I am facing the issue while creating a NAP node pool through Azure portal

Question

I am facing the issue while creating a NAP node pool through Azure portal

cargobuddy 20

Every time I try to create a new NAP pool, I am getting the error that the name is already taken, please choose a different name. No matter what name I use. I have already enabled Karpenter in my AKS cluster using the portal. But what next to do after that to create a scaled node pool?

Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-17T11:49:48.37+00:00
Hello @cargobuddy

Private DNS Zone is not linked to your VM's subnet (most likely root cause).

When AKS creates a private cluster, it automatically creates a Private DNS Zone (something like privatelink.<region>.azmk8s.io) and links it to the AKS VNet. The problem is that this DNS zone is only linked to the AKS VNet by default — not to your VM's subnet, even if both are in the same VNet.

Without this DNS link, your VM can't resolve the private FQDN of the API server, so kubectl commands silently fail or time out, and the portal can't reach the cluster to display nodes/pods either.

To verify and fix this, run:

Check if your VNet has the DNS zone linked

az network private-dns link vnet list \ --resource-group <AKS_NODE_RESOURCE_GROUP> \ --zone-name <your-private-dns-zone>

If your VM's VNet isn't listed, add the link:

az network private-dns link vnet create \ --resource-group <AKS_NODE_RESOURCE_GROUP> \ --zone-name <your-private-dns-zone> \ --name MyVNetLink \ --virtual-network <YOUR_VNET_ID> \ --registration-enabled false

Create a private AKS cluster – DNS configuration options
https://learn.microsoft.com/en-us/azure/aks/private-cluster#options-for-connecting-to-the-private-cluster

2.Azure RBAC role on the AKS resource itself

Yes, to answer your question directly — the identity you're using on the bastion VM does need a specific Azure role on the AKS cluster resource (not just inside Kubernetes). Without Azure Kubernetes Service Cluster User Role (or Cluster Admin Role), you can't even run az aks get-credentials to pull down the kubeconfig.

Assign it like this:

AKS_ID=$(az aks show \ --resource-group <RESOURCE_GROUP> \ --name <CLUSTER_NAME> \ --query id -o tsv) az role assignment create \ --assignee <USER_OR_MANAGED_IDENTITY_PRINCIPAL_ID> \ --role "Azure Kubernetes Service Cluster User Role" \ --scope $AKS_ID

Then pull the credentials *from the bastion VM itself:

az login # or use managed identity: az login --identity az aks get-credentials --resource-group <RG> --name <CLUSTER_NAME> kubectl get pods --all-namespaces

Use Azure RBAC for Kubernetes authorization
https://learn.microsoft.com/en-us/azure/aks/manage-azure-rbac

3.NSG rules between subnets

Since your VM and AKS nodes are in different subnets (even within the same VNet), NSG rules can block the required traffic. Make sure outbound TCP port 443 is allowed from your VM's subnet toward the AKS subnet (where the private endpoint lives). You can verify the private endpoint IP with:

Azure PowerShell

az network private-endpoint show \ --name <PRIVATE_ENDPOINT_NAME> \ --resource-group <AKS_NODE_RESOURCE_GROUP> \ --query 'customDnsConfigs[].ipAddresses'

About the Azure Portal behavior

The fact that the portal shows the cluster but not individual nodes or pods is actually expected for private clusters the portal's backend can't reach your private API server endpoint. To manage it through the portal, you can use the built-in Run Command feature (under the cluster's overview → "Run command"), which proxies commands through Azure without needing direct network access.

Use the Run command for private AKS clusters:
https://learn.microsoft.com/en-us/azure/aks/access-private-cluster

Additional references:

https://learn.microsoft.com/en-us/azure/aks/private-cluster https://learn.microsoft.com/en-us/azure/aks/concepts-identity

https://learn.microsoft.com/en-us/azure/aks/private-cluster#configure-private-dns-zone

Thanks,
Manish.
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-18T10:54:09.2566667+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail
cargobuddy 20 Reputation points

2026-06-18T17:52:58.7666667+00:00

Hi @Manish Deshpande

I think you have answered for the wrong question. The question was, that I am not able to create a NAP node pool, which I want to create for node level auto-scaling. Whereas, you answered me for a question which I asked in a different thread.
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-24T10:32:08.29+00:00
Hello @cargobuddy

The "name is already taken" error is actually expected behavior, and here's why:

When you enable Node Auto Provisioning (NAP) via the portal, AKS automatically provisions a Karpenter NodePool and an AKSNodeClass, both named default, in the background. The portal's NAP creation UI maps directly to these Kubernetes CRDs — so any name you try is colliding with resources Karpenter already created silently.

