Lesson  in  Kubelings — Learn Kubernetes the Rustlings Way

Manifests: write down what you want, then apply it

Real clusters are not built with one-off commands. Write a Pod as a YAML manifest (apiVersion, kind, metadata, spec), give it labels, and create it with kubectl apply, the command every later lesson uses.

The situation

kubectl run is fine for a quick test, but nobody can review it, put it in git, or run it again next week. Real clusters are described in manifests: YAML files that say what should exist. kubectl apply -f file.yaml sends the file to the API server, and Kubernetes makes the cluster match it.

Every manifest has the same four top-level fields:

apiVersion: v1          # which version of the API defines this kind
kind: Pod               # what sort of object
metadata:               # who it is: name, namespace, labels
  name: web
spec:                   # what you want it to be
  containers: [...]

Labels are key/value tags in metadata.labels. They do nothing on their own, but Services, Deployments and kubectl get -l all find pods by label, so almost every pod has them.

YAML is indentation-sensitive. Use spaces, never tabs, and keep items of the same list at the same depth.

Your task

Save this as web.yaml, fill in the three TODO lines, and apply it:

apiVersion: v1
kind: Pod
metadata:
  name: web
  namespace: kubelings
  labels:
    app: web
    # TODO: add a second label, tier, with the value frontend
spec:
  containers:
    - name: web
      image: nginx:1.27-alpine
      ports:
        - containerPort: 0      # TODO: nginx listens on port 80
      resources:
        limits:
          memory: 64Mi          # TODO: raise the limit to 128Mi
kubectl apply -f web.yaml
kubectl -n kubelings get pods --show-labels
Hint

The new label goes directly under app: web, indented the same amount.

Made a mistake? Edit the file and run kubectl apply -f web.yaml again. A few fields of a running pod cannot change (the port is one of them). If apply says the change is forbidden, delete the pod and apply again:

kubectl -n kubelings delete pod web
kubectl apply -f web.yaml
Solution
apiVersion: v1
kind: Pod
metadata:
  name: web
  namespace: kubelings
  labels:
    app: web
    tier: frontend
spec:
  containers:
    - name: web
      image: nginx:1.27-alpine
      ports:
        - containerPort: 80
      resources:
        limits:
          memory: 128Mi
kubectl apply -f web.yaml
kubectl -n kubelings get pods -l tier=frontend

Why apply

kubectl apply is declarative: the file is the record of what you want. Run it twice and nothing changes the second time. Change the file and apply again, and Kubernetes changes only what differs. kubectl create fails if the object already exists, and kubectl run leaves no file behind.

Reading any object back as YAML

kubectl -n kubelings get pod web -o yaml

The output is longer than what you wrote. Kubernetes fills in defaults and adds a status: section describing what is actually happening. Your manifest is the spec, the desired state. status is the observed state.