Description:
- Horizontal Scaling
- HPA updates a workload resource to match demand
- by depploying more/less pods
- doesnt apply for objects that cant be scaled, ex: Kubernetes DaemonSet
- runs inside Kubernetes Control Plane
kubectl get hpa
behaviour?:
- scaledown:
- stabilizationWindowSeconds: helps keep a pod alive after created to prevent thrashing
How does HPA work:
- the interval is set by
--horizontal-pod-autoscaler-sync-periodparameter to the kube-controller-manager (and the default interval is 15 seconds). - metrics-server is required