Managed online endpoint deployment fails with InternalServerError — no container logs, region germanywestcentral

Question

Managed online endpoint deployment fails with InternalServerError — no container logs, region germanywestcentral

Marco Turzi 0

Managed online deployment creation consistently fails with a generic InternalServerError (Code: InternalServerError) with no further detail exposed to the CLI. A prior attempt with the same configuration failed with a different error: "Specified deployment [gdsii-deployment-master] failed during initial provisioning and is in an unrecoverable state. Delete and re-create." That deployment was deleted. Confirmed the workspace's managed identity has the AcrPull role assignment on the target container registry. Subsequent az ml online-deployment create attempts with a clean deployment still return InternalServerError, and az ml online-deployment get-logs returns "There are no logs for this deployment at the moment" — indicating the container never starts. Requesting investigation into the backend provisioning failure using the correlation/trace IDs below.

0 comments

1 answer

Your answer

Answer 1

Hello Marco Turzi

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that your Azure Machine Learning Managed Online Endpoint deployment fails with InternalServerError: No container logs, region germanywestcentral.

The deployment was already deleted and recreated, and AcrPull permissions were confirmed. Since the failure persists on a fresh deployment and continues to return only InternalServerError, the most likely root cause is an Azure Machine Learning provisioning or regional backend issue rather than a customer configuration problem. The InternalServerError for Online Endpoints as a service-side failure requiring backend investigation. - https://learn.microsoft.com/en-us/azure/machine-learning/how-to-troubleshoot-online-endpoints?view=azureml-api-2

What you can do is to:

Verify that the endpoint identity (not only the workspace identity) has the required Azure Container Registry and Storage permissions.
Confirm sufficient quota and VM availability for the selected SKU in germanywestcentral.
Capture deployment telemetry using Azure CLI debug mode.
Collect correlation IDs, request IDs, deployment timestamps, endpoint name, deployment name, workspace name, and region.
Raise an Azure Support ticket for Azure Machine Learning backend investigation using the captured identifiers via your Portal or contact PCS - Priority Customer Service. - https://learn.microsoft.com/en-us/azure/azure-portal/supportability/priority-community-support

At this stage, Azure backend telemetry is the only source capable of identifying the exact provisioning failure because no deployment containers were successfully created and therefore no customer-visible logs exist. - https://learn.microsoft.com/en-us/azure/machine-learning/how-to-troubleshoot-online-endpoints?view=azureml-api-2, and https://learn.microsoft.com/en-us/azure/machine-learning/how-to-manage-quotas?view=azureml-api-2 gives more insights.

I hope this is helpful. Please! Do not hesitate to let me know if you have any other questions, steps or clarifications.

Please do not close the thread by upvoting and accepting the answer if any part of it is helpful.