搭建Hadoop分布式集群

软件和操作系统版本

Hadoop框架是采用Java语言编写，需要java环境（jvm） JDK版本：JDK8版本，本次使用的是 Java: jdk-8u431-linux-x64.tar.gz

Hadoop: hadoop-3.3.6.tar.gz

三台Linux虚拟节点: CentOS-7-x86_64-DVD-2009.iso

通过虚拟机设置共享文件夹将需要的安装包复制到linux虚拟机中。（/opt/software/）

[root@localhost1 software]# ll
total 912972
-rwxrwxrwx. 1 root root 730107476 Dec 23 20:43 hadoop-3.3.6.tar.gz
-rwxrwxrwx. 1 root root 148362647 Dec 23 22:10 jdk-8u431-linux-x64.tar.gz
-rwxrwxrwx. 1 root root      1068 Dec 24 11:29 LICENSE
-rwxrwxrwx. 1 root root  56405224 Dec 24 11:29 VMwareTools-10.3.21-14772444.tar.gz

集群规划

Linux	Localhost1	Localhost2	Localhost3
HDFS	NameNode、DataNode	DataNode	SecondaryNameNode、DataNode
YARN	NodeManager	NodeManager, ResourceManager	NodeManager

安装JDK

tar -zxvf jdk-8u431-linux-x64.tar.gz -C /usr/local/java/

配置环境变量：

vi /etc/profile

# java environment
JAVA_HOME=/usr/local/java/jdk1.8.0_431
JRE_HOME=${JAVA_HOME}/jre
CLASSPATH=.:${JAVA_HOME}/lib:${JRE_HOME}/lib:$CLASSPATH
PATH=$PATH:${JAVA_HOME}/bin:${JRE_HOME}/bin
export JAVA_HOME JRE_HOME CLASS_PATH PATH

让环境变量生效：

source /etc/profile

安装Hadoop

tar -zxvf hadoop-3.3.6.tar.gz -C /usr/local/applications

添加Hadoop到环境变量

vi /etc/profile

#HADOOP
HADOOP_HOME=/usr/local/applications/hadoop-3.3.6
PATH=$PATH:${HADOOP_HOME}/bin:${HADOOP_HOME}/sbin
export HADOOP_HOME PATH

让环境变量生效：

source /etc/profile

验证hadoop

hadoop version

集群配置

cd  /usr/local/applications/hadoop-3.3.6/etc/hadoop/

HDFS集群配置

配置：hadoop-env.sh

export JAVA_HOME=/usr/local/java/jdk1.8.0_431

指定NameNode节点以及数据存储目录（修改core-site.xml）

<!-- 指定HDFS中NameNode的地址 -->
    <property>
        <name>fs.defaultFS</name>
        <value>hdfs://localhost1:9000</value>
    </property>
    <!-- 指定Hadoop运行时产生文件的存储目录 -->
    <property>
        <name>hadoop.tmp.dir</name>
        <value>/usr/local/applications/hadoop-3.3.6/data/tmp</value>
    </property>

指定secondarynamenode节点(修改hdfs-site.xml)

    <!-- 指定Hadoop辅助名称节点主机配置 -->
    <property>
        <name>dfs.namenode.secondary.http-address</name>
        <value>localhost3:50090</value>
    </property>
    <!--副本数量 -->
    <property>
        <name>dfs.replication</name>
        <value>3</value>
    </property>

指定datanode从节点(修改workers文件，每个节点配置信息占一行)

localhost1
localhost2
localhost3

注意：该文件中添加的内容结尾不允许有空格，文件中不允许有空行。

MapReduce集群配置

指定MapReduce使用的jdk路径（修改mapred-env.sh）

export JAVA_HOME=/usr/local/java/jdk1.8.0_431

指定MapReduce计算框架运行Yarn资源调度框架(修改mapred-site.xml)

    <!-- 指定MR运行在Yarn上 -->
    <property>
        <name>mapreduce.framework.name</name>
        <value>yarn</value>
    </property>

Yarn集群配置

指定JDK路径

vi yarn-env.sh

export JAVA_HOME=/usr/local/java/jdk1.8.0_431

指定ResourceMnager的master节点信息(修改yarn-site.xml)

    <!-- 指定YARN的ResourceManager的地址 -->
    <property>
        <name>yarn.resourcemanager.hostname</name>
        <value>localhost2</value>
        </property>
        <!-- Reducer获取数据的方式 -->
    <property>
        <name>yarn.nodemanager.aux-services</name>
        <value>mapreduce_shuffle</value>
    </property>

