Spot instances

The CAST AI autoscaler supports running your workloads on Spot instances.
This guide will help you configure and run it in 5 minutes.

Available configurations


When to use: Spot instances are optional

When a pod is marked only with tolerations, the Kubernetes scheduler could place such a pod/pods on regular nodes as well.

  - key:
    operator: Exists

Node Selectors

When to use: only use Spot instances

If you want to make sure that a pod is scheduled on Spot instances only, add nodeSelector as well as per the example below.
The autoscaler will then ensure that only a Spot instance is picked whenever your pod requires additional workload in the cluster.

  - key:
    operator: Exists
nodeSelector: "true"

Node Affinity

When to use: Spot instances are preferred - if not available, fall back to on-demand nodes

When a Spot instance is interrupted, and on-demand nodes in the cluster have available capacity, pods that previously ran on the Spot instance will be scheduled on the available on-demand nodes if the following affinity rule is applied:

      - weight: 1
          - key:
            operator: Exists

If you want to move pods back to Spot instances, use the Rebalancer feature.

Spot Reliability

When to use: you want to minimize workload interruptions

The Autoscaler can identify which instance types are less likely to be interrupted. You can set a default reliability value in Autoscaler--> Settings. If you want to control that per-workload, e.g. leave most cost-efficient value globally and only choose more stable instances for specific pods, define this in the deployment configuration by setting label on the pod.

Here's an example of how it's done for the typical deployment:

      labels: 10

Reliability is measured by "what is the percentage of reclaimed instances during trailing month for this instance type". This tag specifies an upper limit - all instances below specified reliability value will be considered.

The value is a percentage (range is 1-100), and the meaningful values are:

  • 5: most reliable category; by using this value you'll restrict the Autoscaler to use only the narrowest set of Spot instance types;
  • 10 - 15: reasonable value range to compromise between reliability and price;
  • 25 and above: typically most instances fall into this category.


For AWS, have a look at Spot instance advisor to get an idea which instances correspond to which reliability category.

Spot fallback

CAST AI supports the fallback of Spot instances to on-demand nodes in case there is no Spot instance availability. Our Autoscaler will temporarily add an on-demand node for your Spot-only workloads to run on. Once inventory of Spot instances is again available, on-demand nodes used for the fallback will be replaced with actual Spot instances.

Fallback on-demand instances will be labeled with"true" label.

To enable this feature turn on spot fallback policy use Upsert cluster's policies configuration API.

 "spotInstances": {
    "clouds": [
    "enabled": true,
    "maxReclaimRate": 0,
    "spotBackups": {
      "enabled": true, // this parameter will enable spot fallback feature
      "spotBackupRestoreRateSeconds": 1800 // configure how often CAST AI should try to switch back to spot instances

Step-by-step deployment on Spot Instance

In this step-by-step guide, we demonstrate how to use Spot Instances with your CAST AI clusters.

To do that, we will use an example NGINX deployment configured to run only on Spot instances.

1. Enable relevant policies

To start using spot instances go to Autoscaler menu and enable the following policies:

  • Unschedulable pods policy

    • This policy requests an additional workload to be scheduled based on your deployment requirements (i.e. run on Spot instances).
  • Spot instances policy

    • This policy allows the Autoscaler to use Spot instances.

2. Example deployment

Save the following yaml file, and name it: nginx.yaml:

apiVersion: apps/v1
kind: Deployment
  name: nginx-deployment
    app: nginx
  replicas: 1
      app: nginx
        app: nginx
      nodeSelector: "true"
        - key:
          operator: Exists
        - name: nginx
          image: nginx:1.14.2
            - containerPort: 80
              cpu: '2'
              cpu: '3'

2.1. Apply the example deployment

With kubeconfig set in your current shell session, you can execute the following (or use other means of applying deployment files):

kubectl apply -f nginx.yaml

2.2. Wait several minutes

Once the deployment is created, it will take up to several minutes for the Autoscaler to pick up the information about your pending deployment and schedule the relevant workloads in order to satisfy the deployment needs, such as:

  • This deployment tolerates Spot instances
  • This deployment must run only on Spot instances

3. Spot Instance added

  • You can see your newly added Spot instance in the Node list.