Step 1 — Confirm what's already there

Connect to your cluster and run:

kubectl get nodepools kubectl get aksnodeclass

You'll almost certainly see a default NodePool already in place. This is by design.

Step 2 — Stop using the portal for additional NAP pools

This is the key insight: with NAP enabled, additional node pools are managed as Karpenter NodePool CRDs via kubectl — not through the portal's "Add node pool" UI (which is for traditional agent pools). Here's a minimal example:

apiVersion: karpenter.sh/v1 kind: NodePool metadata: name: my-scaled-pool spec: template: spec: nodeClassRef: group: karpenter.azure.com kind: AKSNodeClass name: default requirements: - key: "karpenter.azure.com/sku-family" operator: In values: ["D"] limits: cpu: 100 disruption: consolidationPolicy: WhenEmptyOrUnderutilized consolidationAfterEmpty: 30s

Apply it with:

kubectl apply -f my-scaled-pool.yaml

Karpenter will then automatically scale nodes up/down based on pending pod demand — no manual scaling needed.

Step 3 — Verify Karpenter is running

Check that the Karpenter controller pods are healthy:

kubectl get pods -n kube-system | grep karpenter

The portal experience for NAP is still maturing. Right now, it's most useful for enabling/disabling NAP and monitoring cluster state — the actual NodePool lifecycle management is best done via kubectl or ARM/Bicep templates.

Node Auto Provisioning overview
https://learn.microsoft.com/en-us/azure/aks/node-autoprovision

Manage NAP NodePools:
https://learn.microsoft.com/en-us/azure/aks/manage-node-auto-provisioning

Karpenter NodePool concepts:
https://karpenter.sh/docs/concepts/nodepools/

AKSNodeClass reference:
https://learn.microsoft.com/en-us/azure/aks/node-autoprovision#configure-node-provisioning-settings

Thanks,
Manish.
cargobuddy 20 Reputation points

2026-06-24T15:07:27.44+00:00

Hi @Manish Deshpande

Thanks for your reply. In the above YAML, can I please know whether I need to specify the values from my existing cluster? If so, which all parameters should I replace with my values?

Also, as you mentioned, the NAP is still evolving for AKS offer, so can you please let me know if I want to use VMSS for auto-scaling for my existing cluster, what are the steps to configure it?
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-25T17:53:08.4066667+00:00
Hello @cargobuddy

Pls go through the suggestions.

Customizing the NodePool YAML with your existing cluster values

When you create a new NodePool under NAP, you don't need to reinvent the wheel — you can mirror most of what your existing cluster is already using. Here's what to update:

-nodeClassRef → name: Keep this as default to reuse the AKSNodeClass that NAP already created for you. If you want different OS settings, subnet, or disk configuration, create a new AKSNodeClass first and point to that instead. You can inspect what's currently configured with:

kubectl get aksnodeclass default -o yaml

karpenter.azure.com/sku-family**: Replace with the VM family your cluster currently uses (e.g., Dfor general purpose,Efor memory-optimized,Ffor compute-optimized). You can also pin specific SKUs usingkarpenter.azure.com/sku-nameif you want exact instance types likeStandard_D4s_v5`.

topology.kubernetes.io/zone**: Add this requirement if your workloads are zone-aware. Use az account list-locations --output table to confirm available zones for your region.

limits.cpu/limits.memory`**: Set these to cap the total resources this NodePool can provision. Think of it as a safety ceiling.

consolidationPolicy**: WhenEmptyOrUnderutilized` is the recommended setting for cost efficiency — it scales down nodes when they're idle or lightly loaded.

A quick example tailored to reuse your default AKSNodeClass:

apiVersion: karpenter.sh/v1 kind: NodePool metadata: name: my-custom-pool spec: disruption: consolidationPolicy: WhenEmptyOrUnderutilized limits: cpu: "100" memory: 200Gi template: spec: nodeClassRef: name: default # reuses your existing AKSNodeClass requirements: - key: karpenter.azure.com/sku-family operator: In values: ["D"] # replace with your preferred VM family - key: karpenter.sh/capacity-type operator: In values: ["on-demand"] - key: kubernetes.io/arch operator: In values: ["amd64"]

