Autoscale a Deployment with an HPA (1 → 5)

The situation

The php-apache Deployment in kubelings runs a single replica and falls over under load. metrics-server is already collecting CPU usage. You need a HorizontalPodAutoscaler so the app scales out on CPU and back in when idle.

         ┌──────────┐     
         │CPU usage │     
         │          │     
         └──────────┘     
             │   ▲        
             │   └───┐    
             │       │    
      metrics-server │    
             │       │    
             ▼       │    
 ┌───────────────┐   │    
 │HPA target 50% │   │    
 │               │   │    
 └───────────────┘   │    
             │       │    
      scale out or in│    
             │       │    
             ▼       │    
        ┌────────────────┐
        │replicas 1 to 5 │
        │                │
        └────────────────┘

Your task

Create an HPA named php-apache that:

Targets the php-apache Deployment.
minReplicas: 1, maxReplicas: 5.
Scales on CPU (e.g. target 50% average utilization).

kubectl -n kubelings top pods          # metrics-server is live
kubectl -n kubelings get hpa

Hint

kubectl -n kubelings autoscale deploy php-apache --cpu=50% --min=1 --max=5

(Generate a load and watch with kubectl -n kubelings get hpa -w to see it climb.)

Solution

Approach

An HPA needs (a) metrics-server (already running) and (b) a CPU request on the target pods to compute utilization against — php-apache has requests.cpu: 200m.

Create the HPA

Imperative (simplest):

kubectl -n kubelings autoscale deploy php-apache --cpu=50% --min=1 --max=5

or declarative (autoscaling/v2):

kubectl -n kubelings apply -f - <<'EOF'
apiVersion: autoscaling/v2
kind: HorizontalPodAutoscaler
metadata:
  name: php-apache
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: php-apache
  minReplicas: 1
  maxReplicas: 5
  metrics:
    - type: Resource
      resource:
        name: cpu
        target:
          type: Utilization
          averageUtilization: 50
EOF

Verify

kubectl -n kubelings get hpa php-apache
# generate load:
kubectl -n kubelings run load --rm -it --image=ghcr.io/iximiuz/labs/busybox:latest -- \
  sh -c "while true; do wget -q -O- http://php-apache; done"
kubectl -n kubelings get hpa php-apache -w   # replicas climb toward 5