Amazon EKS — HPA — Horizontal Pod Autoscaler

1 minute read

Horizontal Pod Autoscaler (HPA)

HPA automatically scales the number of pod replicas based on CPU, memory, or custom metrics.

How HPA Works

HPA Control Loop (every 15 seconds)
┌─────────────────────────────────────────────────────┐
│                                                     │
│  1. HPA checks current metric value                 │
│  2. Compares with target value                      │
│  3. Calculates desired replicas:                    │
│     desired = ceil(current × (currentValue/target)) │
│  4. Scales Deployment up or down                    │
│                                                     │
│  Example:                                           │
│  Current: 3 replicas at 80% CPU                     │
│  Target: 50% CPU                                    │
│  Desired: ceil(3 × 80/50) = ceil(4.8) = 5 replicas │
│                                                     │
└─────────────────────────────────────────────────────┘

Prerequisites: Metrics Server

# Check if metrics server is running
$ kubectl top nodes
# If "error: Metrics API not available", install it:

$ kubectl apply -f https://github.com/kubernetes-sigs/metrics-server/releases/latest/download/components.yaml

Create HPA

CLI

$ kubectl autoscale deployment web --cpu-percent=50 --min=2 --max=20

YAML

apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: web-hpa
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: web
  minReplicas: 2
  maxReplicas: 20
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 50
    - type: Resource
      resource:
        name: memory
        target:
          type: Utilization
          averageUtilization: 70
  behavior:
    scaleDown:
      stabilizationWindowSeconds: 300   # Wait 5 min before scaling down
    scaleUp:
      stabilizationWindowSeconds: 0     # Scale up immediately

Monitor HPA

$ kubectl get hpa
NAME      REFERENCE        TARGETS           MINPODS   MAXPODS   REPLICAS
web-hpa   Deployment/web   35%/50%, 40%/70%  2         20        3

$ kubectl describe hpa web-hpa

Common Mistakes

Mistake Impact
No resource requests on pods HPA can’t calculate utilization percentage
Too low min replicas Slow scale-up during sudden traffic
No scale-down stabilization Flapping (scale up/down repeatedly)
Metrics Server not installed HPA shows <unknown> for metrics

What’s Next?

Right-size pods: VPA.

You may also enjoy