100% 1191   1.2MB/s  00:00  
[hadoop@hadoop34 ~]$ scp /home/hadoop/.ssh/known_hosts hadoop@hadoop39:/home/hadoop/.ssh/known_hosts
hadoop@hadoop39's password:
known_hosts                                    100% 555  511.8KB/s  00:00  
[hadoop@hadoop34 ~]$ scp /home/hadoop/.ssh/known_hosts hadoop@hadoop40:/home/hadoop/.ssh/known_hosts
hadoop@hadoop40's password:
known_hosts                                    100% 555  271.8KB/s  00:00

设置认证文件访问权限（hadoop39和hadoop40）

# 在 hadoop39主机上执行
[hadoop@hadoop39 .ssh]$ chmod 700 /home/hadoop/.ssh
[hadoop@hadoop39 .ssh]$ chmod 600 /home/hadoop/.ssh/authorized_keys
# 在 hadoop40主机上执行
[hadoop@hadoop40 .ssh]$ chmod 700 /home/hadoop/.ssh
[hadoop@hadoop40 .ssh]$ chmod 600 /home/hadoop/.ssh/authorized_keys

SSH免密码登录测试

在 hadoop40 主机上执行 ssh hadoop34 命令就能够免密码登录 hadoop34 主机了。

[hadoop@hadoop40 .ssh]$ ssh hadoop34
Last login: Sat Aug 11 17:56:45 2018 from hadoop34
[hadoop@hadoop34 ~]$ exit
logout
Connection to hadoop34 closed.

SSH免密码登录详细配置，请参考文章：https://blog.csdn.net/pengjunlee/article/details/80919833

搭建Hadoop集群

安装Hadoop

Hadoop安装其实就是一个将下载好的Hadoop压缩包解压到所有服务器上并进行配置的过程，推荐将Hadoop安装在各个节点服务器的相同目录下方便配置，我使用的Hadoop版本是hadoop-2.9.1，指定的Hadoop的安装目录是 /usr/local/hadoop-2.9.1 。
使用如下操作命令将下载好的 hadoop-2.9.1.tar.gz 解压至安装目录：

[hadoop@hadoop34 src]$ tar zxvf hadoop-2.9.1.tar.gz -C /usr/local/

如果解压时使用的是 root 账户，那么还需要执行以下命令将Hadoop的安装目录的所有者和所属组改为 hadoop：

[hadoop@hadoop34 src]$ chown -R hadoop:hadoop /usr/local/hadoop-2.9.1/

编辑 /home/hadoop/.bash_profile 配置文件，增加 Hadoop 相关用户环境变量内容如下：

export HADOOP_HOME=/usr/local/hadoop-2.9.1
PATH=$PATH:$HOME/.local/bin:$HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin
export PATH

所使用到的命令如下：

vim /home/hadoop/.bash_profile       # 编辑 .bash_profile 文件，增加 Hadoop 环境变量
source /home/hadoop/.bash_profile     # 重新加载 .bash_profile 配置文件
hadoop version               # 检查修改，查看 Hadoop 版本信息

配置hadoop集群

Hadoop的Java相关配置由下面两种类型的配置文件共同决定：

只读的缺省配置，包括：core-default.xml, hdfs-default.xml, yarn-default.xml 和 mapred-default.xml 。
节点的个性配置，包括：etc/hadoop/core-site.xml, etc/hadoop/hdfs-site.xml, etc/hadoop/yarn-site.xml 和 etc/hadoop/mapred-site.xml 。

篇幅所限，只读的缺省配置就不再一一列出，想要了解的可以访问以下链接进行查看：
http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-common/core-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-hdfs/hdfs-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-yarn/hadoop-yarn-common/yarn-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-mapreduce-client/hadoop-mapreduce-client-core/mapred-default.xml
接下来，主要讲解如何对各节点进行个性化配置，一般情况下，我们只需要修改下图标示的几个文件即可。

