AWS Glue Local Development with SSO Support on Amazon Linux 2023

This repository provides a streamlined setup for running AWS Glue ETL libraries locally with support for AWS SSO, based on Amazon Linux 2023, in a DevContainer. It resolves common challenges faced when configuring AWS Glue locally, as discussed in various resources.

Key Features

• DevContainer Integration: Fully configured for use with Visual Studio Code or compatible tools.
• AWS SSO Support: Properly handles AWS SSO credential management to facilitate seamless local testing.
• Jupyter Environment: Preconfigured Jupyter server for interactive development.
• Spark Configuration: Includes PySpark setup for ETL development.
• References and Improvements: Inspired by discussions and community solutions, such as:
  • Reddit discussion on AWS Glue with AWS SSO
  • AWS SDK Credentials Profile documentation
  • Glue Libs with SSO by Jerdoe

Getting Started

Prerequisites

• Visual Studio Code installed on your local machine.
• Docker installed and running.
• AWS SSO configuration in your ~/.aws/credentials or environment variables.
• Familiarity with AWS Glue and ETL processes.

Using the DevContainer

1. Clone this repository:

   git clone <repository-url>
   cd <repository-folder>

2. Open the repository in Visual Studio Code.

3. When prompted, open the folder in the DevContainer. Alternatively, you can manually rebuild the container:

   • Press Ctrl+Shift+P (or Cmd+Shift+P on Mac) to open the Command Palette.
   • Select Remote-Containers: Rebuild and Reopen in Container.

4. The DevContainer will start and initialize:

   • AWS credentials will be mounted into the container.
   • The Jupyter server will be started automatically.

Accessing JupyterLab

Once the DevContainer is running, JupyterLab will be accessible at http://localhost:8889. It is preconfigured with all necessary libraries and paths for AWS Glue development.

Key Configuration Details

• DevContainer: Configuration is defined in devcontainer.json for a seamless setup.
• Jupyter: Automatically starts on container launch via the postStartCommand script.
• Spark: Configured with paths to PyGlue.zip and py4j for full Glue functionality.
• Network Ports:
  • 4040: Spark UI
  • 18080: Spark History Server
  • 8998: Livy Server
  • 8889: JupyterLab

Customizing the Environment

Modify the devcontainer.json and Dockerfile to customize dependencies and environment configurations.

Contributing

Feel free to open issues or submit pull requests to improve this repository.