kubectl Cheat Sheet

Resource names can be singular, plural, or abbreviated. A namespace flag narrows namespaced resources, while cluster-scoped resources ignore it.

kubectl get pods
kubectl get pods -o wide
kubectl get deployment web -o yaml
kubectl get pods -A
kubectl describe deployment web
kubectl describe pod POD_NAME

kubectl get all covers a conventional subset of workload resources, not every resource type in the namespace. Use kubectl api-resources when the full discovery list matters.

create -f submits a new object and normally fails if it already exists. apply -f reconciles a declarative manifest, creating a missing object or updating an existing one.

kubectl apply -f deployment.yaml
kubectl apply -f ./manifests/
kubectl create -f one-time-job.yaml
kubectl diff -f deployment.yaml
kubectl delete -f deployment.yaml
kubectl delete pods -l app=web

Standard input is useful when another command generates or transforms a manifest:

kubectl create deployment web --image=nginx:stable \
  --dry-run=client -o yaml | kubectl apply -f -

--dry-run=client -o yaml renders locally without creating the object; apply -f - reads that YAML from standard input.

A context combines a cluster, user, and optional default namespace. use-context changes the persistent current context in the selected kubeconfig.

kubectl config current-context
kubectl config get-contexts
kubectl config use-context staging
kubectl config set-context --current --namespace=team-a
kubectl get pods -n team-b
kubectl get pods --all-namespaces

Before changing a production object, run kubectl config current-context and include -n explicitly when the namespace is operationally important.

Logs are container-specific. A pod with multiple containers needs -c; --previous reads the terminated instance of a restarted container when its prior log is still available.

kubectl logs POD_NAME
kubectl logs -f POD_NAME -c CONTAINER_NAME
kubectl logs --previous POD_NAME -c CONTAINER_NAME
kubectl exec -it POD_NAME -- sh
kubectl exec -it POD_NAME -c CONTAINER_NAME -- sh
kubectl port-forward service/web 8080:80

The -- separator ends kubectl option parsing; everything after it is the command and arguments executed in the container.

Labels are indexed metadata used by Services, controllers, and kubectl selectors. Changing a label can alter which controller or Service selects an object, so inspect selectors first.

kubectl get pods --show-labels
kubectl get pods -l app=web
kubectl get pods -l 'environment in (staging,production)'
kubectl label deployment web environment=staging
kubectl label deployment web environment=production --overwrite

Imperative generators can provide a starting manifest without changing the cluster:

kubectl create service clusterip web --tcp=80:8080 --dry-run=client -o yaml
kubectl create configmap app-config --from-literal=mode=production --dry-run=client -o yaml

Review generated YAML before storing or applying it, because client version and command defaults influence the output.

Deployment rollout commands observe or revise the controller's revision history. A manual replica count can later be changed by a HorizontalPodAutoscaler if one manages the same workload.

kubectl scale deployment/web --replicas=5
kubectl set image deployment/web app=example/web:2.0
kubectl rollout status deployment/web --timeout=2m
kubectl rollout history deployment/web
kubectl rollout undo deployment/web
kubectl rollout undo deployment/web --to-revision=2

top reads the resource metrics API, which is commonly supplied by Metrics Server. An unavailable metrics API makes top fail even when workloads are otherwise healthy.

kubectl top nodes
kubectl top pods
kubectl top pods -A --containers
kubectl api-resources
kubectl api-versions
kubectl explain deployment
kubectl explain deployment.spec.template.spec.containers
kubectl auth can-i create deployments -n team-a

• kubectl documentation
• kubectl command reference
• Kubernetes labels and selectors
• Kubernetes declarative object management