配置core-site.xml

<configuration>
    <!-- The name of the default file system -->
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://hadoop34:9000</value>
    </property>
    <!-- The size of buffer for use in sequence files -->
    <property>
        <name>io.file.buffer.size</name>
        <value>131072</value>
    </property>
    <!-- A base for other temporary directories -->
    <property>
        <name>hadoop.tmp.dir</name>
        <value>file:/usr/local/hadoop-2.9.1/tmp</value>
    </property>
</configuration>

配置hdfs-site.xml

<configuration>
    <property>
        <name>dfs.namenode.name.dir</name>
        <value>file:/usr/local/hadoop-2.9.1/data/namespace</value>
        <final>true</final>
        <description>Path on the local filesystem where the NameNode stores</description>
    </property>
    <property>
        <name>dfs.datanode.data.dir</name>
        <value>/usr/local/hadoop-2.9.1/data/dataspace</value>
        <final>true</final>
        <description>Path on the local filesystem where the DataNode stores Data</description>
    </property>
    <property>
        <name>dfs.namenode.secondary.http-address</name>
        <value>hadoop39:50090</value>
        <final>true</final>
        <description>The secondary namenode http server address and port</description>
    </property>
    <property>
        <name>dfs.webhdfs.enabled</name>
        <value>true</value>
        <final>true</final>
        <description>Enable WebHDFS (REST API) in Namenodes and Datanodes</description>
    </property>
    <property>
        <name>dfs.permissions</name>
        <value>false</value>
        <final>true</final>
        <description>Disable permission checking in HDFS</description>
    </property>
    <property>
        <name>dfs.replication</name>
        <value>2</value>
        <final>true</final>
        <description>Default block replication</description>
    </property>
</configuration>

配置mapred-site.xml

<configuration>
    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
        <final>true</final>
        <description>The runtime framework for executing MapReduce jobs</description>
    </property>
</configuration>

配置yarn-site.xml

<configuration>
    <property>
        <name>yarn.resourcemanager.scheduler.address</name>
        <value>hadoop40:8030</value>
        <final>true</final>
        <description>The address of the scheduler interface</description>
    </property>
    <property>
        <name>yarn.resourcemanager.resource-tracker.address</name>
        <value>hadoop40:8031</value>
        <final>true</final>
    </property>
    <property>
        <name>yarn.resourcemanager.address</name>
        <value>hadoop40:8032</value>
        <final>true</final>
        <description>The address of the applications manager interface in the RM</description>
    </property>
    <property>
        <name>yarn.resourcemanager.admin.address</name>
        <value>hadoop40:8033</value>
        <final>true</final>
        <description>The address of the RM admin interface</description>
    </property>
    <property>
        <name>yarn.resourcemanager.webapp.address</name>
        <value>hadoop40:8088</value>
        <final>true</final>
        <description>The http address of the RM web application</description>
    </property>
        <property> 
                <name>yarn.nodemanager.aux-services</name>
                <value>mapreduce_shuffle</value>
                <final>true</final>
        </property>
</configuration>

配置slaves

hadoop34
hadoop39
hadoop40

配置hadoop-env.sh

# 将
export JAVA_HOME=${JAVA_HOME}
# 修改为：
export JAVA_HOME=/usr/local/jdk1.8.0_144

将配置分发到其他节点

scp -r /usr/local/hadoop-2.9.1/etc/hadoop/* hadoop39:/usr/local/hadoop-2.9.1/etc/hadoop/
scp -r /usr/local/hadoop-2.9.1/etc/hadoop/* hadoop40:/usr/local/hadoop-2.9.1/etc/hadoop/

