ReplicaSet is a controller for horizontal scaling of pods. ReplicaSet ensures that a specific number of pod replicas are running at any given time.

ReplicaSet is the upgraded version of ReplicationController. You should always first try to use ReplicaSet; they should have the same functionality except ReplicaSet supports more complex selectors.

Consider using Deployment controller instead. ReplicaSet can be used independently, but is mainly used with Deployment to manage the underlying pods.

Deleting a ReplicaSet though REST API won't delete the pods. First scale to 0, then delete the controller. kubectl delete will do this automatically as --cascade=true is the default.

Deleting just the ReplicaSet allows you to replace it. Any other ReplicaSet or Deployment can adopt orphaned pods if the label selector matches.

You can remove pods from a ReplicaSet by changing the labels. Useful for debugging and data recovery.

YAML:

• .metadata.labels the same as you have in the pod template
• .spec.replica tells how many pod replicas to maintain
• .spec.selector manage all pods that match this selector, make sure no other Deployment or ReplicaSet has pod labels that would match this selector
• .spec.template defined the pod to replicate
• .spec.template.metadata.labels must match selector

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: frontend
  labels:
    app: guestbook
    tier: frontend
spec:
  # this replicas value is default
  # modify it according to your case
  replicas: 3
  selector:
    matchLabels:
      tier: frontend
    matchExpressions:
      - { key: tier, operator: In, values: [ frontend ] }
  template:
    metadata:
      labels:
        app: guestbook
        tier: frontend
    spec:
      containers:
        - name: php-redis
          image: gcr.io/google_samples/gb-frontend:v3
          resources:
            requests:
              cpu: 100m
              memory: 100Mi
          env:
            - name: GET_HOSTS_FROM
              value: dns
          ports:
            - containerPort: 80

kubectl create -f frontend.yaml
kubectl describe rs/frontend
kubectl get pods

You can use Horizontal Pod Autoscalers (HPA) to scale ReplicaSets or Deployments. This does require Metrics Server to be deployed on your cluster.

apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
  name: frontend-scaler
spec:
  scaleTargetRef:
    kind: ReplicaSet
    name: frontend
  minReplicas: 3
  maxReplicas: 10
  targetCPUUtilizationPercentage: 50

kubectl create -f hpa.yaml

# or, simply
kubectl autoscale rs frontend

Sources

• Kubernetes Docs