Skip to main content

Autoscaling Node Pools

An autoscale node pool automatically adjusts its worker node count between a minimum and a maximum based on a scaling policy. This lets a cluster add capacity under load and release it when idle, so you pay for what your workload actually needs.

You enable autoscaling and pick a policy when you create a cluster or add/edit a node pool. This page explains the policy options.

A pool can use an Elastic Policy, a Scheduled Policy, or both together.

Elastic Policy​

The Elastic Policy scales the pool based on resource utilization. It comes in two types: Default or Custom.

Default​

The default policy scales on a built-in metric:

  • CPU - scales the node count based on CPU utilization. When CPU utilization crosses the threshold, nodes are added; when it falls, nodes are removed.
  • Memory - scales the node count based on memory utilization, the same way.

You define the scale-up and scale-down behavior with:

  • Node Utilization - the threshold values that trigger a scale-up (increase the node count) or scale-down (decrease the node count).
  • Scaling Period Policy:
    • Watch Period - how many periods the metric must hold before scaling.
    • Period Duration - the length of each period, in seconds.
    • Cooldown - how long to wait after a scaling action before scaling again, in seconds.

Custom​

A custom policy scales on an attribute you define instead of CPU or memory - for example, network traffic (NETTX) or disk write operations (DISKWRIOPS).

  • Policy Parameter Name - the name of the custom attribute to monitor. It must contain only uppercase letters, digits, and single underscores between segments (it cannot be CPU or MEMORY).
  • The custom parameter's value defaults to 0 at launch.
  • The same Node Utilization and Scaling Period Policy fields apply.

When the custom parameter crosses the threshold, nodes are added; when it drops below, nodes are removed. You update the parameter's value yourself using a generated cURL command (see below).

Update a Custom Parameter with cURL​

  1. After the cluster is Running, open the Node Pool tab and expand the autoscale pool.
  2. Select Get cURL to generate the command for that pool's custom parameter.
  3. Copy the command. It contains placeholders for your API key and Bearer token - replace them with your real values (find them in the API section of the portal).
  4. Set the custom parameter value in the request body and run the command (in your terminal, a script, a cron job, or Postman).

You can only update the custom parameter's value with this command - not other pool settings. The dialog also shows the parameter's current value.

Scheduled Policy​

A scheduled policy scales the pool on a recurring schedule, which suits predictable traffic patterns (for example, scaling up during business hours and down overnight).

  • Recurrence - when the scaling action runs, expressed as a cron schedule.
  • Upscale Recurrence - the number of nodes to scale to at the scheduled upscale time. This value must be within the pool's maximum.
  • Downscale Recurrence - the number of nodes to scale to at the scheduled downscale time, within the pool's minimum.

Elastic and Scheduled Policy​

Select Elastic and Scheduled Policy to combine both: the pool responds to utilization in real time and follows the schedule. Configure the parameters for each, then proceed.

ResourceUse it for
Create a ClusterSet up an autoscale pool at creation.
Node PoolsEdit, resize, and view pool policies.
Auto ScalingStandalone (non-Kubernetes) auto scaling.
Last updated on June 24, 2026.