分发配置

远程复制文件夹：将localhost1 配置好的Hadoop 复制到localhost2 和localhost3

scp -r /usr/local/applications/hadoop-3.3.6 root@localhost2:/usr/local/applications/hadoop-3.3.6
scp -r /usr/local/applications/hadoop-3.3.6 root@localhost3:/usr/local/applications/hadoop-3.3.6

配置SSH 密钥登陆

生成Key Pairs

在三台节点分别生成 Key Pairs，默认名字 id_rsa，千万不要改。

ssh-keygen -t rsa -C "root@localhost1"
ssh-keygen -t rsa -C "root@localhost2"
ssh-keygen -t rsa -C "root@localhost3"

key pairs 默认生产到 ~/.ssh/

[root@localhost1 .ssh]# ll
total 16
-rw-------. 1 root root 1191 Jan  5 20:22 authorized_keys
-rw-------. 1 root root 1679 Jan  5 20:04 id_rsa
-rw-r--r--. 1 root root  397 Jan  5 20:04 id_rsa.pub
-rw-r--r--. 1 root root  564 Jan  5 10:40 known_hosts

把三个节点的公钥互相发送到每个节点

添加公钥到远程服务器

ssh-copy-id root@localhost2
ssh-copy-id root@localhost3

添加公钥到本地

cat localhost1.pub >> authorized_keys

添加完成以后三个节点都有了所有节点的公钥

ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDTmy6v6Xsv4893ofLyMVyct439T4ePv/OfEtpMPtBqLsobyiDV1Ih0H17oxI8Mk4UvZIEgokyLDq8lz8n3v4Vr5RfaCLRvbFEmn7SEzV+tBeRd9tO9rccaSjBT5NfOlFpLecucbA1JP3sS1PgVmyGvD6xE490gD1EGKWDi04fxifwvrEJPxkIRS/FKKik1fB4IRtReKQlAkh/JwE3mvBdNbL0jeOKiB01ofQfV2xoOgooBJSkQNhRqxtPccWuH6GrKK3uTTpPZWZvCoW1vHKOaOH59AGFuRlrO6yobLdQueS/VE8jx7VdlZonICorgQuxUwQsCR9Ys+xKRwcpTtWin root@localhost1
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDdRLWUfYgCZ5KS2yC6BaOtCz/nsnb6EjO8jt6q8o9deen3+/jN/hnqhHFAEPRFGKq4WwPy4awYlo1sqDxgc8GrpzUUUnkIsnERhXp9L0df99X360t9QgNwNiz3ayTa40oAT0Zg1GbftjRH/foJIReEUA1IjGTAphTQYnKwWMsTOXeIqI7A2QeUUQvxVkNWJ/lTO59vCxIA0NJUkgdMgXImfGlb23gtX/RlJ0NCqdoOm83qrNLHcC+iIwOrKpFx9v88qdajiDi+pI1AqWz1fQD6LpIx9OojM9ZPb2VYvCSt4smvCECxH+r3/HOB68+bumynm2w26XQumZA/utpQgJaH root@localhost2
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDAZ36gymNo4jmEkeS9fuwSlZ5mceFPOQcmgKnfSEi4LWzRdg1hmHK3bOzFabnRLaKGZvtK1GBFtZFHF8LAmP4/BXyDAxpU2/kFkWjn+z/Cu4bfAUl98M1Gqj7L3iv360wpGLz/VB65YIBxn88sac4hEZZUSFNMZxbRm2MIdN05YVgN7z84nLph3VNLzfpdfqnn6fLP8Fu6qwd8NaE91M95Lf7xBIdO1k0+/4/RLjTWfnpo4pKw1YM1CgGQEuz0JW6+yNuDyDwtlgNXDZeOni9JBjda/UQckguQMDDFrRzgsFTK9j5tywJQSlE0TDXIqyIUSgiajO6V//XGFP1weVLP root@localhost3

关闭防火墙

三个linux节点都需要关闭防火墙，这个Hadoop集群才能互相访问。

临时关闭

systemctl stop firewalld

永久关闭

systemctl stop firewalld

systemctl disable firewalld

启动集群

1，Namenode执行格式化

hadoop namenode -format

2，启动HDFS

在namenode 节点执行：

cd /usr/local/applications/hadoop-3.3.6/

sbin/start-dfs.sh

3，启动YARN

在ResouceManager节点执行：

cd /usr/local/applications/hadoop-3.3.6/

sbin/start-yarn.sh

4，启动完成后

查看三台节点进程

[root@localhost1 hadoop-3.3.6]# jps
5232 Jps
4775 DataNode
4623 NameNode
5119 NodeManager


[root@localhost2 hadoop-3.3.6]# jps
3001 ResourceManager
3161 NodeManager
3514 Jps
2796 DataNode


[root@localhost3 ~]# jps
2793 DataNode
2907 SecondaryNameNode
3115 Jps
2991 NodeManager