DLT-Asset bundle : Pipelines do not support a setting a run_as user that is different from the owner

[erigaud]

Hello !

We're using Databricks asset bundles to deploy to several environments using a devops pipeline. The service principal running the CICD pipeline and creating the job (owner) is not the same as the SPN that will be running the jobs (run_as).

This is not an issue for workflows but somehow it is an issue for DLT pipelines as we run into this error : 

Pipelines do not support a setting a run_as user that is different from the owner

Is there a way to solve this issue or is it going to be supported by the DLT pipelines ? It seems like a very big constraint.

[saurabh18cs]

Hi @erigaud 

yes observed same behaviour either remove run_as and it will automatically take the same information as owner.

OR

if owner can be mapped to run_as

[NandiniN]

By default who creates pipeline is the owner/run_as.

However, yes this is a feature in progress.

I wouls still encorage you to share your request her https://docs.databricks.com/en/resources/ideas.html#ideas 

[erigaud · Accepted solution]

Hello again,

I've created an idea for that issue if someone wants to give it an upvote. It can be a very blocking issue for architectures where you cannot allow to use the same SPN for deployment and runs

https://feedback.azure.com/d365community/idea/61e1a9e4-99e3-ef11-b542-00224854717c

[no2]

I can't believe I have to say this, but this issue should not be marked as "Solved" until the actual problem is... solved. 

[Coffee77]

Maybe I'm not catching this or missing something else but I've got the following job in one of my demo workspaces:

Creator is my user and the job runs as a service principal account. Those are different identities. I got this by deploying the job with Databricks Asset Bundles and changing in there the "run_as" property while keeping same job identifiers.