kubectl Cheat Sheet

• minikube start
  • Starts a local Kubernetes cluster
  • Automatically configures kubectl to point to the cluster
• minikube stop
  • Stops the cluster without deleting it
• minikube delete
  • Completely deletes the Kubernetes cluster
  • Removes all namespaces, system pods, and workloads
  • This is the correct way to fully reset Kubernetes

• kubectl get nodes
  • Confirms kubectl can connect to the cluster
  • Verifies node readiness
• kubectl get pods
  • Lists pods in the current namespace
• kubectl get pods -A
  • Lists pods across all namespaces
  • Includes system namespaces like kube-system

• kubectl describe pod
  • Shows detailed pod configuration and runtime state
  • Displays events (very important for debugging)
  • First command to use when something is broken

• doskey k=kubectl $*
  • Creates a temporary shortcut so “k” can be used instead of “kubectl”
  • Only lasts for the current Command Prompt session
  • Windows CMD does not support Linux-style aliases

• kubectl get pods
  • Default, human-readable summary output
• kubectl get pods -o wide
  • Adds pod IP and node name
• kubectl get pod -o yaml
  • Shows the full Kubernetes object definition
  • Useful for learning resource structure
• kubectl get pods -o name
  • Outputs only resource names
  • Useful for scripting
• kubectl get pods -o custom-columns=NAME:.metadata.name,PHASE:.status.phase
  • Displays selected fields in custom columns
• kubectl get pods -o jsonpath=“{.items[*].metadata.name}“
  • Extracts specific values from Kubernetes API output
  • Foundation for automation and scripting

• Imperative commands
  • Tell Kubernetes what to do immediately
  • Fast but not repeatable or version-controlled
• Examples:
  • Creating deployments
  • Scaling replicas directly
• Declarative workflow
  • Define desired state in YAML
  • Kubernetes reconciles actual state to match
• Workflow:
  • Export YAML from a resource
  • Edit YAML using Notepad or VS Code
  • Apply YAML with kubectl apply
• Declarative is preferred in production environments.

Standard debugging flow:
1. kubectl get pods → Identify failing or pending pods
2. kubectl describe pod → Look at events and status
3. kubectl logs → View application logs
4. kubectl logs –previous → View logs from a crashed container
5. kubectl exec -it – sh → Access a running container shell
6. kubectl debug -it –image=busybox –target=[PODNAME] → Attach an ephemeral debug container
Important distinction:

• Pod name and container name are different
• Container names are found via describe or JSONPath

• kubectl create namespace sandbox
  • Creates an isolated workspace
• kubectl get pods -n sandbox
  • Lists resources in a specific namespace
• kubectl config set-context –current –namespace=sandbox
  • Sets default namespace for the current context
• kubectl config current-context
  • Shows which cluster/context kubectl is using
• Namespaces are the safest unit for testing and cleanup.

• kubectl delete deployment demo
  • Removes user-created workloads
• kubectl delete namespace sandbox
  • Deletes everything inside that namespace
• minikube delete
  • Fully resets Kubernetes
  • Only supported way to remove kube-system indirectly

• Created automatically by Kubernetes
• Contains control plane and core system components
• Cannot and should not be deleted
• A clean cluster always includes kube-system pods
• If you want kube-system gone, delete the entire cluster instead.

• Windows CMD does not support alias; use doskey
• YAML editing is done with external editors (Notepad, VS Code)
• JSONPath requires double quotes in CMD
• Many Kubernetes tutorials assume Linux shells and need adjustment

• Thinking kubectl installs or runs Kubernetes
  • kubectl is only a client
• Expecting an empty cluster after cleanup
  • System pods in kube-system always exist
• Trying to delete kube-system
  • This is intentionally forbidden
• Forgetting namespaces when resources seem missing
  • Default namespace may be empty
• Confusing pod names with container names
  • Multi-container pods require explicit container targeting
• Using imperative commands for long-term management
  • Declarative YAML is the production standard
• Assuming Linux shell commands work on Windows CMD
  • Alias, quoting, and editors differ
• Editing live resources instead of source YAML
  • kubectl edit is convenient but not Git-friendly

• kube-system = operating system processes
• Your namespaces = application folders
• kubectl delete deployment = uninstalling an app
• minikube delete = reinstalling the OS

AI Home