Is there any way to run jobs from github actions and catch the results?

[Yuki]

Hi all,

Is there any way to run jobs from github actions and catch the results?

Of course, I can do this if I use the API or CLI.
But I found the actions for notebook: https://github.com/marketplace/actions/run-databricks-notebook  

Compared to this, writing code from scratch for jobs is complicated. Is there a better way?

[ManojkMohan]

@Yuki 

you can run Databricks jobs from GitHub Actions and capture their results using the "Run Databricks Notebook" GitHub Action. This tool abstracts much of the manual code and API work, making integration with your CI/CD pipeline easier

https://github.com/marketplace/actions/run-databricks-notebook

The databricks/run-notebook GitHub Action allows you to trigger Databricks notebooks directly from workflow YAML files.

You can configure your workflow to parse and utilize output values for subsequent automation.

The Action awaits completion, so you can perform checks or notifications

REST API or CLI from a custom GitHub Actions workflow step requires more manual coding

[Yuki]

@ManojkMohan 
Thank you for your response, but I knew that.
My problem is that I can't find similar tools for "Databricks Jobs & Pipelines".

[ManojkMohan]

@Yuki There is currently no GitHub Action that provides a no-code experience in the same way as the "Run Databricks Notebook" Action does for individual notebooks.

Alternate work arounds:

1. Databricks Asset Bundles + GitHub Actions
Package job and pipeline definitions as YAML files using Databricks Asset Bundles (DABs).

Use CLI automation within GitHub Actions

https://docs.databricks.com/aws/en/dev-tools/ci-cd/

use Azure Data Factory to trigger Databricks jobs, with monitoring and outputs passed back to GitHub Actions as part of a broader workflow

https://learn.microsoft.com/en-us/answers/questions/5514533/databricks-repos-or-alternatives

[Yuki]

OK, thank you for your advices, I will consider to use asset bundles for this.