Apache Spark lab (Spark 2.0, Scala 2.11)
Branch: master
Clone or download
Pull request Compare This branch is even with jayway:master.
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
gradle/wrapper
src
.gitignore
LICENSE
README.md
build.gradle
gradlew
gradlew.bat
settings.gradle

README.md

Spark lab

Clone this repository!

Installation

In this lab we will use Gradle, Spark and Scala but you do not have to install any of them yourself - it should all be taken care of by the included Gradle wrapper!

Setup IntelliJ IDEA (recommended)

Install IntelliJ IDEA Community.

Open project in IntelliJ IDEA

If you haven't used IntelliJ IDEA before, be sure to install the Scala plugin when you reach "Download featured plugins" during the setup wizard. Default settings are ok otherwise.

  • Start IntelliJ IDEA.
  • If this is your first JVM project:
  • Click Configure > Project Defaults > Project Structure.
  • Under Project and Project SDK, click New... and select JDK.
    • Windows: C:\Program Files\Java\jdk1.8.0_<version>
    • OS X: /Library/Java/JavaVirtualMachines/jdk1.8.0_<version>.jdk/Contents/H‌​ome
    • Linux: /usr/lib/jvm(/java-8-oracle) (can be different depending on your distribution)
  • Click "Open".
  • Locate project folder spark-lab and select it.
  • Click "OK".
  • Use the default settings but make sure Gradle JVM is 1.8.
  • Click "OK".
  • Wait for everything to download and build - it will take a little while.

If you haven't already been asked to install the Scala plugin yet, do it now (Preferences > Plugins > Install JetBrains plugin...)

Running tests

The code for part0 is already implemented, you should make sure the test runs as expected (see below).

IntelliJ IDEA

IntelliJ IDEA need to download the dependencies and index the project. In the bottom right corner, if you have a text displaying <nr> process(es) running..., wait for it to finish (otherwise the Run menu will not be visible).

In the left side-bar (Cmd + 1 or Alt + 1 if not visible), expand spark-lab > src > test > scala.

Right click on the part0 package, click Run > ScalaTests in 'part0... (the bottom one with the little Scala icon).

It's recommended to use ScalaTests ... (the bottom one with the little Scala icon) and not Test ..., otherwise things can get a little screwed up if you wish to run individual tests later.

Optional

You can also open any test source file and right click either on the class name (AlreadyImplementedTest in part0 to run all the tests in that class), or on the line test("... (to run a specific test). As before, click Run > ScalaTests in 'part0... (the bottom one with the little Scala icon).

Command line

./gradlew test --tests part0* (OS X / Linux)

gradlew.bat test --tests part0* (Windows)

Optional

Run all tests:

./gradlew test (OS X / Linux)

gradlew.bat test (Windows)

Advanced options: http://stackoverflow.com/a/31468902

Doing the lab

The lab is divided into several parts, you should preferably do them in order although it is not required.

To get started, navigate to the first package src > main > scala > part1 and double-click the source file.

You'll find further instructions about what to do in each source file.

Implement all the methods by replacing the ??? with your (hopefully) working code.

Verify that your solutions are correct by running the corresponding unit tests, src > test > scala > part1.

Spark UI (optional)

While it's not needed, it can be interesting to look at the Spark UI while the tests are running. Unfortunately, the test will probably finish before you'll have a chance to look at it.

To overcome this, you can go to src > test > scala > setup and open the SparkTest trait. In the afterAll() method, before sparkContext.stop(), add the following line Thread.sleep(Long.MaxValue).

While running a specific test, look for the following line in the output: 16/09/12 19:49:48 INFO SparkUI: Bound SparkUI to 0.0.0.0, and started at http://192.168.255.147:4040

Click the link to see the Spark UI.

Remember to remove the Thread.sleep when you're done, otherwise your tests will continue to hang indefinitely.

Good links