Skip to content
This repository has been archived by the owner on Jul 17, 2024. It is now read-only.

noklam/kedro-softfail-runner

Repository files navigation

IMPORTANT NOTE: kedro-softfail-runner HAS MOVEVD TO A NEW HOME. THIS REPO IS DEPRECATED.

kedro-softfail-runner

!!!!! Deprecated, for kedro>0.19 support it is moved to https://github.com/noklam/kedro-runners as a collection of runners

This file will become your README and also the index of your documentation.

Install

pip install kedro_softfail_runner

How to use

Fill me in please! Don’t forget code examples:

kedro run --runner=kedro_softfail_runner.SoftFailRunner

Why is it useful?

The team created a soft-fail runner to transform errors into warnings, allowing the pipeline to continue executing to the best of its ability while providing a report of any nodes that failed, so that data issues can be addressed. At that point, the pipeline run can be finalised by executing only those missing nodes separately, using appropriate Kedro syntax. https://kedro.org/blog/build-a-custom-kedro-runner

It’s most useful in two different scenarios: 1. Development - you want to detect all problematic nodes in one go 2. Deployment - you want to run as much node as you can before the Kedro pipeline is stopped.

Why is it not in Kedro yet?

Note that the SoftFailRunner does not return anything, that is result = session.run() normally returns a dictionary of free output, but it will be always None for SoftFailRunner due to implementation problem.

Other than that, there are no other known issues yet. In order to merge this into the core library, this problem need to be fixed and the runner should be test thoroughly.