Kubectl

Kubectl#

The kubectl utility is a primary way to interact with k8s cluster.

Different k8s resources have slightly different configurations. However, kubectl is designed in such a way that different resources often have many features in common. This page addresses the questions about managing the entire cluster with kubectl, as well as common details about managing different resource types. It is not always very detailed regarding questions specific to particular resources - there are specific pages for such details.

Install&Config#

Check installation guide for kubectl. For me, using the ubuntu option snap install kubectl --classic worked well.

To enable TAB completion add source the script printed with kubectl completion bash: source <(kubectl completion bash). To enable completion at the start of the shell, add line into ~/.bashrc file: echo 'source <(kubectl completion bash)' >> ~/.bashrc.

Config file#

The behavior of the kubectl is typically defined by the configuration file that is located in the ~/.kube/config directory.

The following cell displays a few lines of the kubectl configuration.

cat ~/.kube/config | head -n 10

apiVersion: v1
clusters:
- cluster:
    certificate-authority: /home/user/.minikube/ca.crt
    extensions:
    - extension:
        last-update: Mon, 13 Oct 2025 09:07:40 CEST
        provider: minikube.sigs.k8s.io
        version: v1.37.0
      name: cluster_info

Context#

A k8s context is a configuration setup under a convenient name. You can switch between different kubectl configurations, which provide access to different clusters or different setups for the same cluster.

The following table shows useful kubectl commands that can be beneficial for managing the context.

Command	Purpose
`kubectl config current-context`	Displays the name of the context that is currently active.
`kubectl config get-contexts`	Lists all the contexts defined in your kubeconfig file.
`kubectl config use-context <name>`	Sets the specified context as the current-context, switching the cluster/user/namespace your commands target.
`kubectl config set-context <name> --cluster=<cluster> --user=<user> --namespace=<namespace>`	Creates a new context or modifies an existing one to link a specific cluster, user, and/or default namespace.
`kubectl config delete-context <name>`	Deletes the specified context from your kubeconfig file.

Minikube, which is primarily used as an example k8s setup, automatically creates a kubectl context for each profile. For example, consider how to select a kubectl context from the options provided by minikube.

minikube start --profile one-node &> /dev/null
minikube start --profile two-nodes &> /dev/null

The following cell shows the available kubectl contexts.

kubectl config get-contexts

CURRENT   NAME        CLUSTER     AUTHINFO    NAMESPACE
          one-node    one-node    one-node    default
*         two-nodes   two-nodes   two-nodes   default

The two-nodes context is marked as CURRENT. The following cell switches contexts and displays list of contexts list again.

kubectl config use-context one-node
kubectl config get-contexts

Switched to context "one-node".
CURRENT   NAME        CLUSTER     AUTHINFO    NAMESPACE
*         one-node    one-node    one-node    default
          two-nodes   two-nodes   two-nodes   default

Now there is different context specified as CURRENT.

minikube delete --profile one-node &> /dev/null
minikube delete --profile two-nodes &> /dev/null

Cluster information#

The kubectl cluster-info command displays the information about the cluster that kubectl currently connected to.

kubectl cluster-info

Kubernetes control plane is running at https://192.168.49.2:8443
CoreDNS is running at https://192.168.49.2:8443/api/v1/namespaces/kube-system/services/kube-dns:dns/proxy

To further debug and diagnose cluster problems, use 'kubectl cluster-info dump'.

Proxy#

Since kubectl is just convenient wrapper around the Kubernetes API, it can communicate with the kubernetes API and allow you to create a locally hosted proxy for it. Use the kubectl proxy command for this.

The following cell runs the proxy.

kubectl proxy &
export proxy_pid=$!

[1] 101050

Now you are able to access api:

curl localhost:8001/api

Starting to serve on 127.0.0.1:8001
{
  "kind": "APIVersions",
  "versions": [
    "v1"
  ],
  "serverAddressByClientCIDRs": [
    {
      "clientCIDR": "0.0.0.0/0",
      "serverAddress": "192.168.49.2:8443"
    }
  ]
}

kill $proxy_pid

Display resouces#

The kubectl get command displays given resources. It have subcommands for all resources types. The most common are:

Command	Resource Kind	Description
`kubectl get pods`	Pod	The smallest deployable unit — runs one or more containers.
`kubectl get deployments`	Deployment	Manages Pods and ReplicaSets; handles rolling updates.
`kubectl get services`	Service	Exposes a set of Pods on a network endpoint (ClusterIP, NodePort, etc.).
`kubectl get replicasets`	ReplicaSet	Ensures a specified number of Pods are running.
`kubectl get daemonsets`	DaemonSet	Runs one Pod on each (or selected) node.
`kubectl get statefulsets`	StatefulSet	Manages Pods with stable identities (useful for databases).
`kubectl get jobs`	Job	Runs Pods to completion (batch jobs).
`kubectl get cronjobs`	CronJob	Schedules Jobs to run periodically.
`kubectl get nodes`	Node	Shows the cluster’s worker nodes.
`kubectl get namespaces`	Namespace	Lists logical partitions of the cluster.

Output format#

The important thing is how you specify the output format. Use the -o option followed by the instructions on how to display the output.

There following types of instructions:

Format	Description	Example
`wide`	Adds more columns (e.g., node name, IP).	`kubectl get pods -o wide`
`name`	Prints only resource names.	`kubectl get pods -o name` → `pod/nginx-1234`
`yaml`	Full YAML definition (like what’s stored in the API).	`kubectl get pod nginx -o yaml`
`json`	Same as above but in JSON.	`kubectl get pod nginx -o json`
`jsonpath`	Extracts specific fields using JSONPath syntax.	`kubectl get pods -o jsonpath='{.items[*].metadata.name}'`
`jsonpath-as-json`	Like `jsonpath`, but ensures valid JSON output.	`kubectl get pods -o jsonpath-as-json='{.items[*].metadata.name}'`
`go-template`	Uses Go templates to define custom output.	`kubectl get pods -o go-template='{{.items[0].metadata.name}}'`
`custom-columns`	Displays only selected fields as table columns.	`kubectl get pods -o custom-columns=NAME:.metadata.name,STATUS:.status.phase`

For example, consider the output of the raw kubectl get pods command and the configuration option that shows additional information.

kubectl create deployment temp --image=nginx:1.22.1 --replicas=3 &> /dev/null

The following cell represents regular output:

kubectl get pods

NAME                    READY   STATUS    RESTARTS   AGE
temp-5dcd5945c4-255xn   1/1     Running   0          17s
temp-5dcd5945c4-rfvt7   1/1     Running   0          17s
temp-5dcd5945c4-rrm9p   1/1     Running   0          17s

The wide output adds some additional columns to the output table.

kubectl get pods -o wide

NAME                    READY   STATUS    RESTARTS   AGE   IP           NODE         NOMINATED NODE   READINESS GATES
temp-5dcd5945c4-255xn   1/1     Running   0          34s   10.244.0.4   daemon       <none>           <none>
temp-5dcd5945c4-rfvt7   1/1     Running   0          34s   10.244.1.3   daemon-m02   <none>           <none>
temp-5dcd5945c4-rrm9p   1/1     Running   0          34s   10.244.2.3   daemon-m03   <none>           <none>

kubectl delete deployment temp &> /dev/null

Management types#

The k8s supports two management types:

The imperative: operates directly on live objects in a cluster.
The declarative: operates through configuration files that declare the objects.

Check more in the k8s Object Management page of the official documentation.