Deletion of managed MLflow Artifacts

Christoph Stumpf 1 Reputation point
2020-11-20T08:50:37.143+00:00

When using Workspace experiments in Azure Databricks with the default managed MLflow artifact location dbfs:/databricks/mlflow-tracking configured, we see the following message when deleting an MLflow experiment run:

Deleted runs are restorable for 30 days, after which they are purged along with associated metrics, params and tags.
Artifacts are not automatically purged and must be manually deleted.

which suggests that the artifacts are not automatically deleted. Does this also hold true for the managed artifact location?

If so, how are we supposed to delete these artifacts as access via dbutils or DBFS API are prohibited for this location (see docs) and the MLflow client does not support deletion of artifacts?

Thanks!

Azure Databricks
Azure Databricks
An Apache Spark-based analytics platform optimized for Azure.
2,150 questions
{count} votes

Your answer

Answers can be marked as Accepted Answers by the question author, which helps users to know the answer solved the author's problem.