Pytest-xdist-kubernetes

The pytest-xdist-plugin extends pytest-xdist with new kubernetes pod communication. It is capable of creating Kubernetes deployment using the given namespace and docker image, run tests from pods, and display the result at the end from the terminal.

How to Install

In order to use the plugin, >= Python 3.9 is required. Can be installed with the following command:

pip install pxk

How to Use

By giving --tx='pod' as part of the xdist command, it triggers Pytest-xdist-kubernetes plugin.

pytest {test files to run} -n {number of pods per deployment} --tx='pod'

On top of the existing pytest-xdist library, various options have been added to support running tests from kubernetes pods remotely.

pytest --namespace='{custom namspace}' --custom_image='{custom image}' {test files to run} -n {number of pods per deployment} --tx='pod'

When there are multiple images, multiple deployments get created and each deployment will contain its own image. The plugin will then try its best to evenly distribute the workers as much as possible. If the number of workers is less than the number of custom images, an exception would occur. Multiple custom images can be provided through the command without spaces in between images:

pytest --namespace='{custom namspace}' --custom_image='{custom image1,custom image2}' {test files to run} -n {number of pods per deployment} --tx='pod'

Logger is included as part of the functions added by the plugin. You can check the progress of the plugin by specifying:

--log-cli-level INFO

Since the plugin relies on xdist library's task scheduler for distributing tasks, you can specify how you would like to distribute tasks across multiple pods. By default, it evenly distributes test files.
But if you would like each pod to run all the specified test files:

pytest {testfiles to run} -n {number of pods per deployment} --tx='pod' --dist=each

All the testfiles must be from the same folder.
The plugin may have issues in Windows OS.

Reference

The plugin uses Kubernetes API to create/delete kubernetes deployments.
https://github.com/kubernetes-client/python/blob/master/kubernetes/README.md