操作Hadoop集群

格式化NameNode

第一次启动HDFS需要先进行格式化，使用如下命令：

# $HADOOP_PREFIX/bin/hdfs namenode -format <cluster_name>
hdfs namenode -format hadoop_cluster

HDFS格式化完成之后，使用如下命令启动HDFS集群：

# $HADOOP_PREFIX/sbin/start-dfs.sh
start-dfs.sh

在集群的任意 HDFS 节点上执行 start-dfs.sh 命令，控制台打印的 HDFS集群的启动流程如下：

# 在HDFS集群中的任意节点上执行 start-dfs.sh
[hadoop@hadoop34 hadoop-2.9.1]$ start-dfs.sh
Starting namenodes on [hadoop34]
hadoop34: starting namenode, logging to /usr/local/hadoop-2.9.1/logs/hadoop-hadoop-namenode-hadoop34.out
hadoop34: starting datanode, logging to /usr/local/hadoop-2.9.1/logs/hadoop-hadoop-datanode-hadoop34.out
hadoop40: starting datanode, logging to /usr/local/hadoop-2.9.1/logs/hadoop-hadoop-datanode-hadoop40.out
hadoop39: starting datanode, logging to /usr/local/hadoop-2.9.1/logs/hadoop-hadoop-datanode-hadoop39.out
Starting secondary namenodes [hadoop39]
hadoop39: starting secondarynamenode, logging to /usr/local/hadoop-2.9.1/logs/hadoop-hadoop-secondarynamenode-hadoop39.out

HDFS集群启动完成之后，此时各个节点上的进程如下所示：

# hadoop34
[hadoop@hadoop34 hadoop-2.9.1]$ jps
11123 DataNode
11351 Jps
10969 NameNode
# hadoop39
[hadoop@hadoop39 data]$ jps
660 DataNode
822 SecondaryNameNode
893 Jps
# hadoop40
[hadoop@hadoop40 data]$ jps
14811 DataNode
15003 Jps

在浏览器中输入 http://172.16.250.234:50070 ，查看HDFS的管理界面。

注意：使用 start-dfs.sh 命令启动HDFS集群时，在集群的任意节点上执行该命令都能够启动集群。

启动 YARN 使用如下命令：

# $ $HADOOP_PREFIX/sbin/start-yarn.sh
start-yarn.sh

在被指派为 ResourceManager 的主机上执行 start-yarn.sh 命令，控制台打印的 YARN 的启动流程如下：

# 必须在 ResourceManager 所在的 hadoop40 主机上执行start-yarn.sh
[hadoop@hadoop40 hadoop-2.9.1]$ start-yarn.sh
starting yarn daemons
starting resourcemanager, logging to /usr/local/hadoop-2.9.1/logs/yarn-hadoop-resourcemanager-slave2.out
hadoop34: starting nodemanager, logging to /usr/local/hadoop-2.9.1/logs/yarn-hadoop-nodemanager-hadoop34.out
hadoop40: starting nodemanager, logging to /usr/local/hadoop-2.9.1/logs/yarn-hadoop-nodemanager-hadoop40.out
hadoop39: starting nodemanager, logging to /usr/local/hadoop-2.9.1/logs/yarn-hadoop-nodemanager-hadoop39.out

待YARN启动完成之后，再次查看各个节点上运行的进程：

# hadoop34
[hadoop@hadoop34 hadoop-2.9.1]$ jps
11504 NameNode
11940 Jps
11624 DataNode
11784 NodeManager
# hadoop39
[hadoop@hadoop39 data]$ jps
1347 NodeManager
1076 DataNode
1593 Jps
1197 SecondaryNameNode
# hadoop40
[hadoop@hadoop40 hadoop-2.9.1]$ jps
16264 NodeManager
15721 DataNode
16682 Jps
16141 ResourceManager

在浏览器中输入 http://172.16.250.240:8088 ，查看YARN的管理界面。

注意：

在使用 start-yarn.sh 命令启动YARN集群时，必须在ResourceManager所在节点上执行该命令才能够启动 ResourceManager；

当 Namenode和ResourceManger不在同一台主机时，就不能在NameNode上启动 YARN了，应该在ResouceManager所在的机器上启动YARN。

MapReduce测试

为了验证 hadoop集群是否安装成功，以及初体验一把MapReduce，接下来，我们以 Hadoop 安装包中自带的MapReduce 示例为例，简单统计一下 $HADOOP_HOME/etc/hadoop 目录下的配置文件中匹配 'dfs[a-z.]+'的字符串都有哪些以及它们在配置文件中出现了多少次。

