Formerly known as Azure AI Services or Azure Cognitive Services is a unified collection of prebuilt AI capabilities within the Microsoft Foundry platform
Hi Evgeny Pinigin,
Thanks for sharing your issue.
From your description, when a Foundry project is stuck in a non deletable state and all delete attempts return InternalServerError, it usually means the problem is happening on the backend side and not due to your configuration. [microsoftl....github.io]
What is happening
- Delete fails in portal, CLI, and API
- Project still exists even after delete attempts
- Even empty or new projects may show the same behavior
This clearly indicates a temporary service side issue or a project stuck in an inconsistent state.
Troubleshooting steps you can try
- Retry after some time
This is the most common fix
Example
- You try to delete and get InternalServerError
- Retry after some time
- In many cases, deletion works automatically once backend cleanup completes
- Try deletion using Azure CLI
Use the CLI instead of portal
az cognitiveservices account project delete \
--name <project-name> \
--account-name <foundry-account-name> \
--resource-group <resource-group-name>
Example
az cognitiveservices account project delete \
--name my-project \
--account-name my-foundry-account \
--resource-group my-rg
- Verify if project is still present
After delete, check if project still exists
az cognitiveservices account project list \
--account-name <foundry-account-name> \
--resource-group <resource-group-name>
If project is still listed, it means deletion is not completed yet
- Check for hidden or dependent resources
Even if the project looks empty, there can be linked resources
Example
- Open the project and check agents model deployments endpoints connections
If any exist, delete them first, Azure does not allow deleting a project if dependent resources are still present.
- Try deleting the resource group
If project deletion keeps failing
az group delete --name <resource-group-name> --yes --no-wait
Example
az group delete --name my-rg --yes --no-wait
Note
- This deletes all resources inside the group
- If the project is stuck, this may also fail until backend state is fixed
How to identify the root cause
- Delete fails in all methods → backend issue
- Retry works later without changes → temporary platform issue
- Only some projects fail → those projects are stuck in corrupted state
Important point
If all delete attempts return InternalServerError and nothing changes, then
- The project is likely in a partially deleted or inconsistent state
- Deletion is controlled by backend services
- This cannot be fixed from the user side
This behavior is known and usually resolves after backend stabilization.
I hope this helps. Please let me know if you have any further questions.
Thankyou!