Bits of .NET

📈 HPA Scales Pods. VPA Scales Pod Resources.

Guessing CPU/memory requests is hard. Vertical Pod Autoscaler recommends or automatically adjusts resource requests based on actual usage.

📝 Install VPA

git clone https://github.com/kubernetes/autoscaler.git
cd autoscaler/vertical-pod-autoscaler
./hack/vpa-up.sh

kubectl get pods -n kube-system | grep vpa

🎯 VPA Configuration

apiVersion: autoscaling.k8s.io/v1
kind: VerticalPodAutoscaler
metadata:
  name: myapp-vpa
spec:
  targetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: myapp
  updatePolicy:
    updateMode: "Auto"  # or "Off", "Initial", "Recreate"
  resourcePolicy:
    containerPolicies:
    - containerName: '*'
      minAllowed:
        cpu: 100m
        memory: 256Mi
      maxAllowed:
        cpu: 2
        memory: 4Gi

# updateMode:
# - Off: Only recommendations, no changes
# - Initial: Set requests at creation only
# - Auto: Automatically update pods (may restart)
# - Recreate: Restart pods when needed

✅ Check Recommendations

kubectl describe vpa myapp-vpa

# Output shows:
# Recommendation:
#   Container Recommendations:
#     Container Name:  myapp
#     Lower Bound:     cpu: 250m, memory: 512Mi
#     Target:          cpu: 500m, memory: 1Gi
#     Upper Bound:     cpu: 1,    memory: 2Gi
#     Uncapped Target: cpu: 800m, memory: 1.5Gi

# Apply recommendations manually
kubectl edit deployment myapp
# Update requests to match Target values

💡 Best Practices

• Start with updateMode: “Off” to see recommendations first
• Set minAllowed to prevent under-provisioning
• Set maxAllowed to prevent runaway pods
• Don’t use VPA with HPA on same metrics (CPU/memory)
• VPA may restart pods to apply changes