How to Install Spark on Ubuntu

Introduction

Apache Spark is a framework used in cluster computing environments for analyzing big data. This platform became widely popular due to its ease of use and the improved data processing speeds over Hadoop.

Apache Spark is able to distribute a workload across a group of computers in a cluster to more effectively process large sets of data. This open-source engine supports a wide array of programming languages. This includes Java, Scala, Python, and R.

In this tutorial, you will learn how to install Spark on an Ubuntu machine. The guide will show you how to start a master and slave server and how to load Scala and Python shells. It also provides the most important Spark commands.

Prerequisites

• An Ubuntu system.
• Access to a terminal or command line.
• A user with sudo or root permissions.

Install Packages Required for Spark

Before downloading and setting up Spark, you need to install necessary dependencies. This step includes installing the following packages:

• JDK
• Scala
• Git

Open a terminal window and run the following command to install all three packages at once:

sudo apt install default-jdk scala git -y

You will see which packages will be installed.

Once the process completes, verify the installed dependencies by running these commands:

java -version; javac -version; scala -version; git --version

The output prints the versions if the installation completed successfully for all packages.

Download and Set Up Spark on Ubuntu

Now, you need to download the version of Spark you want form their website. We will go for Spark 3.0.1 with Hadoop 2.7 as it is the latest version at the time of writing this article.

Use the wget command and the direct link to download the Spark archive:

wget https://downloads.apache.org/spark/spark-3.0.1/spark-3.0.1-bin-hadoop2.7.tgz

When the download completes, you will see the saved message.

Now, extract the saved archive using tar:

tar xvf spark-*

Let the process complete. The output shows the files that are being unpacked from the archive.

Finally, move the unpacked directory spark-3.0.1-bin-hadoop2.7 to the opt/spark directory.

Use the mv command to do so:

sudo mv spark-3.0.1-bin-hadoop2.7 /opt/spark

The terminal returns no response if it successfully moves the directory. If you mistype the name, you will get a message similar to:

mv: cannot stat 'spark-3.0.1-bin-hadoop2.7': No such file or directory.

Configure Spark Environment

Before starting a master server, you need to configure environment variables. There are a few Spark home paths you need to add to the user profile.

Use the echo command to add these three lines to .profile:

echo "export SPARK_HOME=/opt/spark" >> ~/.profile
echo "export PATH=$PATH:$SPARK_HOME/bin:$SPARK_HOME/sbin" >> ~/.profile
echo "export PYSPARK_PYTHON=/usr/bin/python3" >> ~/.profile

You can also add the export paths by editing the .profile file in the editor of your choice, such as nano or vim.

For example, to use nano, enter:

nano .profile

When the profile loads, scroll to the bottom of the file.

Then, add these three lines:

export SPARK_HOME=/opt/spark

export PATH=$PATH:$SPARK_HOME/bin:$SPARK_HOME/sbin

export PYSPARK_PYTHON=/usr/bin/python3

Exit and save changes when prompted.

When you finish adding the paths, load the .profile file in the command line by typing:

source ~/.profile

Start Standalone Spark Master Server

Now that you have completed configuring your environment for Spark, you can start a master server.

In the terminal, type:

start-master.sh

To view the Spark Web user interface, open a web browser and enter the localhost IP address on port 8080.

http://127.0.0.1:8080/

The page shows your Spark URL, status information for workers, hardware resource utilization, etc.

The URL for Spark Master is the name of your device on port 8080. In our case, this is ubuntu1:8080. So, there are three possible ways to load Spark Master’s Web UI:

1. 127.0.0.1:8080
2. localhost:8080
3. deviceName:8080

Start Spark Slave Server (Start a Worker Process)

In this single-server, standalone setup, we will start one slave server along with the master server.

To do so, run the following command in this format:

start-slave.sh spark://master:port

The master in the command can be an IP or hostname.

In our case it is ubuntu1:

start-slave.sh spark://ubuntu1:7077

Now that a worker is up and running, if you reload Spark Master’s Web UI, you should see it on the list:

Specify Resource Allocation for Workers

The default setting when starting a worker on a machine is to use all available CPU cores. You can specify the number of cores by passing the -c flag to the start-slave command.

For example, to start a worker and assign only one CPU core to it, enter this command:

start-slave.sh -c 1 spark://ubuntu1:7077

Reload Spark Master’s Web UI to confirm the worker’s configuration.

Similarly, you can assign a specific amount of memory when starting a worker. The default setting is to use whatever amount of RAM your machine has, minus 1GB.

To start a worker and assign it a specific amount of memory, add the -m option and a number. For gigabytes, use G and for megabytes, use M.

For example, to start a worker with 512MB of memory, enter this command:

start-slave.sh -m 512M spark://ubuntu1:7077

Reload the Spark Master Web UI to view the worker’s status and confirm the configuration.

Test Spark Shell

After you finish the configuration and start the master and slave server, test if the Spark shell works.

Load the shell by entering:

spark-shell

You should get a screen with notifications and Spark information. Scala is the default interface, so that shell loads when you run spark-shell.

The ending of the output looks like this for the version we are using at the time of writing this guide:

Type :q and press Enter to exit Scala.

Test Python in Spark

If you do not want to use the default Scala interface, you can switch to Python.

Make sure you quit Scala and then run this command:

pyspark

The resulting output looks similar to the previous one. Towards the bottom, you will see the version of Python.

To exit this shell, type quit() and hit Enter.

Basic Commands to Start and Stop Master Server and Workers

Below are the basic commands for starting and stopping the Apache Spark master server and workers. Since this setup is only for one machine, the scripts you run default to the localhost.

To start a master server instance on the current machine, run the command we used earlier in the guide:

start-master.sh

To stop the master instance started by executing the script above, run:

stop-master.sh

To stop a running worker process, enter this command:

stop-slave.sh

The Spark Master page, in this case, shows the worker status as DEAD.

You can start both master and server instances by using the start-all command:

start-all.sh

Similarly, you can stop all instances by using the following command:

stop-all.sh

Conclusion

This tutorial showed you how to install Spark on an Ubuntu machine, as well as the necessary dependencies.

The setup in this guide enables you to perform basic tests before you start configuring a Spark cluster and performing advanced actions.

Our suggestion is to also learn more about what Spark DataFrame is, the features, how to use Spark DataFrame when collecting data and how to create a Spark DataFrame.