2. Introduction, and Getting Set Up

GETTING SET UP (Install Java, Spark, Scala, Eclipse)

Install A Java Development Kit(JDK)

• From oracle.com

• Accept the default configurations

Install Spark (pre-built)

• From spark.apache.org

• Choose a Spark release later than or equal to 2.0.0

• Choose a package type: Pre-built for Hadoop 2.7 and later

• Download the Spark tgz file (Unix compression format on Windows)

• So you might have to download a 3rd party tgz programme on Windows to uncompress tgz file

• 3rd party tgz programme

• Open the tgz spark tgz file and extract to this folder, copy all the files inside to a new folder spark in (C:) drive

• Change the configuration file for spark, go to spark/conf and change log4j.properties.template to log4j.properties

• Open log4j.properties with whatever word editor you have (I uses Sublime 3)

• Change the following settings

  # Set everything to be logged to the console
  log4j.rootCategory=ERROR, console

• Install winutils.exe and HADOOP_HOME

• Create a new folder call winutils, and create a new folder called bin in (C:) drive

• Copy winutils.exe into C:/winutils/bin

Set Up SPARK_HOME JAVA_HOME And PATH environment Variables

• Setting up the Windows environment, right click on the windows icon on the left hand corner and go into Control Panel

• Click on Systems and Security, then onto System and then Advanced system settings. Click on Environment Variables

• Click on New under User Variables for User

• Input in the following details for New User Variable

    Variable name: SPARK_HOME
    Variable value: C:\spark
  
    Variable name: JAVA_HOME
    Variable value: C:\Program Files\Java\jdk1.8.0_172
  
    Variable name: HADOOP_HOME
    Variable value: C:\winutils

• Now click Edit on the Path under User Variables and for User and click on New for the following environment variables

    %SPARK_HOME%\bin
  
    %JAVA_HOME%\bin

• Press okay for all of the settings

Install Scala IDE (bundled with Eclipse)

• From scala-ide.org
• Extract and copy the eclipse folder to a new folder eclipse in (C:/)

• Create a Desktop shortcut to eclipse.exe in C:/eclipse/eclipse.exe

• Open up a Windows Command Prompt in Adminstrator

    # To see your spark directory 
    cd C:\spark
    dir
  
    # spark-shell.cmd is inside the bin folder in spark directory
    cd bin
  
    # To write a new simple Scala Spark Application
    spark-shell
  
    # To see if Spark works with RDD
    val rdd=sc.textFile("../README.md")
  
    # Counting the number of lines in that file
    rdd.count()

• To exit just hit Ctrl-D

Detailed, Written Steps At SunDog Website