DatabricksInstall

Overview

The node interacts with the Databricks API, specifically supporting operations related to "Model Serving" among other resources. For the Update Serving Endpoint operation under the Model Serving resource, it updates an existing model serving endpoint by configuring which models are served and how traffic is routed among them.

This node is beneficial in scenarios where you need to programmatically manage machine learning model deployment endpoints on Databricks, such as updating the set of models behind an endpoint or adjusting traffic distribution for A/B testing or gradual rollouts.

Practical example:
You have a serving endpoint named "recommendation-service" currently serving version 1 of your recommendation model. You want to update it to serve both version 1 and version 2, directing 70% of traffic to version 2 and 30% to version 1. Using this node's Update Serving Endpoint operation, you provide the updated list of served models and traffic configuration in JSON format to achieve this.

Properties

Output

The node outputs a JSON object representing the response from the Databricks API after attempting to update the serving endpoint. This typically includes details about the updated endpoint configuration, confirmation of success, or error information if the update failed.

No binary data output is indicated for this operation.

Dependencies

• Requires an API authentication token credential for Databricks with appropriate permissions to manage model serving endpoints.
• The node uses the Databricks REST API base URL and authorization header derived from the provided credentials.
• No additional external dependencies beyond the Databricks API.

Troubleshooting

• Common issues:

  • Invalid JSON format in the Served Models or Traffic Config properties can cause request failures.
  • Incorrect or missing endpoint name will result in errors as the API cannot identify which endpoint to update.
  • Insufficient permissions or invalid API token will lead to authorization errors.
  • Network connectivity problems may prevent reaching the Databricks API.

• Error messages:

  • API Error: <status code> <status text>: Indicates the API responded with an error status. Check the detailed message for specifics (e.g., 404 Not Found if the endpoint does not exist).
  • Network Error: No response received from server: Suggests network issues or incorrect API host configuration.
  • JSON parsing errors if the input JSON is malformed.

• Resolution tips:

  • Validate JSON inputs before running the node.
  • Verify the endpoint name exists and is correctly spelled.
  • Ensure the API token has necessary permissions.
  • Confirm network access to the Databricks API host.

Links and References

• Databricks Model Serving API Documentation
• Databricks REST API Reference