Configure Node Pools for NAP [https://learn.microsoft.com/en-us/azure/aks/node-auto-provisioning-node-pools](https://learn.microsoft.com/en-us/azure/aks/node-auto-provisioning-node-pools) Configure AKSNodeClass [https://learn.microsoft.com/en-us/azure/aks/node-auto-provisioning-aksnodeclass](https://learn.microsoft.com/en-us/azure/aks/node-auto-provisioning-aksnodeclass) 2. Using VMAS for auto-scaling — important heads-up Unfortunately, VMAS (Virtual Machine Availability Sets) is **no longer a viable option** for AKS autoscaling. Microsoft officially retired VMAS support in AKS on **September 30, 2025**. As of that date, clusters still using Availability Sets are considered out of support, and you can no longer create new clusters with VMAS either. If your existing cluster is still on VMAS, the recommended path is to migrate to **Virtual Machine node pools**, which is a one-command operation: ```bash az aks update \ --name <your-cluster> \ --resource-group <your-rg> \ --migrate-vmas-to-vms

This also automatically upgrades a Basic Load Balancer to Standard if needed.

Once migrated to VM node pools, you have two solid auto-scaling options:

NAP (Node Auto Provisioning) — what you're already exploring. It's now generally available and handles right-sized node selection automatically based on pending pod pressure. Best for dynamic, mixed workloads.

Cluster Autoscaler with VM/VMSS node pools — the traditional approach, great if you want explicit control over node pool sizes:

az aks nodepool update \ --resource-group <your-rg> \ --cluster-name <your-cluster> \ --name <nodepool-name> \ --enable-cluster-autoscaler \ --min-count 1 \ --max-count 10

VMAS Deprecation on AKS
https://learn.microsoft.com/en-us/azure/aks/availability-sets-on-aks
NAP Overview
https://learn.microsoft.com/en-us/azure/aks/node-auto-provisioning
Cluster Autoscaler on AKS
https://learn.microsoft.com/en-us/azure/aks/cluster-autoscaler

Thanks,
Manish.
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-26T04:06:05.8266667+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail.
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-27T18:08:25.59+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail.

1 answer

Your answer

Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-18T10:54:09.2566667+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail
cargobuddy 20 Reputation points

2026-06-18T17:52:58.7666667+00:00

Hi @Manish Deshpande

I think you have answered for the wrong question. The question was, that I am not able to create a NAP node pool, which I want to create for node level auto-scaling. Whereas, you answered me for a question which I asked in a different thread.
cargobuddy 20 Reputation points

2026-06-24T15:07:27.44+00:00

Hi @Manish Deshpande

Thanks for your reply. In the above YAML, can I please know whether I need to specify the values from my existing cluster? If so, which all parameters should I replace with my values?

Also, as you mentioned, the NAP is still evolving for AKS offer, so can you please let me know if I want to use VMSS for auto-scaling for my existing cluster, what are the steps to configure it?
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-26T04:06:05.8266667+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail.
Manish Deshpande 7,520 Reputation points Microsoft External Staff Moderator

2026-06-27T18:08:25.59+00:00

Hello @cargobuddy

I wanted to check if my last response made sense. I’d be glad to assist further or explain anything in more detail.

Answer 1

SUNOJ KUMAR YELURU 18,336 MVP Volunteer Moderator

Hello @cargobuddy,

Thank you for reaching out Q&A forum.

When creating a Node Auto-Provisioning (NAP) node pool in Azure Kubernetes Service (AKS) and encountering an error stating that the name is already taken, it is important to ensure that the name you are trying to use is unique within the scope of your AKS cluster. If you have already enabled Karpenter in your AKS cluster, the next steps to create a scaled node pool include:

Verify Node Pool Configuration: Ensure that your NodePool configuration is correct and that it does not conflict with existing node pools. You can check the existing node pools using the Azure CLI or the Azure portal.
Use Unique Names: Make sure that the name you are trying to use for the new NAP node pool is not already in use by another node pool. Each NodePool must have a unique name within the AKS cluster. If you are unsure, try using a completely different name.
Configure AKSNodeClass: If you want to customize the node pool further, you can configure AKSNodeClass resources to define Azure-specific settings for NAP. This includes settings like VM image, OS disk size, and maximum pods per node. Each NodePool must reference an AKSNodeClass using spec.template.spec.nodeClassRef.

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.

cargobuddy 20 Reputation points

2026-06-18T18:17:42.4866667+00:00

But I tried all sort of names. Even any random name, if I provide, it returns with the error that the name is already taken. This is quite confusing to me.
cargobuddy 20 Reputation points

2026-06-18T19:01:43.5833333+00:00

@SUNOJ KUMAR YELURU for any given name, it results into the same error. No matter what name I specify. I gave a random name like onvpowqhef22safwqf, but it still results in the same error. So I think the underlying root cause is different. Further, in my AKS cluster, I already have nodes with manual scaling in place. I want to switch to the auto-scaled nodes with NAP. For that only, I raised this question.