Can we use multiple git repos for a job running multiple tasks?

[thib]

I have a job running multiple tasks :

• Task 1 runs a machine learning pipeline from git repo 1
• Task 2 runs an ETL pipeline from git repo 1

Task 2 is actually a generic pipeline and should not be checked in repo 1, and will be made available in another repo where various ETL pipelines are stored.

However I see now that git repos are set up per Job, and not per Task.

Does anyone have a solution for this, which I would suppose is quite a common scenario?

[Prabakar]

hi @Thibault Daoulas​ if task2 is generic and doesnt depend on task1, could you try removing the depends on filed in task2 and run the pipeline?

[thib]

Hi @Prabakar Ammeappin​ , task2 depends on task1, as it sends out the results from task1.

I also tried removing the dependency and edited the git reference, but it changes the git reference for all the tasks in the job.

Do you know if that is a feature missing from jobs, or if it is intentional?

[trijit]

The way to go about this would be to create Databricks repos in the workspace and then use that in the task formation. This way we can refer multiple repos in different tasks.

[tors_r_us]

Had this same problem. Fix was to have two workflows with no triggers, each pointing to the respective git repo. Then setup a 3rd workflow with appropriate triggers/schedule which calls the first 2 workflows. 
A workflow can run other workflows. 

[jeremy98]

Yes, but this can cause an issue on waiting more time to have the cluster up.. How did u fix this?