Model logging issues - Azure Databricks

[Ivan_]

I am running job with service principal ( principal e1077c1b-1dcf-4975-8164-d14f748d4774). Mentioned job has a task that will log (mlflow.pyfunc.log_model) newly trained model to unity catalog, but I get PERMISSION DENIED error. I do not get why principal has to be owner of the model? If it has all privileges, why do I get this error?

Pls help 🙂

Error

[Kaniz_Fatma]

Hi @Ivan_, 

• Check Artifact Permissions: Verify that the service principal has write permissions to the artifact store location where you’re trying to log the model.
• Experiment Permissions: With the extension of MLflow experiment permissions to artifacts, explicit permission is required to access experiment artifacts.
• Experiment Owner Permissions: If you’re not the experiment owner, ask the owner to grant you the app...¹.
• Remember, even though the service principal has privileges, the artifact store permissions are separate and crucial for successful artifact logging. Ensure that the service principal has the necessary rights to write to the specified location, and you should be able to resolve the PERMISSION DENIED issue. 🚀

[Ivan_]

@Kaniz_Fatma thanks for reply, but can you explain how can I check (set) artifact permissions?

Experiment permissions are set correctly (mentioned principal has needed permissions), but  how do I change artifact (store) permissions?

This is artifact path: dbfs:/databricks/mlflow-tracking/2878245870427067/945abed8808c4ed6961da2008290b8c3/artifacts/artifacts/item_recommender_12754.

How can I access artifact store permissions?