Skip to content

Nodes and Node Groups

NETWAYS Managed Kubernetes clusters consist of two or more node groups: one node group for the control plane and one or more node groups for workers.

Deletion of Node Groups

The node groups created when a NETWAYS Managed Kubernetes cluster is provisioned cannot be deleted.

For the worker node group, however, the number of nodes can be scaled down to 0.

Scaling Node Groups

Nodes are organized into node groups. If you want to create nodes or delete existing ones, you need to scale the node group. The node group determines which Flavor is used for new nodes.

To scale a node group, select the Nodegroups tab and click Resize in the context menu. Here you choose the desired number of nodes and finally click Resize Nodegroup.

Scaling the Control Plane Node Group

It is not possible to scale the control‑plane node group. When a NETWAYS Managed Kubernetes cluster is created, this group is fixed at 1 or 3 nodes.

Creating Node Groups

To create a new node group, click the Create Nodegroup button and adjust the node group to your preferences.

Deleting Nodes

Deleting unhealthy nodes differs depending on the NETWAYS Managed Kubernetes® variant you're running. You can identify the variant by looking at your NETWAYS Managed Kubernetes® project name in MyNWS:

  • gardener in the project name: see information for NETWAYS Managed Kubernetes® (Gardener) below.
  • openstack in the project name: see information for NETWAYS Managed Kubernetes® (Legacy) below.

Both, NETWAYS Managed Kubernetes® as well as its Legacy variant (with autoscaling enabled) detect unhealthy nodes automatically after some time and will delete unhealthy nodes on their own. Manual deletion should only be the fallback scenario.

Node Autoscaling

Depending on your cluster's autoscaling settings, nodes might be added back to your cluster automatically after deleting one or more nodes.

Deleting nodes manually should only be done if nodes are unhealthy. In a healthy cluster, remove nodes by scaling node groups down instead.

NETWAYS Managed Kubernetes® (Gardener)

In the newer NETWAYS Managed Kubernetes® variant based on Gardener, you can delete unhealthy nodes by removing the respective Kubernetes API object:

kubectl delete node <node>

This will trigger the following actions:

  1. The node gets marked for deletion.
  2. The node VM status gets evaluated.
  3. The node gets drained.
  4. Node finalizers get deleted.
  5. The node VM gets deleted.

Automatic Node Draining

The automatic draining of nodes follows a best-effort model. Depending on things like PodDisruptionBudgets, node taints, and other configuration within Kubernetes, you might have to clean up/drain the node manually before deleting it.

NETWAYS Managed Kubernetes® (Legacy)

You can delete unhealthy nodes by deleting it in MyNWS in your cluster settings under Servers.

Draining Nodes

To make sure that scaling down a node group will delete the desired node, you have to exclude it from scheduling and drain it first:

kubectl cordon <nodename>
kubectl drain <nodename>

After successfully deleting the node, NETWAYS Managed Kubernetes® will automatically remove the corresponding API object from your cluster.