I have a deployment that is supposed to scale between 5 and 20 pods. For the sake of simplicity, lets say that the pod requests 128Mi and has a limit of 512Mi, so I want the memory usage of the pod between 305Mi and 410Mi.

If the avarage memory usage is above 410Mi a new pod should go up. If the average memory usage is below 305Mi, one pod should be shutdown.

by using:

spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: myDeployment
  minReplicas: 5
  maxReplicas: 20
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50
        #type: AvarageValue
        #avarageValue: 305
        #avarageValue: 410

and the problem is that im getting A LOT of flapping:

1. a new pod goes up,
2. avg load drops below “target value”,
3. pod gets removed,
4. avg load goes above “target value”,
5. new pod provisioned,
6. and it keeps flapping all day long until about 1am when service utilization drops