> ## Documentation Index
> Fetch the complete documentation index at: https://docs.rafftechnologies.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Scale a Kubernetes node pool

> Add or remove worker nodes by hand, or set autoscaling bounds

<sub>Updated August 24, 2026</sub>

Scale a pool between 2 and 20 nodes. Scaling down never kills your workloads abruptly: each node is cordoned and drained first, respecting PodDisruptionBudgets.

## Steps

<Steps>
  <Step title="Open the pool's scale control">
    On the **Node pools** tab, use the +/− stepper on the pool card, or click the count to open the scale dialog with a cost preview.

    <Frame>
      <img src="https://mintcdn.com/rafftechnologiesllc/pydLUQqixwIwkrCw/images/products/compute/kubernetes/scale-pool.png?fit=max&auto=format&n=pydLUQqixwIwkrCw&q=85&s=25cce5450d530c046b4f34e90850cd6c" alt="Node pools tab with the scale control and per-node pricing" width="1663" height="849" data-path="images/products/compute/kubernetes/scale-pool.png" />
    </Frame>
  </Step>

  <Step title="Confirm the new count">
    Scale-up deploys new nodes (\~3 minutes each). Scale-down runs in the background: newest nodes are drained one at a time, then removed. The pool shows a **Scaling** badge until it settles.
  </Step>
</Steps>

<Note>
  **Zero-downtime rule:** a node is only removed once every pod on it has been evicted. If a PodDisruptionBudget blocks eviction for 10 minutes, scaling stops, the node is restored, and the pool's status message names the blocking pod. Workloads with a **single replica** still restart during scale-down — there is no second copy to serve while the pod moves; run 2+ replicas for anything that must stay up.
</Note>

## Autoscaling

Enable autoscaling per pool with min/max bounds (**Edit autoscale** on the pool card). The autoscaler adds nodes as soon as pods can't be scheduled for lack of CPU/memory, and removes one when a node's pod requests stay under 50% for 10 minutes and its pods fit elsewhere — always inside your bounds, with at most one scale-down per 10 minutes.

## Next steps

<CardGroup cols={2}>
  <Card title="Node pools & autoscaling" icon="layer-group" href="/products/build/kubernetes/concepts/node-pools-and-autoscaling">
    The rules the autoscaler follows
  </Card>

  <Card title="Scale via API" icon="code" href="/api-reference/kubernetes/scale-node-pool">
    Automate scaling with the REST API
  </Card>
</CardGroup>