Skip to main content

Autoscaling Node Pools

Kube-Hetzner supports autoscaling node pools powered by the Kubernetes Cluster Autoscaler.

Enabling Autoscaling​

To enable autoscaling, add at least one map to the array of autoscaler_nodepools in your configuration:

kube.tf
autoscaler_nodepools = [
  {
    name        = "autoscaler-agent"
    server_type = "cx22"
    location    = "nbg1"
    min_count   = 1
    max_count   = 10
    labels = {
      "node.kubernetes.io/type" = "autoscaler"
    }
    taints = [
      {
        key    = "dedicated"
        value  = "autoscaler"
        effect = "NoSchedule"
      }
    ]
  }
]

Note: The old autoscaler_labels and autoscaler_taints variables are deprecated. Use the labels map and taints list of objects directly within the autoscaler_nodepools block as shown above.

Important Considerations​

The nodes are booted based on a snapshot that is created from the initial control_plane. So please ensure that the disk of your chosen server type is at least the same size (or bigger) as the one of the first control_plane.

Configuration Options​

Required Parameters​

  • name: Name of the nodepool
  • server_type: Hetzner Cloud server type
  • location: Datacenter location
  • min_count: Minimum number of nodes
  • max_count: Maximum number of nodes

Optional Parameters​

  • labels: Labels to apply to nodes
  • taints: Taints to apply to nodes
  • instance_type: Specific instance type (alternative to server_type)
  • placement_group: Placement group for the nodes

Autoscaler Configuration​

You can configure additional autoscaler settings:

kube.tf
autoscaler_enabled = true
autoscaler_scale_down_enabled = true
autoscaler_scale_down_delay = "10m"
autoscaler_scale_down_unneeded_time = "10m"

Monitoring Autoscaler​

To monitor the autoscaler, check the logs of the autoscaler pod:

Terminal
kubectl -n kube-system logs -l app=cluster-autoscaler

Best Practices​

  1. Set appropriate limits: Choose min_count and max_count based on your workload requirements
  2. Use labels and taints: Properly label autoscaled nodes for workload scheduling
  3. Monitor costs: Keep an eye on the number of nodes to control costs
  4. Test scaling: Test both scale-up and scale-down scenarios to ensure proper behavior