Skip to main content
Prophecy requires you to install additional Python libraries on your Livy cluster to enable selective data sampling.

Prerequisites

Before installing the libraries, identify which Prophecy library version your project uses. The bundle version must match your project’s Prophecy library version. To check your project’s Prophecy library version:
  1. Open your project metadata.
  2. Click on the Dependencies tab.
  3. Find the ProphecyPythonLibs entry and note the version number.

Required libraries

The following Python libraries are required for selective sampling on Livy clusters running Spark 3.5:

Installation

Prophecy provides a bundled ZIP file containing all required dependencies. This simplifies installation by including all libraries in a single package. Access the bundle from the Prophecy public S3 bucket:
Replace <prophecy-libs-version> with the version number from your project dependencies.
The bundle version must match your project’s ProphecyPythonLibs version.
1

Set the fabric's resolution mode

You’ll need to configure your Livy fabric to use the bundled library file. There are two ways to do this.Option A: Use the Prophecy public bucket as the artifactory.
  • Set Resolution mode to Custom Artifactory.
  • Set the Artifactory URL to the provided bucket URL.
Livy resolution
mode
Option B: Host your own artifactory.
  • Download the bundle file from the bucket.
  • Upload it to your own infrastructure.
  • Set the Resolution mode to File System.
  • Set the Path to the location of the bundled library file.
Set the resolution mode to File System and provide the path to the bundled library file.
2

Verify installation

After installation, verify that the libraries are available by running a test import in a Python session on your Livy cluster.