MLflow API 2.0
Azure Databricks provides a managed version of the MLflow tracking server and the Model Registry, which host the MLflow REST API. You can invoke the MLflow REST API using URLs of the form
https://<databricks-instance>/api/2.0/mlflow/<api-endpoint>
replacing <databricks-instance> with the workspace URL of your Azure Databricks deployment.
Migration guide lists the MLflow release packaged in each Databricks Runtime version and a link to the respective documentation.
Important
To access Databricks REST APIs, you must authenticate.
Rate limits
The MLflow APIs are rate limited as four groups, based on their function and maximum throughput. The following is the list of API groups and their respective limits in qps (queries per second):
- Low throughput experiment management (list, update, delete, restore): 7 qps
- Search runs: 7 qps
- Log batch: 47 qps
- All other APIs: 127 qps
In addition, there is a limit of 20 concurrent model versions in Pending status (in creation) per workspace.
If the rate limit is reached, subsequent API calls will return status code 429. All MLflow clients (including the UI) automatically retry 429s with an exponential backoff.
API reference
The MLflow API is provided as an OpenAPI 3.0 specification that you can download and view as a structured API reference in your favorite OpenAPI editor.
- Download the OpenAPI specification
- View in Redocly: this link immediately opens the OpenAPI specification as a structured API reference for easy viewing.
- View in Postman: Postman is an app that you must download to your computer. Once you do, you can import the OpenAPI spec as a file or URL.
- View in Swagger Editor: In the online Swagger Editor, go to the File menu and click Import file to import and view the downloaded OpenAPI specification.
Important
To access Databricks REST APIs, you must authenticate.
Feedback
Submit and view feedback for