[hadoop@hadoop34 hadoop]$ hdfs dfs -mkdir /input        # 在HDFS根目录下创建input目录
[hadoop@hadoop34 hadoop]$ hdfs dfs -mkdir /output       # 在HDFS根目录下创建output目录
[hadoop@hadoop34 hadoop]$ hdfs dfs -ls /                # 查看HDFS根目录下文件列表
Found 2 items
drwxr-xr-x  - hadoop supergroup     0 2018-08-12 00:20 /input
drwxr-xr-x  - hadoop supergroup     0 2018-08-12 00:23 /output
# 将本地主机上的 hadoop 文件夹上传到 HDFS的 input目录
[hadoop@hadoop34 hadoop]$ hdfs dfs -put /usr/local/hadoop-2.9.1/etc/hadoop /input
[hadoop@hadoop34 hadoop]$ hdfs dfs -ls /input/hadoop    # 查看HDFS中/input/hadoop目录下文件列表
Found 29 items
-rw-r--r--  2 hadoop supergroup    7861 2018-08-12 00:20 /input/hadoop/capacity-scheduler.xml
-rw-r--r--  2 hadoop supergroup    1335 2018-08-12 00:20 /input/hadoop/configuration.xsl
# 此处省略其余 25个文件信息 
-rw-r--r--  2 hadoop supergroup    4876 2018-08-12 00:20 /input/hadoop/yarn-env.sh
-rw-r--r--  2 hadoop supergroup    1971 2018-08-12 00:20 /input/hadoop/yarn-site.xml
# 使用 hadoop 提供的mapreduce示例程序，统计上述29个文件中匹配'dfs[a-z.]+'的字符串出现的次数
[hadoop@hadoop34 hadoop]$ hadoop jar /usr/local/hadoop-2.9.1/share/hadoop/mapreduce/hadoop-mapreduce-examples-2.9.1.jar grep /input/hadoop /output/first 'dfs[a-z.]+'
[hadoop@hadoop34 hadoop]$ hdfs dfs -ls /output/first   # 查看结果存放目录/output/first 下的文件列表
Found 2 items
-rw-r--r--  2 hadoop supergroup     0 2018-08-12 00:52 /output/first/_SUCCESS
-rw-r--r--  2 hadoop supergroup    338 2018-08-12 00:52 /output/first/part-r-00000
# 你可以将结果文件目录从HDFS拷贝到本地主机的 /usr/local/hadoop-2.9.1/tmp/ 目录下
[hadoop@hadoop34 hadoop]$ hdfs dfs -get /output/first /usr/local/hadoop-2.9.1/tmp/
[hadoop@hadoop34 hadoop]$ ls /usr/local/hadoop-2.9.1/tmp/first/
part-r-00000 _SUCCESS
[hadoop@hadoop34 hadoop]$ cat /usr/local/hadoop-2.9.1/tmp/first/part-r-00000
# 或者直接查看HDFS中mapreduce任务的统计结果
[hadoop@hadoop34 hadoop]$ hdfs dfs -cat /output/first/part-r-00000
# mapreduce任务执行完毕后的的统计结果内容如下:
6 dfs.audit.logger
4 dfs.class
3 dfs.logger
3 dfs.server.namenode.
2 dfs.audit.log.maxfilesize
2 dfs.period
2 dfs.audit.log.maxbackupindex
1 dfsmetrics.log
1 dfsadmin
1 dfs.webhdfs.enabled
1 dfs.servers
1 dfs.replication
1 dfs.permissions
1 dfs.namenode.secondary.http
1 dfs.log
1 dfs.file
1 dfs.datanode.data.dir
1 dfs.namenode.name.dir

注意： 在使用 hadoop jar 命令执行mapreduce任务的时候，用来存放计算结果的输出目录必须是一个尚未创建的目录。

参考文章：

http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-common/SingleCluster.html
http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-common/ClusterSetup.html
http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-common/core-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-project-dist/hadoop-hdfs/hdfs-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-yarn/hadoop-yarn-common/yarn-default.xml
http://hadoop.apache.org/docs/r2.9.1/hadoop-mapreduce-client/hadoop-mapreduce-client-core/mapred-default.xml
http://hadoop.apache.org/docs/current/api/index.html