Installing and Configuring Spark Operator

This section describes how to install and configure Spark Operator on HPE Ezmeral Runtime Enterprise.

Prerequisites

  • Log in as a Kubernetes Cluster Administrator or Platform Administrator in HPE Ezmeral Runtime Enterprise.

About this task

In HPE Ezmeral Runtime Enterprise, you can install Spark Operator using GUI or manually using the Helm chart.

Learn more about supported Spark versions by Spark Operator at Interoperability Matrix for Spark.

Installing Spark Operator Using the GUI

About this task

Install Spark Operator during the Kubernetes Cluster creation step using the HPE Ezmeral Runtime Enterprise GUI. See Creating a New Kubernetes Cluster.

Procedure

  1. Set up Host Configurations, Cluster Configurations, and Authentication for Kubernetes Cluster.
  2. In Application Configurations, select Enable Spark Operator.
  3. Click Next, review the summary of resources to be assigned to Kubernetes cluster.
  4. To install the Kubernetes cluster, click Submit. This triggers the installation process for Spark Operator.

Results

The GUI installs the Spark Operator.

Starting from HPE Ezmeral Runtime Enterprise 5.4.0, selecting Enable Spark Operator option does not trigger the installation for Livy, Spark History Server, Spark Thrift Server, and Hive Metastore. You must install Livy, Spark History Server, Spark Thrift Server, and Hive Metastore using the GUI or manually using the helm charts.

Installing Spark Operator Using the Helm

Prerequisites

  1. Install and configure Helm 3.
  2. Install and configure kubectl.

About this task

Install the Spark Operator by using the Helm chart.

Procedure

To install the Spark Operator for Apache Spark in an existing namespace, run the following command: 
helm install -f <path-to-values.yaml-file> <spark-operator-name> ./<path-to-spark-operator-chart>/ \
--namespace <cluster-namespace> \
--set sparkJobNamespace=<cluster-namespace> \
--set webhook.namespaceSelector=hpe.com/tenant=<cluster-namespace> \
--set fullnameOverride=<spark-operator-name>
Autoticket generator webhook is not installed by default in the cluster namespace. To enable the installation of autoticket generator, set the following flag:
--set autotix.enable=true

Running the helm install installs Spark Operator in a cluster namespace.

Example

To install Spark Operator in the compute namespace, run the following command:
helm install -f spark-operator-chart/values.yaml spark-operator-compute ./spark-operator-chart/ \
--namespace compute \
--set sparkJobNamespace=compute \
--set webhook.namespaceSelector=hpe.com/tenant=compute \
--set fullnameOverride=spark-operator-compute