Create and configure an Azure Kubernetes Services (AKS) cluster to use virtual nodes using Azure CLI

Article
10/27/2023

Virtual nodes enable network communication between pods that run in Azure Container Instances (ACI) and AKS clusters. To provide this communication, you create a virtual network subnet and assign delegated permissions. Virtual nodes only work with AKS clusters created using advanced networking (Azure CNI). By default, AKS clusters are created with basic networking (kubenet). This article shows you how to create a virtual network and subnets, then deploy an AKS cluster that uses advanced networking.

This article shows you how to use the Azure CLI to create and configure virtual network resources and an AKS cluster enabled with virtual nodes.

Before you begin

Important

Before using virtual nodes with AKS, review both the limitations of AKS virtual nodes and the virtual networking limitations of ACI. These limitations affect the location, networking configuration, and other configuration details of both your AKS cluster and the virtual nodes.

You need the ACI service provider registered with your subscription. You can check the status of the ACI provider registration using the az provider list command.

az provider list --query "[?contains(namespace,'Microsoft.ContainerInstance')]" -o table

The Microsoft.ContainerInstance provider should report as Registered, as shown in the following example output:

Namespace                    RegistrationState    RegistrationPolicy
---------------------------  -------------------  --------------------
Microsoft.ContainerInstance  Registered           RegistrationRequired

If the provider shows as NotRegistered, register the provider using the az provider register.

az provider register --namespace Microsoft.ContainerInstance

If using Azure CLI, this article requires Azure CLI version 2.0.49 or later. Run az --version to find the version. If you need to install or upgrade, see Install Azure CLI.

Launch Azure local Shell

If you prefer to install and use the CLI locally, this article requires Azure CLI version 2.0.49 or later. Run az --version to find the version. If you need to install or upgrade, see Install Azure CLI.

Create a resource group

An Azure resource group is a logical group in which Azure resources are deployed and managed.

Create a resource group using the az group create command.

az group create --name myResourceGroup --location eastus

Create a virtual network

Important

Virtual node requires a custom virtual network and associated subnet. It can't be associated with the same virtual network as the AKS cluster.

Create a virtual network using the az network vnet create command. The following example creates a virtual network named myVnet with an address prefix of 10.0.0.0/8 and a subnet named myAKSSubnet. The address prefix of this subnet defaults to 10.240.0.0/16.
```
az network vnet create \
    --resource-group myResourceGroup \
    --name myVnet \
    --address-prefixes 10.0.0.0/8 \
    --subnet-name myAKSSubnet \
    --subnet-prefix 10.240.0.0/16
```
Create an extra subnet for the virtual nodes using the az network vnet subnet create command. The following example creates a subnet named myVirtualNodeSubnet with an address prefix of 10.241.0.0/16.
```
az network vnet subnet create \
    --resource-group myResourceGroup \
    --vnet-name myVnet \
    --name myVirtualNodeSubnet \
    --address-prefixes 10.241.0.0/16
```

Create an AKS cluster with managed identity

Get the subnet ID using the az network vnet subnet show command.

az network vnet subnet show --resource-group myResourceGroup --vnet-name myVnet --name myAKSSubnet --query id -o tsv

Create an AKS cluster using the az aks create command and replace <subnetId> with the ID obtained in the previous step. The following example creates a cluster named myAKSCluster with five nodes.
```
az aks create \
    --resource-group myResourceGroup \
    --name myAKSCluster \
    --node-count 5 \
    --network-plugin azure \
    --vnet-subnet-id <subnetId>
```
After several minutes, the command completes and returns JSON-formatted information about the cluster.

For more information on managed identities, see Use managed identities.

Enable the virtual nodes addon

Enable virtual nodes using the az aks enable-addons command. The following example uses the subnet named myVirtualNodeSubnet created in a previous step.

az aks enable-addons \
    --resource-group myResourceGroup \
    --name myAKSCluster \
    --addons virtual-node \
    --subnet-name myVirtualNodeSubnet

Connect to the cluster

Configure kubectl to connect to your Kubernetes cluster using the az aks get-credentials command. This step downloads credentials and configures the Kubernetes CLI to use them.
```
az aks get-credentials --resource-group myResourceGroup --name myAKSCluster
```
Verify the connection to your cluster using the kubectl get command, which returns a list of the cluster nodes.
```
kubectl get nodes
```
The following example output shows the single VM node created and the virtual node for Linux, virtual-node-aci-linux:
```
NAME                          STATUS    ROLES     AGE       VERSION
virtual-node-aci-linux        Ready     agent     28m       v1.11.2
aks-agentpool-14693408-0      Ready     agent     32m       v1.11.2
```

