The Apache Superset community extensively uses Docker for development, release, and productionizing Superset. This page details our Docker builds and tag naming schemes to help users navigate our offerings.

Images are built and pushed to the Superset Docker Hub repository. Different sets of images are created for:

  • Published releases (release): with tags like 3.0.0 and the latest tag. Those are multi-platform (arm+amd). More on that later.
  • Pull request iterations (pull_request):, each identified by tags starting with a SHA like 8a2f7d378ab13c156fa183d9284b607ed69f5ecc, and pr-3454, referencing the pull request ID.
  • Merges to the main branch (push): resulting in new SHAs, with tags prefixed with master for the latest master version.

Each CI build run has multiple builds for different purposes, identified by suffixes:

  • Build preset: We offer various images for different needs:
    • lean: The default Docker image, including both frontend and backend. Tags without a build_preset are lean builds, e.g., latest.
    • dev: For development, with a headless browser and root access.
    • py310, e.g., Py310: Similar to lean but with a different Python version (in this example, 3.10).
    • ci: For certain CI workloads.
    • websocket: For Superset clusters supporting advanced features.
    • dockerize: Used by Helm.
  • Platform: We build for linux/arm64 and linux/amd64. The -arm suffix indicates ARM builds (e.g., latest-arm), while tags without a suffix are for AMD (e.g., latest).

Key Image Tags and Examples

  • latest: The latest official release build, implicitly the lean build on linux/amd64.
  • latest-dev: the -dev image of the latest official release build, with a headless browser and root access.
  • master: The latest build from the master branch, implicitly lean on linux/amd64.
  • master-dev: Similar to master but includes a headless browser and root access.
  • pr-5252: The latest commit in PR 5252.
  • 30948dc401b40982cb7c0dbf6ebbe443b2748c1b-dev-arm: A linux/arm64 build for this specific SHA, which could be from a pull request, master merge, or release.
  • 30948dc-dev-arm: Same as above, but SHA truncated to 7 characters for a shorter handle on the same image
  • websocket-latest: The WebSocket image for use in a Superset cluster.

For insights or modifications to the build matrix and tagging conventions, check the build_docker.py script and the docker.yml GitHub action.

Caching

To accelerate builds, we follow Docker best practices and use apache/superset-cache.

About database drivers

Our docker images come with little to zero database driver support since each envrionment requires different drivers, and mataining a build with wide database support would be both challenging (dozens of databases, python drivers, and os dependencies) and inefficient (longer build times, larger images, lower layer cache hit rate, …).

For production use cases, we recommend that you derive our lean image(s) and add database support for the database you need.

On supporting arm64 AND amd64

Only the release builds are multi-platform, supporting linux/arm64 and linux/amd64. This enables higher level constructs like helm and docker-compose to point to these images and effectively be multi-platform as well.

Pull requests and master builds are one-image-per-platform so that they can be parallized and the build matrix for those is more sparse as we don’t need to build every build preset on every platform, and generally can be more selective here. For those builds, we suffix tags with -arm where it applies.

Working with Apple silicon

Apple’s current generation of computers uses ARM-based CPUs, and Docker running on MACs seem to require linux/arm64/v8 (at least one user’s M2 was configured in that way). Setting the environment variable DOCKER_DEFAULT_PLATFORM to linux/amd64 seems to function in term of leveraging, and building upon the Superset builds provided here.

  1. export DOCKER_DEFAULT_PLATFORM=linux/amd64

Presumably, linux/arm64/v8 would be more optimized for this generation of chips, but less compatible across the ARM ecosystem.