VMSS scale-up fails with OverconstrainedAllocationRequest although pinned cluster has available Standard_ND96amsr_A100_v4 capacity

Question

Hi,

I'm trying to scale out a Uniform VMSS in West US 2:

• VMSS: A100-ND80G-Linux-WestUS2-ForAdsOnly
• SKU: Standard_ND96amsr_A100_v4
• Region: West US 2
• singlePlacementGroup = false
• platformFaultDomainCount = 1
• securityType = Standard
• No PPG configured (proximityPlacementGroup = null)
• No Availability Zone configured (zones = null)

The VMSS has been recreated from scratch and currently has 0 instances.

The VMSS is expected to be pinned to cluster MWH20PrdGPC01. Internal capacity data shows:

• Tenant: MWH20PrdGPC01
• VM Type: GPCv4A100S80_ND96amsr
• Available VM count: 60

However, scaling the VMSS from 0 to 1 instance fails immediately with:Hi,

I'm trying to scale out a Uniform VMSS in West US 2:

• VMSS: A100-ND80G-Linux-WestUS2-ForAdsOnly
• SKU: Standard_ND96amsr_A100_v4
• Region: West US 2
• singlePlacementGroup = false
• platformFaultDomainCount = 1
• securityType = Standard
• No PPG configured (proximityPlacementGroup = null)
• No Availability Zone configured (zones = null)

The VMSS has been recreated from scratch and currently has 0 instances.

The VMSS is expected to be pinned to cluster MWH20PrdGPC01. Internal capacity data shows:

• Tenant: MWH20PrdGPC01
• VM Type: GPCv4A100S80_ND96amsr
• Available VM count: 60

However, scaling the VMSS from 0 to 1 instance fails immediately with:

OverconstrainedAllocationRequest

Allocation failed. VM(s) with the following constraints cannot be allocated, because the condition is too restrictive.

Constraints applied are:

• VM Size

A standalone VM deployment using the same SKU (Standard_ND96amsr_A100_v4) in West US 2 also fails, but with a different error:

AllocationFailed

We do not have sufficient capacity for the requested VM size in this region.

What does "Constraints applied are: - VM Size" specifically mean in this scenario? Does it indicate that the requested SKU (Standard_ND96amsr_A100_v4) is not considered allocatable on the target cluster, or are there additional placement/allocation filters causing all candidate hosts for this VM size to be excluded?

My questions are:

1. My questions are:
   1. In the context of scaling this VMSS, what exactly does "Constraints applied are: - VM Size" mean? Does it indicate that the requested SKU (Standard_ND96amsr_A100_v4) is not considered allocatable for this VMSS placement request, even though the SKU is available in the region and the pinned cluster reports available capacity?
   2. If the pinned cluster still shows 60 available GPCv4A100S80_ND96amsr slots, why can the VMSS not scale to a single instance?
   3. Could this indicate a placement/filtering issue rather than a pure capacity shortage?
   4. What backend checks or diagnostics would you recommend to determine why the allocation is being rejected?

Thanks in advance for any guidance.

Answer 1

Hey Yanpeng — that error just means no host was found, not that the SKU is wrong. Your standalone test failing too suggests West US 2 is out of ND96amsr capacity. Grab the correlation ID from Activity Log and raise a support case. Capacity Reservations help longer term.