Deploy a sample app

Create a file named virtual-node.yaml and copy in the following YAML. The YAML schedules the container on the node by defining a nodeSelector and toleration.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: aci-helloworld
spec:
  replicas: 1
  selector:
    matchLabels:
      app: aci-helloworld
  template:
    metadata:
      labels:
        app: aci-helloworld
    spec:
      containers:
      - name: aci-helloworld
        image: mcr.azk8s.cn/azuredocs/aci-helloworld
        ports:
        - containerPort: 80
      nodeSelector:
        kubernetes.io/role: agent
        beta.kubernetes.io/os: linux
        type: virtual-kubelet
      tolerations:
      - key: virtual-kubelet.io/provider
        operator: Exists
      - key: azure.com/aci
        effect: NoSchedule

Run the application using the kubectl apply command.
```
kubectl apply -f virtual-node.yaml
```
Get a list of pods and the scheduled node using the kubectl get pods command with the -o wide argument.
```
kubectl get pods -o wide
```
The pod is scheduled on the virtual node virtual-node-aci-linux, as shown in the following example output:
```
NAME                            READY     STATUS    RESTARTS   AGE       IP           NODE
aci-helloworld-9b55975f-bnmfl   1/1       Running   0          4m        10.241.0.4   virtual-node-aci-linux
```
The pod is assigned an internal IP address from the Azure virtual network subnet delegated for use with virtual nodes.

Note

If you use images stored in Azure Container Registry, configure and use a Kubernetes secret. A current limitation of virtual nodes is you can't use integrated Microsoft Entra service principal authentication. If you don't use a secret, pods scheduled on virtual nodes fail to start and report the error HTTP response status code 400 error code "InaccessibleImage".

Test the virtual node pod

Test the pod running on the virtual node by browsing to the demo application with a web client. As the pod is assigned an internal IP address, you can quickly test this connectivity from another pod on the AKS cluster.
Create a test pod and attach a terminal session to it using the following kubectl run -it command.
```
kubectl run -it --rm testvk --image=mcr.azk8s.cn/dotnet/runtime-deps:6.0
```

Install curl in the pod using apt-get.

apt-get update && apt-get install -y curl

Access the address of your pod using curl, such as http://10.241.0.4. Provide your own internal IP address shown in the previous kubectl get pods command.
```
curl -L http://10.241.0.4
```
The demo application is displayed, as shown in the following condensed example output:
```
<html>
<head>
  <title>Welcome to Azure Container Instances!</title>
</head>
[...]
```
Close the terminal session to your test pod with exit. When your session is ends, the pod is deleted.

Remove virtual nodes

Delete the aci-helloworld pod running on the virtual node using the kubectl delete command.
```
kubectl delete -f virtual-node.yaml
```

Disable the virtual nodes using the az aks disable-addons command.

az aks disable-addons --resource-group myResourceGroup --name myAKSCluster --addons virtual-node

Remove the virtual network resources and resource group using the following commands.

# Change the name of your resource group, cluster and network resources as needed
RES_GROUP=myResourceGroup
AKS_CLUSTER=myAKScluster
AKS_VNET=myVnet
AKS_SUBNET=myVirtualNodeSubnet

# Get AKS node resource group
NODE_RES_GROUP=$(az aks show --resource-group $RES_GROUP --name $AKS_CLUSTER --query nodeResourceGroup --output tsv)

# Get network profile ID
NETWORK_PROFILE_ID=$(az network profile list --resource-group $NODE_RES_GROUP --query "[0].id" --output tsv)

# Delete the network profile
az network profile delete --id $NETWORK_PROFILE_ID -y

# Grab the service association link ID
SAL_ID=$(az network vnet subnet show --resource-group $RES_GROUP --vnet-name $AKS_VNET --name $AKS_SUBNET --query id --output tsv)/providers/Microsoft.ContainerInstance/serviceAssociationLinks/default

# Delete the service association link for the subnet
az resource delete --ids $SAL_ID --api-version 2021-10-01

# Delete the subnet delegation to Azure Container Instances
az network vnet subnet update --resource-group $RES_GROUP --vnet-name $AKS_VNET --name $AKS_SUBNET --remove delegations

Next steps

In this article, you scheduled a pod on the virtual node and assigned a private internal IP address. You could instead create a service deployment and route traffic to your pod through a load balancer or ingress controller. For more information, see Create a basic ingress controller in AKS.

Virtual nodes are often one component of a scaling solution in AKS. For more information on scaling solutions, see the following articles: