Azure Machine Learning: Workspace deletion fails: Internal Server Error (unused workspace, no compute, no locks)

Question

Service: Azure Machine Learning (workspace), Azure CLI

Scenario: I created an Azure ML workspace as part of an initial setup. The workspace was never actually used — no compute instances, clusters, or jobs were ever created. I now need to delete it because it is blocking a later step in my deployment and I want to recreate it cleanly.

The workspace originally had an auto-created Application Insights instance and an auto-created Log Analytics workspace linked to it. Both of those were deleted successfully beforehand. The remaining ML workspace itself cannot be deleted.

Result: Deletion via the Azure Portal (Workspace → Overview → Delete) fails with: "Internal Server Error"

Deletion via Azure CLI fails with the same error: az ml workspace delete --name <workspace-name> --resource-group <rg-name> --yes

Environment:

• Resource group also contains a Key Vault and a Storage Account that are shared with other resources. The Key Vault must NOT be deleted
• For this reason I have intentionally avoided the --all-resources flag
• No resource locks present on workspace or resource group

Troubleshooting efforts:

• Checked for resource locks (Resource Group → Locks) — none found
• Confirmed no compute instances/clusters exist under the workspace
• Deleted the linked Application Insights and Log Analytics workspace successfully first
• Retried deletion via Portal after browser refresh — same error
• Retried deletion via Azure CLI (az ml workspace delete without --all-resources) — same Internal Server Error

Reproducible steps:

1. Navigate to the ML workspace in the Azure Portal
2. Click Delete
3. Confirm by typing the workspace name
4. Error appears: "Internal Server Error"

Supporting materials: Happy to provide more details.

Thanks in advance.

Answer 1

Hello @Laurenz Merth

Thanks for reaching out to Microsoft Q&A. We’re looking into the issue and will get back to you as soon as possible. In the meantime, could you please share the details requested in Private Message?

Alternate Workaround:

Access your workspace by using the Azure Machine Learning Studio or Azure CLI/SDK to identify any managed online endpoint and/or batch endpoint resources associated with this workspace.

Delete the managed online endpoint(s) and/or batch endpoint(s), as nested resources of this workspace.

Then try to delete the workspace again after all nested resources are deleted.

PG Assisted :

Workspace life cycle topic running out of space, my team has freed up some stale messages, please retry the deletion 

Answer 2

Hi @Laurenz Merth

Thank you for reaching out to Microsoft Q&A. I understand you are having issues deleting a workspace in Azure Machine Learning.

With the troubleshooting you have already completed, it appears to be a backend Azure Machine Learning workspace deletion issue rather than a configuration problem.

I recommend you open an Azure support request at https://azure.microsoft.com/support/create-ticket/ and include the details above, along with a note that the workspace had no compute resources, no locks, and that the linked Application Insights and Log Analytics resources have already been removed.

To give the support team the best chance of investigating and mitigating it, you should provide the subscription ID, resource group name, workspace name, the exact time of the failed delete attempt, the full portal or CLI error output, and any correlation or request ID returned by the failed operation.

I am very sorry you are experiencing this issue, but I expect this to be resolved soon for you once a backend engineer gets involved.

If you find this to be helpful, please 'upvote' and mark the answer as 'accepted', as this will help others who experience the same issue.

Best regards,

Andrew Taylor