Job deployments fail with "The job must have exactly one owner." error

[russell-mcdaniel]

Describe the issue

Beginning with Databricks CLI 0.274.0, all of our job deployments fail with the following error:

Deploying resources...
Error: Terraform error during deployment: exit status 1

Error: cannot update permissions: The job must have exactly one owner.

  with databricks_permissions.job_[job name redacted],
  on bundle.tf.json line 2095, in resource.databricks_permissions.job_[job name redacted]:
2095:       },

Version 0.275.0 exhibits the same behavior. We've reverted to 0.273.0 in the meantime.

Configuration

We use Databricks in Azure. Bundles are deployed using a Microsoft Entra ID service principal that is configured as a Databricks workspace administrator. Databricks job owners are also Microsoft Entra ID service principals (non-administrator). This error can also be reproduced directly by developers using their Entra ID user accounts.

Steps to reproduce the behavior

1. Run databricks bundle deploy ...
2. See error.

Expected Behavior

All jobs should deploy successfully with no error.

Actual Behavior

All jobs fail to deploy with the error specified in the description.

OS and CLI version

The error is observed on both Windows and Linux beginning with Databricks CLI 0.274.0.

Is this a regression?

Yes. This error does not occur in Databricks CLI 0.273.0 or earlier.

Debug Logs

Bundle source and logs can be provided upon request; there is too much to redact to post publicly.

[denik]

Thanks for the report! Could you sent databricks.yml + included yml and debug logs if you have them to dabs-feedback@databricks.com

[denik]

Fixed by #3850 (will be released next week).