Summary - ReplicaSet

Combine YAML, selector and scale to manage a ReplicaSet.

Summary: ReplicaSet

In this group:

A ReplicaSet keeps the number of pods matching its selector at replicas; it does no version management.
It isn't used directly; a Deployment creates it, and old ReplicaSets are kept for rollback.
It owns pods by label; a pod whose label changes is detached and a new one opens in its place.
Scaling is done through the top owner (Deployment); --current-replicas scales conditionally.
Self-healing: the kubelet restarts containers, the ReplicaSet recreates pods; logs --previous shows the crashed one.

kubectl get rs -l app=web
kubectl label pod <pod> app=inceleme --overwrite
kubectl logs <pod> --previous

Exercise A ReplicaSet with matchExpressions

Write a ReplicaSet YAML: name api, replicas: 2; the selector uses matchExpressions with key: app, operator: In, values: [api]; template label app: api; container named api, image api:1.0.

Show example solution

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: api
spec:
  replicas: 2
  selector:
    matchExpressions:
      - key: app
        operator: In
        values: [api]
  template:
    metadata:
      labels:
        app: api
    spec:
      containers:
        - name: api
          image: api:1.0

Test yourself

When a ReplicaSet template's image changes, what happens to the existing pods?
- They are deleted
- They update gradually
- Nothing; only newly created ones get the new image
- They are recreated immediately
Show answer

Nothing; only newly created ones get the new image — Rolling updates are the Deployment's job.

Show the pods with their labels.

kubectl get pods ____

Show answer

kubectl get pods --show-labels

Match.

ReplicaSet

Replica count
Deployment

Version transitions
label

Ownership criterion
reconciliation loop

Observe, compare, fix

Next lesson: What is a Service?

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies