Azure Cognitive Services Account Stuck in Failed State — Delete/Purge Returns InternalServerError

farhan kavi 20 Reputation points
2026-08-31T01:21:40.88+00:00

I am experiencing an issue with an Azure AI Services (Cognitive Services) that was created in Korea Central. The account provisioning failed and is currently stuck in the Failed.

I ran "adz down" successfully which removed the application resources but the failed Cognitive Services account remains and cannot be deleted. Attempts to delete or purge the account using Azure CLI consistently return:

                      InternalServerError: Encountered internal server error. Please try again
```I also attempted to recover the account, but Azure returned:

```yaml
                      CanNotRestoreAnActiveResource: Could not restore an active account
```delete from portal.azure.com - Resource group(rg-dev-trail) - delete resource group, azure returning:

```yaml
                       (Code: ResourceGroupDeletionTimeout) Encountered internal server error. Please try again. (Code: InternalServerError).
```Tracking ID:  '0ee63a97-32b9-4e56-8077-f9c9fd693781'

Resource ID:  /subscriptions/<redacted>/resourceGroups/rg-dev-trail/providers/Microsoft.CognitiveServices/accounts/ai-account-bjz4aegnph2sg

Microsoft Foundry
Microsoft Foundry

A unified Azure platform for creating and managing AI models, agents, and applications with built‑in enterprise security, monitoring, and governance


Answer accepted by question author
Ganesh Gurram 7,405 Reputation points Microsoft External Staff Moderator
2026-09-01T07:00:04.15+00:00

Hi @farhan kavi

We have checked with the PG team, and below is the response we received from them:

"The account cannot be deleted because its backing AML hub has a child AI project stuck in Active state. The CogSvc RP's delete path calls AML RP to soft-delete the backing hub; AML RP rejects that call with a 400 ValidationError, and the CogSvc RP client converts any non-success downstream status into a hardcoded InternalServerError exception that goes unhandled — which is the 500 the customer sees. The 500 is a masked 400; it is not a transient platform fault, so retrying can never succeed.

We have manually cleaned Hub, please retry the operation."

I Hope this helps. Do let me know if you have any further queries.


If this answers your query, please do click "Accept Answer" and "Yes" for was this answer helpful.

Thank you!

Was this answer helpful?

2 people found this answer helpful.
0 comments No comments

0 additional answers

Sort by: Most helpful

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.