I have followed the k8s document for setup the HPA based on CPU percentage(CPU >50%) for one of the my deployment and Configured scaling behavior and specify a stabilization window that prevents flapping the replica count for a scaling target.

behavior:
  scaleDown:
    stabilizationWindowSeconds: 180
    policies:
    - type: Percent
      value: 10
      periodSeconds: 30
  scaleUp:
    stabilizationWindowSeconds: 0
    policies:
    - type: Pods
      value: 4
      periodSeconds: 30
    selectPolicy: Max

After created HPA, increased the load the deployment to test the HPA and it’s was working as per the configuration.

NAME         REFERENCE                     TARGET      MINPODS   MAXPODS   REPLICAS   AGE
php-apache   Deployment/php-apache/scale   105% / 50%  1         10        3          3m

Based on the scaleUp stabilizationWindowSeconds configuration, it is creating the only 4 pods in 30 seconds period if the CPU % is more then 50(as per the above o/p the CPU value is 105%) and it not importing the application/user.

If suddenly traffic increased and CPU percentage will reached to more the 300%, above scaleUp conditions are impacting the end users and requests are in queue till new required pods come to online to serve the traffic.

Is there any way to set the behavior.scaleUp.policies.value value dynamically based on CPU percentages?

I want to scale up the Pods count in 15 sec like below