Question

You deployed a model to a managed online endpoint in Azure Machine Learning. The deployment succeeded, but every scoring request returns an HTTP 424 error. You need to find the root cause quickly by inspecting the container's stdout/stderr output where the scoring script's init() and run() functions write their messages. Which action should you take first?

Accepted Answer

Retrieve the deployment's console logs using 'az ml online-deployment get-logs' for the specific endpoint and deployment.

Answer

Increase the 'request_timeout_ms' value in the deployment's request settings and redeploy.

Answer

Shift all traffic to a new blue-green deployment to isolate the failing container.

Answer

Enable Application Insights on the workspace and wait for aggregated metrics to appear in the dashboard.