Min and max replicas

Learn minReplicas and maxReplicas fields.

Min and max replicas

minReplicas is the base replica count kept even when load drops; for high availability, at least 2 is usually chosen. maxReplicas is the ceiling limiting cost and back-end (database) load. The calculated value is clipped to this range:

# calculated 14, maxReplicas 10
$ kubectl describe hpa api | grep ScalingLimited
  ScalingLimited  True  TooManyReplicas  the desired replica count is more than the maximum replica count

Scaling behaviour

The behavior field adjusts the speed of scaling. By default there's a 5-minute stabilization window for scaling down: pods aren't closed immediately on short dips (prevents flapping):

  behavior:
    scaleUp:
      policies:
        - type: Percent
          value: 100          # at most double every 15 s
          periodSeconds: 15
    scaleDown:
      stabilizationWindowSeconds: 300
      policies:
        - type: Pods
          value: 1            # reduce at most 1 pod per minute
          periodSeconds: 60

Exercise An HPA that shrinks slowly

Write an HPA's spec section (starting with spec:): target apps/v1 Deployment api; minReplicas: 3, maxReplicas: 20; behavior.scaleDown.stabilizationWindowSeconds: 600.

Show example solution

spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: api
  minReplicas: 3
  maxReplicas: 20
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 600

Test yourself

If you see ScalingLimited True and TooManyReplicas, what does it mean?
- minReplicas is too high
- There are no metrics
- The load exceeds what maxReplicas can handle; the ceiling was hit
- The HPA is broken
Show answer

The load exceeds what maxReplicas can handle; the ceiling was hit — The ceiling must be raised or pod efficiency improved.

Give a waiting window before scaling down.

    scaleDown:
      ____: 300

Show answer

    scaleDown:
      stabilizationWindowSeconds: 300

Match.

minReplicas

Floor
maxReplicas

Ceiling
stabilizationWindowSeconds

Prevents flapping
TooManyReplicas

Hit the ceiling

Next lesson: Metrics server

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies