Foundry provisioning failed

Question

Foundry provisioning failed

Jeff Brown 0

Microsoft Foundry AIServices resource jeff-5442-resource failed during provisioning in westus3. Resource group: forgewarden-dev. Subscription: Azure subscription 1. Activity Log reports InternalServerError at 2026-08-12T05:16:47.7608772Z. Resource Health reports Unknown with transient chronicity. No model deployments exist. Please advise whether this resource can be repaired or must be recreated.

0 comments

3 answers

Your answer

Answer 1

A failed Foundry AIServices resource is not documented as repairable in place from the provided guidance. The supported validation step is to confirm whether the resource actually exists and whether provisioning completed successfully.

Use Azure PowerShell:

Sign in:
```
   Connect-AzAccount
```

Select the subscription that contains the resource:

   Set-AzContext -Subscription "Azure subscription 1"

Check the resource:

   Get-AzCognitiveServicesAccount -Name jeff-5442-resource -ResourceGroupName forgewarden-dev | Select-Object AccountName, Kind, @{Name="ProvisioningState"; Expression={$_.Properties.ProvisioningState}}, Endpoint

Interpret the result as follows:

If Kind is AIServices and ProvisioningState is Succeeded, the resource is usable.
If the resource is missing, or ProvisioningState is not Succeeded, the provided documentation does not describe an in-place repair path.

Based on the available troubleshooting guidance for Foundry provisioning failures, the practical recovery path is to recreate the resource. If recreation is chosen, use a new globally unique account name if the original name is still reserved or if creation reports a name conflict. For cleanup, deleting the resource group is destructive and removes all resources in that group.

If deployment tooling is involved, also check these known provisioning blockers before recreating:

AccountNameInvalid: the Foundry resource name must be globally unique, 2–64 characters, and use only lowercase letters, numbers, and hyphens.
Region mismatch: explicitly pass the intended location if deployment defaults to another region.
CLI bug in Azure CLI 2.74–2.75: rerun with --debug and inspect Exception Details: if the visible error is masking the real cause.

If the failed resource must be removed first, the documented cleanup commands are:

Remove-AzResourceGroup -Name exampleRG

or

az group delete --name exampleRG

Warning: both commands are destructive for the entire resource group.

Answer 2

kagiyama yutaka 5,330

I think a Foundry AIServices stuck in Failed won’t move to a working state from the user side, and the safe path is to delete and recreate it. The Deployments blade only shows where it stopped and doesn’t offer recovery.

0 comments

Answer 3

Hi @Jeff Brown,

Thank you for providing the resource details.

Based on the information shared, the AIServices resource provisioning failed with an InternalServerError during resource creation, and the resource health status is showing Unknown. Since no model deployments exist and the failure occurred during the initial provisioning phase, this may indicate a backend provisioning issue rather than a configuration problem in the resource itself.

Before recreating the resource, could you please share the following details:

The full Activity Log error details (JSON view if available).
The Correlation ID / Operation ID associated with the failed deployment.
Whether the resource currently remains in a Failed provisioning state in the Azure portal.

As an initial troubleshooting step, please try selecting Redeploy or Retry if the option is available in the portal. If the resource remains in a failed state and cannot be recovered, creating a new AIServices resource may be required. However, we recommend reviewing the deployment diagnostics first to determine whether the failure was caused by a transient platform issue.

Once you provide the Activity Log details and Correlation ID, we can help further investigate whether the resource can be recovered or if recreation is the recommended path.