Provisioning error with wrong quota message

Question

Provisioning error with wrong quota message

Giorgio Severi 0 Microsoft Employee

Hello,

I have an Azure ML cluster compute in East US 2 that has been standing for some months. It has 5 nodes of type: Standard_ND96isrf_H100_v5. My subscription has quota for 480 cores of Standard NDv5H100Flex Family Cluster Dedicated vCPUs.

Recently I have had some region capacity errors while trying to dynamically allocate nodes - and have been limited to 1 for the last week.

Today I am receiving the error below, which is factually incorrect.

 Provisioning error
The subscription SUB-ID has insufficient core quota for VM Family standardNDv5H100FlexFamily. Contact your subscription admin or refer to https://docs.microsoft.com/azure/machine-learning/how-to-manage-quotas#request-quota-increases to increase core quota for VM family standardNDv5H100FlexFamily in region eastus2 for your subscription

Does anyone know what is happening, and how to fix this?

SRILAKSHMI C 19,715 Reputation points Microsoft External Staff Moderator

2026-07-21T09:31:08.2433333+00:00
Hello @Giorgio Severi

Thank you for reaching out.

Based on the information you've provided, it appears that your subscription already has 480 cores of Standard NDv5H100Flex Family Cluster Dedicated vCPUs in East US 2, which should be sufficient to support your existing Azure Machine Learning compute cluster. Since the reported quota exceeds the requested capacity, the "insufficient core quota" message may not accurately reflect the underlying issue.

Azure Machine Learning provisioning errors that reference quota can sometimes be triggered by either actual quota exhaustion or regional capacity/backend provisioning constraints, which may present as quota-related errors even when sufficient quota is available.

What to check

1. Verify all applicable quota types

Azure Machine Learning uses multiple quota types, including:

VM family dedicated/low-priority vCPU quota (per VM family, per region)

Azure ML cluster quota, which is separate from the VM family quota

Managed compute resource limits that are independent of VM family quotas

Even if your VM family quota is sufficient, provisioning can still fail if another Azure ML quota has been reached.

Please verify the following in Azure Portal > Subscriptions > Usage + quotas:

The quota for Standard NDv5H100Flex Family Cluster Dedicated vCPUs in East US 2.

The Azure Machine Learning cluster quota for East US 2.

If either quota is insufficient, you can request a quota increase by following the guidance below:

https://learn.microsoft.com/azure/machine-learning/how-to-manage-quotas#request-quota-increases

2. Differentiate quota issues from regional capacity issues

Since you mentioned that you were previously experiencing regional capacity errors and have recently been limited to provisioning only a single node, this strongly suggests that the issue may be related to capacity availability for the Standard_ND96isrf_H100_v5 SKU in East US 2.

Regional capacity limitations can sometimes result in generic quota-related provisioning errors, even though the subscription quota itself is sufficient.

As a temporary mitigation, you may consider:

Scaling the cluster in smaller increments.

Using a different compatible VM SKU, if your workload permits.

Testing in another Azure region (if feasible) to determine whether the issue is isolated to regional capacity.

3. Review the detailed provisioning failure

The provisioning error shown in the Azure portal is often a generic wrapper. The underlying provisioning operation may contain more specific details.

Please review the failed operation in:

Azure Portal → Resource Group → Activity Log

Open the failed provisioning operation and review the detailed JSON response for any additional provisioning or capacity-related error codes.

If the provisioning is part of an online endpoint deployment, deployment logs may also provide more specific information regarding quota validation, configuration, or regional capacity issues.

Please refer this

Azure ML quotas overview (VM quota vs cluster quota, dedicated/low-priority, etc.): https://learn.microsoft.com/en-us/azure/machine-learning/how-to-manage-quotas?view=azureml-api-2

Troubleshoot UserErrorFromQuotaService (quota + regional capacity limitations, verification steps): https://learn.microsoft.com/azure/machine-learning/how-to-troubleshoot-online-endpoints?view=azureml-api-2

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

Thank you!

Provisioning error with wrong quota message

Your answer