renku-frontend-buildpacks

This project provides a comprehensive set of Cloud Native Buildpacks and a specialized selector builder designed to streamline the deployment of Renku frontend applications and development environments. Whether you're setting up JupyterLab, RStudio, or Vscodium in your Renku project, these buildpacks simplify the process by intelligently detecting your project's needs and configuring the appropriate environment.

The selector builder acts as an orchestrator, integrating various frontend frameworks and essential tools (like Python dependency management and Jupyter kernel installers) to create ready-to-use images. This allows Renku users to focus on their data science work without deep knowledge of underlying containerization.

For automated image building within your CI/CD pipelines, you can leverage the provided GitHub Action actions/build-image. This action simplifies the pack build process, allowing you to easily specify the desired frontend, tags, builder-version, and run-image directly in your workflows.

To get started with manual builds, you'll primarily interact with the pack CLI, using the selector builder to build your project. For example:

pack build my-user/my-renku-environment --builder ghcr.io/swissdatasciencecenter/renku-frontend-buildpacks/selector:0.1.0 --path .

This command will leverage the selector builder (using version 0.1.0 as an example) to automatically detect and configure your Renku environment based on your project's files.

Directory Structure

• builders: Contains builder definitions. A builder defines the environment and buildpacks used to build an application. For now we only maintain the selector builder.
• buildpacks: Contains individual buildpacks for different frontend frameworks. Each buildpack provides the necessary scripts and configurations to detect and build applications.
• jupyterlab: Buildpack for JupyterLab frontend.
• kernel-installer: Buildpack for installing the correct kernel for the environment.
• samples: Contains sample applications for different frontend frameworks. These samples can be used to test the buildpacks and builders.

Makefile Targets

The Makefile provides several targets for building and running the project:

• all: Builds buildpacks, builders, and sample images.
• buildpacks: Builds all buildpacks defined in the buildpacks directory using pack buildpack package.
• builders: Builds all builders defined in the builders directory using pack builder create.
• samples: Builds sample images using the buildpacks and builders. It utilizes the pack build command with the specified builder image and environment variables.
• run: Runs a sample image with Docker, publishing port 8000.

Building the Project

Note you have to have experimental features enabled on the pack CLI in order to be able to use and build image extensions.

pack config experimental true

To build the project, run:

make

This will build all buildpacks, builders, and sample images.

To build only the buildpacks, run:

make buildpacks

To build only the builders, run:

make builders

To build only the sample images, run:

make samples

Please note that the builders must be set and that you may set the BUILDER_IMAGE variable to select a builder

make samples BUILDER_IMAGE=selector

Running a Sample Image

To run a sample image, use the run target:

make run

You can set the SAMPLE_IMAGE and FRONTEND. For example:

make run SAMPLE_IMAGE=conda FRONTEND=jupyterlab

Dependencies

This project requires the following tools:

• pack
• Docker