Releasing Apache Kafka

This directory contains the tools used to publish a release.

Requirements

• python 3.12
• git
• gpg 2.4

The full instructions for producing a release are available in https://cwiki.apache.org/confluence/display/KAFKA/Release+Process.

Setup

Create a virtualenv for python, activate it and install dependencies:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

Usage

To start a release, first activate the virtualenv, and then run the release script.

source .venv/bin/activate

You'll need to setup PUSH_REMOTE_NAME to refer to the git remote for apache/kafka.

export PUSH_REMOTE_NAME=<value>

It should be the value shown with this command:

git remote -v | grep -w 'github.com' | grep -w 'apache/kafka' | grep -w '(push)' | awk '{print $1}'

Then start the release script:

python release.py

Should you encounter some problem, where re-running the script doesn't work, look at the following steps:

• The script remembers data inputted previously if you need to correct it, it is saved under the .release-settings.json file in the release folder.
• If the script is interrupted you might need to manually delete the tag named after the release candidate name and branch named after the release version.

Docker workflow triggers

After the RC tag is pushed, the script triggers the Docker image build/test and RC release workflows on GitHub Actions.

GitHub Personal Access Token

Triggering the workflows requires a GitHub Personal Access Token. To generate one:

1. Go to https://github.com/settings/tokens
2. Click “Generate new token” → “Generate new token (classic)”
3. Set a name (e.g. kafka-release)
4. Set an expiration (7 days is sufficient for a release cycle)
5. Select the repo scope (this includes the actions write permission)
6. Click “Generate token” and copy the token (starts with ghp_...)

The token is cached in .release-settings.json so it only needs to be entered once per release cycle. To reset the saved token, remove the github_token entry from .release-settings.json or delete the file entirely.

Optional environment variables

• GITHUB_REPO: target repository for the workflow dispatches. Defaults to apache/kafka. Set this to your fork (e.g. myuser/kafka) when testing the release script end-to-end without affecting apache/kafka.
• GITHUB_DRY_RUN: when set to true, prints the GitHub API calls that would be made instead of executing them. Useful for verifying the flow without a token or network access.

GITHUB_DRY_RUN=true GITHUB_REPO=myuser/kafka python release.py