RTX instance reboots multiple times and eventually fails

Question

Attempting to create a 24core RTX GPU VM on US West 1. All peripherals provision successfully and immediately, but the VM itself loops until it eventually reaches a failed state. The error messages are completely unhelpful so I have no idea what the problem is:

"At least one resource deployment operation failed. Please list deployment operations for details. Please see https://aka.ms/arm-deployment-operations for usage details.",

"The resource write operation failed to complete successfully, because it reached terminal provisioning state 'Failed'."

I've attempted to create this VM multiple times from both the portal and the CLI, all with the same result. The quota team confirmed it was not a quota issue.

Answer 1

Hello @Kristina Kruglyak,

Thank you for using Q&A forum.

1. Specific VM size unavailable in that zone/fault domain 24-core RTX GPUs (likely Standard_NV24ads_A10_v5 or similar NVv4/NCas series) may simply not be available in the specific fault domain or availability zone you're targeting.
2. Image compatibility issue Certain marketplace images, especially GPU-optimized ones, have driver or generation requirements that silently fail during provisioning.

Step 1 — Get the Real Error

The ARM message is a wrapper. Get the actual inner error:

powershell

Replace with your deployment name and resource group

az deployment operation group list \

--resource-group <your-rg> \

--name <deployment-name> \

--query "[?properties.provisioningState=='Failed'].{op:properties.targetResource.resourceType, error:properties.statusMessage.error}" \

--output table

Step 2 — Check Actual SKU Availability

powershell

Check if your target size is actually available in West US

az vm list-skus \

--location westus \

--size Standard_NV \

--query "[].{Name:name, Restrictions:restrictions[*].reasonCode}" \

--output table

Step 3 — Try These Workarounds in Order

Switch to West US 2 or West US 3 Both are newer regions with more GPU capacity. If this is not a hard requirement, this is the fastest fix.

powershell

az vm create \

--resource-group <your-rg> \

--name <vm-name> \

--location westus2 \ # try westus2 or westus3

--size <your-gpu-sku> \

--image <your-image>

Remove availability zone pinning If your template specifies zones: ["1"] or similar, remove it and let Azure pick. Pinning to a zone that has no GPU capacity is a common silent failure.

Try a different but equivalent SKU If you're on Standard_NV24ads_A10_v5, try Standard_NC24ads_A100_v4 or the nearest equivalent.

---

If this answers your query, do click Accept Answer and Up-Vote for the same. And, if you have any further query do let us know.

Answer 2

Hello Kristina Kruglyak,

Welcome to the Microsoft Q&A and thank you for posting your questions here.

I understand that your RTX instance reboots multiple times and eventually fails.

This is Azure compute allocation or GPU SKU capacity unavailability for the selected RTX VM size in the selected West US region or zone, not by quota. Because error shown is a generic Azure Resource Manager deployment failure, so the actual root cause must be confirmed from the failed deployment operation details. - https://docs.azure.cn/en-us/azure-resource-manager/templates/deployment-history, https://github.com/MicrosoftDocs/azure-docs/blob/main/articles/azure-resource-manager/troubleshooting/find-error-code.md

What you need to do is to:

• Retrieve the failed deployment operation details and confirm the exact Microsoft.Compute/virtualMachines error.
• If the error is AllocationFailed, SkuNotAvailable, ZonalAllocationFailed, or capacity-related, stop retrying the same RTX VM in the same region/zone.
• Check whether the requested GPU VM SKU is restricted or unavailable in the selected location.
• Redeploy the workload using a supported, unrestricted GPU VM SKU in another available zone or region.
• If the same GPU SKU and same region/zone are mandatory, create an On-demand Capacity Reservation for that exact VM size. If the reservation fails, Azure does not currently have allocatable capacity for that exact SKU/location, and Azure Support can then be engaged for backend capacity validation.

After moving the RTX workload to an available GPU SKU, region, or zone, or after successfully reserving capacity for the exact required SKU, the VM deployment should complete successfully. If capacity reservation fails for the same VM size and location, the issue is confirmed as a backend capacity/SKU availability limitation rather than a quota or portal/CLI problem. - https://learn.microsoft.com/en-us/azure/azure-resource-manager/troubleshooting/common-deployment-errors, https://learn.microsoft.com/en-us/troubleshoot/azure/virtual-machines/windows/troubleshoot-deployment-new-vm-windows, and https://learn.microsoft.com/en-us/azure/virtual-machines/capacity-reservation-create Use the associated resource links for more reading and steps.

I hope this is helpful. Please! Do not hesitate to let me know if you have any other questions, steps or clarifications.

---

Please do not close the thread by upvoting and accepting the answer if any part of it is helpful.