Placement Groups

Hetzner Cloud Placement Groups allow you to control how your Kubernetes nodes are distributed across physical hosts within Hetzner’s infrastructure. By assigning nodes to a placement group of type spread, you ensure that each node is scheduled on a different physical machine, increasing fault tolerance and availability.

This feature is especially useful for high-availability Kubernetes clusters, as it minimizes the risk of multiple nodes failing due to hardware issues on a single host.

Creating placement groups#

In the Cluster resource, you can create placement groups in the spec.topology.variables section. The following example creates two placement groups:

yaml

		- name: hcloudPlacementGroups
  value:
    - name: myPlacementGroup1
      type: spread
    - name: myPlacementGroup2
      type: spread

Note

As of now, Hetzner Cloud only supports the spread placement group type. If you want to know more, you can read the Hetzner documentation.

Adding placement groups to machine deployments#

To add placement groups to your machine deployments, you can set them in variables.overrides:

yaml

		machineDeployments:
  - class: workeramd64hcloud
    name: md-0
    replicas: 5
    failureDomain: nbg1
    variables:
      overrides:
        - name: workerMachineTypeHcloud
          value: cpx42
        - name: workerMachinePlacementGroupNameHcloud
          value: myPlacementGroup1

The name should match an existing placement group, created in your Cluster resource topology.variables:

Adding placement groups to control plane nodes#

You can set control plane placement groups in spec.topology.variables:

yaml

		- name: controlPlanePlacementGroupNameHcloud
  value: myPlacementGroup2

Here is an example of a complete Cluster resource with placement groups:

cluster.yamlyaml

		apiVersion: cluster.x-k8s.io/v1beta2
kind: Cluster
metadata:
  name: mycluster
spec:
  clusterNetwork:
    services:
      cidrBlocks: ["10.128.0.0/12"]
    pods:
      cidrBlocks: ["192.168.0.0/16"]
    serviceDomain: "cluster.local"
  topology:
    classRef:
      name: hetzner-apalla-1-35-v4
    version: v1.35.7
    controlPlane:
      replicas: 3
    variables:
      - name: hcloudPlacementGroups
        value:
          - name: myPlacementGroup1
            type: spread
          - name: myPlacementGroup2
            type: spread
      - name: controlPlanePlacementGroupNameHcloud
        value: myPlacementGroup2
    workers:
      machineDeployments:
        - class: workeramd64hcloud
          name: md-0
          replicas: 5
          failureDomain: nbg1
          variables:
            overrides:
              - name: workerMachineTypeHcloud
                value: cpx42
              - name: workerMachinePlacementGroupNameHcloud
                value: myPlacementGroup1

Rebooting bare metal nodes

Reboot bare metal servers for maintenance without triggering a reprovisioning in Syself Autopilot.

Taints

Why taints are not supported.