Skip to content
Run Apache Spark on Openshift
Shell Dockerfile
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
docker-images Fixed autocompletion problem (#17) Sep 18, 2019
LICENSE Initial commit Aug 16, 2018 Extrastorage (#5) Dec 3, 2018 Update custom libaries installation docs Nov 22, 2018
spark-template.yml Update image tag to 2.4 (#16) Jul 16, 2019


Run Apache Spark on Openshift. Based on


The template provisions a Spark cluster based on the configuration on Openshift. NOTE: Make sure, you create a new openshift project in order to run this. It is also recommended that you have only one cluster per openshift project

If you are looking for instructions on how to install custom spark libraries on top of your spark cluster, click here

There are 3 deployments created -

  1. Spark Master: Serves as the master of the cluster and runs Spark UI
  2. Spark Worker: Runs worker instances in the cluster
  3. Jupyter Notebook: Serves as the Spark Driver, where one writes the code and submits it to the master


Listed below are some of the variables that should be changed.

Please NOTE : The values for the CPU and Memory should only be changed (to avoid errors) after checking the project quota allocated to your Openshift project. You should increase or request the admins to increase it for you, if needed.

Mandatory Required Values:

  • Cluster Name: Unique identifier for your cluster

  • Username: Username for authenticating and logging into your Spark cluster and Jupyter (Recommended: create a new username, don't use any existing one)

  • Password: Password for authenticating and logging into your Spark cluster and Jupyter (Recommended: create a new password, don't use any existing one)

  • Worker Replicas: Number of workers to have (Default: 4)

  • Storage Size: Persistent storage volume size (Default: 10Gi)

Optional Required Values:

  • Enable Jupyter Lab: Specify whether if you want to use Jupyter Lab instead of the default Jupyter Notebook (Default: false)

  • Master CPU: Number of cores for the master node of the cluster

  • Master Memory: Memory for the master node of the cluster

  • Worker CPU: Number of cores for each worker of the cluster (Default: 2)

  • Worker Memory: Memory of each worker of the cluster (Default: 4G)

  • Executor Default Cores: Default value for Spark Executor Cores (See official Spark documention for more) (Default: 2)

  • Executor Default Memory: Default value for Spark Executor Memory (Should always be less than the Worker memory!) (Default: 3G)

  • Driver CPU: Number of cores for the driver (Jupyter Notebook)

  • Driver Memory: Memory of the driver (Jupyter Notebook)

Do not change the following variables, unless you know what you're doing

  • Master Image: Docker Image for the Master
  • Worker Image: Docker Image for the Worker
  • Driver Image: Docker Image for the Driver
  • Application Hostname Suffix: The exposed hostname suffix that will be used to create routes for Spark UI and Jupyter Notebook

NOTE: The template assumes that the request and the limits are same for all the containers. If you wish to have different limits, it's recommended to edit the template

If running through the Command line:

  • Download the oc client for openshift
  • oc login
  • oc new-project
  • Change the variables
  • oc process -f spark-template.yml | oc apply -f -


  • oc delete all -l app=spark
  • oc delete configmap -l app=spark
  • oc delete secret -l app=spark
  • You might also want to delete the persistent volume created by the setup by oc delete pvc -l app=spark

Adding more storage from OpenShift UI

From OpenShift console

  • open Storage -> Create Storage -> Fill required fields and press Create button.
  • Application -> Deployments -> For each of the items go Configuration tab -> Add Storage -> Fill desired Mount Path ie. /mnt/data -> Type Volume Name or leave empty for automatically generated -> press Add.

Automatic redeployment starts and after repeating above steps to all items, new pvc will be mounted to application.

You can’t perform that action at this time.