Create and use a custom load balancer in Azure Kubernetes Service on Azure Stack HCI and Windows Server

Applies to: AKS on Azure Stack HCI 22H2, AKS on Windows Server

This article covers how to create and use a custom load balancer. In Azure Kubernetes Service (AKS) on Azure Stack HCI and Windows Server, you use load balancers to send requests to the Kubernetes API server and to manage traffic to application services. When using a custom load balancer, kube-vip automatically deploys to manage the load balancing of requests to the Kubernetes API server and to make sure that it's highly available.

Note

You can also use a other load balancers, such as MetalLB or Software Defined Networking (SDN) load balancing, to load balance traffic to application services.

Before you begin

You must have installed AKS on Azure Stack HCI and Windows Server and provided a range of virtual IP addresses for the load balancer during the network configuration step during installation.

Configure a custom load balancer

Warning

If you choose to deploy your own load balancer, the Kubernetes cluster will be unreachable after installation. If you deploy any services with type=LoadBalancer, the services will also be unreachable until you configure your load balancer.

This configuration assumes you want to leverage a custom load balancer in your cluster. In this case, the workload cluster is deployed without a load balancer.

1. Create a load balancer configuration using the New-AksHciLoadBalancerSetting cmdlet and then select none for the loadBalancerSku parameter:

   $lbCfg=New-AksHciLoadBalancerSetting -name "myLb" -loadBalancerSku "none" 
   

2. Deploy a workload cluster without providing the load balancer configuration using the following command:

   New-AksHciCluster -name "summertime" -nodePoolName mynodepool -nodeCount 2 -OSType linux -nodeVmSize Standard_A4_v2 -loadBalancerSettings $lbCfg 
   

3. Use Get-AksHciCluster to verify that the cluster is successfully deployed with the control plane nodes running kube-vip and that the API server requests are reachable.

4. Manually configure your load balancer.

If you run an upgrade, the load balancer configuration (loadBalancerSku and count) you defined during installation will remain the same after the upgrade completes. However, if you want to update loadBalancerSku during an upgrade, you must redeploy your workload clusters. If you have existing clusters running a HAProxy-based load balancer, you can continue running your workloads and the upgrade will successfully complete.

Important

If you change from using a custom load balancer to using the default load balancer, you're required to redeploy your workload cluster with the new load balancer configuration. For instructions on how to configure the default load balancer, see Configure load balancer.

Next steps

To learn more about Kubernetes services, see the Kubernetes services documentation.