Skip to content

Contributing to KubeStellar#

Greetings! We are grateful for your interest in joining the KubeStellar community and making a positive impact. Whether you're raising issues, enhancing documentation, fixing bugs, or developing new features, your contributions are essential to our success.

To get started, kindly read through this document and familiarize yourself with our code of conduct. If you have any inquiries, please feel free to reach out to us on the KubeStellar-dev Slack channel.

We can't wait to collaborate with you!

Contributing Code#

Prerequisites#

Go#

Install Go 1.19+. See this gist for another way to install Go. Please note that the go language version numbers in these files must exactly agree:

Your local go/go.mod file, kcp/.ci-operator.yaml, and in all the kcp/.github/workflows yaml files that specify go-version.

- In ./ci-operator.yaml the go version is indicated by the "tag" attribute.
- In go.mod it is indicated by the "go" directive.
- In the .github/workflows yaml files it is indicated by "go-version"

Check out our QuickStart Guide

Other packages#

  • GNU make
  • ko (required for compiling KubeStellar Syncer)
  • slsa-verifier needed in Ubuntu for ko signing

Issues#

Prioritization for pull requests is given to those that address and resolve existing GitHub issues. Utilize the available issue labels to identify meaningful and relevant issues to work on.

If you believe that there is a need for a fix and no existing issue covers it, feel free to create a new one.

As a new contributor, we encourage you to start with issues labeled as good first issues.

Your assistance in improving documentation is highly valued, regardless of your level of experience with the project.

To claim an issue that you are interested in, kindly leave a comment on the issue and request the maintainers to assign it to you.

Committing#

We encourage all contributors to adopt best practices in git commit management to facilitate efficient reviews and retrospective analysis. Your git commits should provide ample context for reviewers and future codebase readers.

A recommended format for final commit messages is as follows:

{Short Title}: {Problem this commit is solving and any important contextual information} {issue number if applicable}

Pull Requests#

When submitting a pull request, clear communication is appreciated. This can be achieved by providing the following information:

  • Detailed description of the problem you are trying to solve, along with links to related GitHub issues
  • Explanation of your solution, including links to any design documentation and discussions
  • Information on how you tested and validated your solution
  • Updates to relevant documentation and examples, if applicable

The pull request template has been designed to assist you in communicating this information effectively.

Smaller pull requests are typically easier to review and merge than larger ones. If your pull request is big, it is always recommended to collaborate with the maintainers to find the best way to divide it.

Approvers will review your PR within a business day. A PR requires both an /lgtm and then an /approve in order to get merged. You may /approve your own PR but you may not /lgtm it. Automation will add the PR it to the OpenShift PR merge queue. The OpenShift Tide bot will automatically merge your work when it is available.

Congratulations! Your pull request has been successfully merged! 👏

If you have any questions about contributing, don't hesitate to reach out to us on the KubeStellar-dev Slack channel.

Testing Locally#

Our QuickStart guide shows a user how to install a local KCP server and install the KubeStellar components and run an example. As a contributor you will want a different setup flow, including git clone of this repo instead of fetching and unpacking a release archive. The same example usage should work for you, and there is a larger example at this link.

Testing changes to the KubeStellar central container image#

If you make a change that affects the container image holding the central components then you will need to build a new image; perhaps surprisingly, this is not included in make build. The regular way to build this image is with the following command. It builds a multi-platform image, for all the platforms that KubeStellar can run its central components on, and pushes it to quay.io. Read the remarks below before trying this.

make kubestellar-image

The set of target platforms can be specified by setting the CORE_PLATFORMS variable. The following command is equivalent to the default behavior.

make kubestellar-image CORE_PLATFORMS=linux/amd64,linux/arm64,linux/ppc64le

NOTE VERY SHARP AND BURIED EDGE: IF the target platforms include linux/amd64 --- either because you explicitly set that or you let the default setting apply --- then you MUST issue this command on a machine (real or virtual) with the x86-64-v2 instructions. "x86-64-v2" is a shorthand for a bundle of instruction set features that have been appearing in x86 chips for many years now (any real machine that you are likely to use today has them) but still do not all appear by default in some common emulators. See QEMU configuration recommendations, for example. If the machine lacks the v2 instructions then the build will fail when it tries to use the glibc in the redhat/ubi9 image. Cross-platform building when the builder is NOT x86 and the target IS x86 is beyond the ken of modern technology (see here and here). If you somehow succeed to build for the target platform linux/amd64/v2 and successfully test on real x86 hardware you still are not done: when you try to use this image in OpenShift on x86 you may get inexplicable failures to pull the image.

The command shown above will only succeed if you have done docker login to quay.io with credentials authorized to write to the kubestellar/kubestellar repository. Look on quay.io to find the image you just pushed, you will soon need to use one of its tags. This make target pushes the image with two tags, one based on build timestamp and one based on git metadata.

If you are not authorized to write to quay.io/kubestellar/kubestellar then you can specify an alternate image repository: put it in the make variable named CORE_IMAGE_REPO. For example, you might invoke make kubestellar-image CORE_IMAGE_REPO=docker.io/myacct/ksctr.

Another variable that you might like to use is EXTRA_CORE_TAG. This causes the make command to push the image with a third tag that you supply in that variable. For example, if you want to tag the image with a release tag you might invoke make kubestellar-image EXTRA_CORE_TAG=v0.42.7.

For a less pushy alternative you can build a single-platform image and not push it, using the following command. It also supports the CORE_IMAGE_REPO and EXTRA_CORE_TAG variables. But it only builds for your local "platform"; you can use this if you have podman pretending to be docker.

make kubestellar-image-local

Follow that with docker images to find the tag of the image you just built. Get that image:tag known where you are going to run the central container; for example, if that will be in a local kind cluster then you can use kind load.

To get the image you just built used in your testing, edit outer-scripts/kubectl-kubestellar-deploy and update the line that defines $image_tag; follow this with your make build. For the sake of future users of a merged change, your last edit like this should refer to a tag that you pushed to quay.io/kubestellar/kubestellar.

Testing changes to the bootstrap script#

The quickstart says to fetch the bootstrap script from the release-0.14 branch of the KubeStellar repo; if you want to contribute a change to that script then you will need to test your changed version. Just run your local copy (perhaps in a special testing directory, just to be safe) and be sure to add the downloaded bin at the front of your $PATH (contrary to what the scripting currently tells you) so that your git clone's bin does not shadow the one being tested.

Note that changes to the bootstrap script start being used by users as soon as your PR merges. Since this script can only fetch a released version of the executables, changes to this script can not rely on any behavior of those executables that is not in the currently latest release. Also, a change that restricts the range of usable releases needs to add checking for use of incompatible releases.

Testing the bootstrap script against an upcoming release#

Prior to making a new release, there needs to be testing that the current bootstrap script works with the executable behavior that will appear in the new release. To support this we will add an option to the bootstrap script that enables it to use a local release archive instead of fetching an archive of an actual release from github.

Licensing#

KubeStellar is Apache 2.0 licensed and we accept contributions via GitHub pull requests.

Please read the following guide if you're interested in contributing to KubeStellar.

Certificate of Origin#

By contributing to this project you agree to the Developer Certificate of Origin (DCO). This document was created by the Linux Kernel community and is a simple statement that you, as a contributor, have the legal right to make the contribution. See the DCO file for details.