前置环境

Hadoop集群必须部署完成，如果还没有搭建请先前往>>Hadoop全分布搭建笔记

程序版本

scala-2.11.8spark-2.0.0-bin-hadoop2.6

官网下载：Index of /dist/spark

组件介绍

Spark是加州大学伯克利分校AMP实验室（Algorithms, Machines, and People Lab）开发的通用内存并行计算框架。

Spark使用Scala语言进行实现，它是一种面向对象、函数式编程语言，能够像操作本地集合对象一样轻松地操作分布式数据集。

操作流程

上传Scala & Spark

使用FTP工具（xftp）上传Scala和Spark的程序包到master

[root@master ~]# ls

scala-2.11.8.tgz spark-2.0.0-bin-hadoop2.6.tgz

解压Scala & Spark

[root@master ~]# tar xf scala-2.11.8.tgz -C /usr/local/src/

[root@master ~]# tar xf spark-2.0.0-bin-hadoop2.6.tgz -C /usr/local/src/

解压后，cd进入解压目录

[root@master ~]# cd /usr/local/src/

[root@master src]# ls

hadoop jdk scala-2.11.8 spark-2.0.0-bin-hadoop2.6

修改文件夹名称

[root@master src]# mv scala-2.11.8/ scala

[root@master src]# mv spark-2.0.0-bin-hadoop2.6/ spark

[root@master src]# ls

hadoop jdk scala spark

配置环境变量

使环境变量仅对root用户生效，编辑root用户的环境变量

[root@master src]# vi /root/.bash_profile

修改后的文件状态

.bash_profile

Get the aliases and functions

if [ -f ~/.bashrc ]; then

    . ~/.bashrc

fi

User specific environment and startup programs

export JAVA_HOME=/usr/local/src/jdk

export HADOOP_HOME=/usr/local/src/hadoop

export SCALA_HOME=/usr/local/src/scala

export SPARK_HOME=/usr/local/src/spark

PATH=$PATH:$HOME/bin:$JAVA_HOME/bin:$HADOOP_HOME/bin:$HADOOP_HOME/sbin:$SCALA_HOME/bin:$SPARK_HOME/bin

export PATH

生效环境变量

[root@master src]# source /root/.bash_profile

配置Spark

进入Spark配置目录：cd /usr/local/src/spark/conf/

复制Spark配置模板：cp spark-env.sh.template spark-env.sh

编辑Spark配置文件：vi spark-env.sh

添加配置文件

提供Java、Hadoop、Scala相关文件位置

export JAVA_HOME=/usr/local/src/jdk

export HADOOP_HOME=/usr/local/src/hadoop

export HADOOP_CONF_DIR=/usr/local/src/hadoop/etc/hadoop

export SCALA_HOME=/usr/local/src/scala

设置主节点名称 (主机名或IP地址都可)

export SPARK_MASTER_HOST=master

设置主节点端口

export SPARK_MASTER_PORT=7077

设置每个Worker使用的核心数 (默认使用所有核心)

export SPARK_WORKER_CORES=1

设置每个Worker使用的内存大小 (默认使用1G内存)

export SPARK_WORKER_MEMORY=1G

参考文献：https://www.cnblogs.com/xupccc/p/9800380.html

配置Slaves文件

[root@master conf]# mv slaves.template slaves

[root@master conf]# vi slaves

文件里有 localhost 的请先删掉

slave1

slave2

拷贝配置好的程序到slave：scp网络拷贝命令 -r 拷贝文件夹

拷贝Scala环境

[root@master ~]# scp -r /usr/local/src/scala/ slave1:/usr/local/src/

……

[root@master ~]# scp -r /usr/local/src/scala/ slave2:/usr/local/src/

……

拷贝Spark环境

[root@master ~]# scp -r /usr/local/src/spark/ slave1:/usr/local/src/

……

[root@master ~]# scp -r /usr/local/src/spark/ slave2:/usr/local/src/

……

拷贝环境变量

[root@master ~]# scp /root/.bash_profile slave1:/root/

.bash_profile 100% 359 35.9KB/s 00:00

[root@master ~]# scp /root/.bash_profile slave2:/root/

.bash_profile 100% 359 38.1KB/s 00:00

切换到slave服务机，生效环境变量

[root@slave1 ~]# source /root/.bash_profile

[root@slave2 ~]# source /root/.bash_profile

启动Spark集群

启动 spark 脚本，由于与 Hadoop 脚本同名，需要进入到目录执行

[root@master ~]# cd /usr/local/src/spark/sbin/

注意脚本名前面有 ./ 不要少打了

[root@master sbin]# ./start-all.sh

starting org.apache.spark.deploy.master.Master, logging to /usr/local/src/spark/logs/spark-root-org.apache.spark.deploy.master.Master-1-master-01.out

slave2: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/src/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave2-01.out

slave1: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/src/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-slave1-01.out

master: starting org.apache.spark.deploy.worker.Worker, logging to /usr/local/src/spark/logs/spark-root-org.apache.spark.deploy.worker.Worker-1-master-01.out

查看集群状态

查看Spark进程状态：jps

master

[root@master ~]# jps

3903 SecondaryNameNode

4692 Master

3448 DataNode

4037 ResourceManager

4138 NodeManager

4892 Jps

3328 NameNode

slave1

[root@slave1 ~]# jps

2613 NodeManager

2865 Worker

2918 Jps

2528 DataNode

slave2

[root@slave2 ~]# jps

2253 NodeManager

2440 Jps

2385 Worker

2168 DataNode

测试Spark

[root@master bin]# spark-shell

Setting default log level to "WARN".

To adjust logging level use sc.setLogLevel(newLevel).

22/02/16 16:31:46 WARN util.NativeCodeLoader: Unable to load native-hadoop library for your platform... using builtin-java classes where applicable

22/02/16 16:31:49 WARN spark.SparkContext: Use an existing SparkContext, some configuration may not take effect.

Spark context Web UI available at http://172.16.2.11:4040

Spark context available as 'sc' (master = local[*], app id = local-1645047109135).

Spark session available as 'spark'.

Welcome to

  ____              __

 / __/__  ___ _____/ /__

_\ \/ _ \/ _ `/ __/  '_/

// ./_,// //_\ version 2.0.0

  /_/

     

Using Scala version 2.11.8 (Java HotSpot(TM) 64-Bit Server VM, Java 1.7.0_80)

Type in expressions to have them evaluated.

Type :help for more information.

scala>

查看Spark Web页面状态（master IP地址加上8080端口）：172.16.2.11:8080

至此——Spark集群搭建完成