Vertical Pod Autoscaler

The Vertical Pod Autoscaler (VPA) automatically adjusts the CPU and memory resources of pods. It continuously analyzes the actual consumption of workloads, then recommends or applies adjustments.

Component	Role
recommender	Analyzes metrics and recommends resources for pods
updater	Recreates pods when the recommendations change
admissionController	Applies the recommended resources when pods are created

In the console

The Vertical Pod Autoscaler is part of the Advanced Configuration of the Addons step: it is always present in the cluster and cannot be disabled. You can only override its configuration.

1. At creation (Addons step) or from Edit > Extensions & Addons, expand the Vertical Pod Autoscaler block in the Advanced Configuration section.
2. Enter your values in Helm Configuration (YAML) — optional.
3. Confirm with Next then Create cluster (creation) or Save (modification).

note

Clearing the field, or a line of the override, from Edit does not remove a value already applied to the cluster. Contact support to return to the default values.

On the cluster detail page, the VPA line of the Network section shows VPA when the addon is configured.

Override the configuration

The YAML value is passed to the VPA Helm chart, under the vertical-pod-autoscaler key. For example, to disable the updater and only use the recommendations:

vpa-override.yaml

vertical-pod-autoscaler:
  updater:
    enabled: false

The available options are described in the Vertical Pod Autoscaler Helm chart.

Usage in the cluster

Create a VerticalPodAutoscaler object for each workload to track:

vpa-my-app.yaml

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: my-app
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: my-app
  updatePolicy:
    updateMode: "Off"

kubectl apply -f vpa-my-app.yaml

# Read the recommendations
kubectl describe vpa my-app

Best practices

• Start with updateMode: "Off" to observe the recommendations before applying them.
• Do not use the VPA and a HorizontalPodAutoscaler on the same metric (CPU or memory) of the same workload.
• Combine the VPA with node group autoscaling to adapt both the pods and the cluster capacity.