Enhance your Kubernetes experience with custom kubectl plugins. Streamlining, automating and simplifying everyday tasks for efficient cluster management.

_custom-kubectl-subcommands.md

Custom Kubectl Subcommands

With the introduction of custom subcommands in Kubernetes version 1.12, the kubectl command line tool became even more powerful for DevOps engineers. These plugins extend kubectl’s functionality, allowing you to streamline repetitive tasks, partially automate processes, and improve your day-to-day workflow as a Kubernetes operator.

To create or install a plugin for kubectl, simply place an executable or script in your system’s PATH or a directory specified by the $PATH environment variable. For example, kubectl-all installed in /usr/local/bin, can effortlessly invoked as a subcommand: kubectl all. No additional configuration or registration is necessary. Detailed instructions can be found in the official documentation.

An ecosystem has already developed around the Krew plugin manager, and there are many great plugins available, but here are some neat subcommands that I use frequently myself for day-to-day troubleshooting, monitoring, and operational tasks.

kubectl-all

#!/usr/bin/env bash
# Get all Pods across all namespaces, including status and node name.

kubectl get pod -o=custom-columns=NAME:.metadata.name,STATUS:.status.phase,NODE:.spec.nodeName --all-namespaces

kubectl-kill

#!/usr/bin/env bash
# Kills a pod forcibly without waiting.
set -eo pipefail

if [[ -z "${1// /}" ]]
then
    echo >&2 "error: resource(s) were provided, but no name was specified"
    exit 1
fi

kubectl delete pod "${1}" --grace-period=0 --force

kubectl-readyornot

#!/usr/bin/env bash
# Lists deployments which have less than desired pods ready/available.
set -eo pipefail

if ! command -v jq &>/dev/null
then
    echo >&2 "error: unable to find 'jq', please install it to use this plugin."
    exit 1
fi

kubectl get deployments -o json | jq -r '.items[] | select(.status.unavailableReplicas > 0) | .metadata.name'

kubectl-shell

#!/usr/bin/env bash
# Start a one-off containers for inspection and debugging, image and command is configurable.
set -eo pipefail

image=${1:-alpine}
command=${2:-sh}
kubectl run -it --rm --restart=Never "shell-$(date +%s)" --image="${image}" -- ${command}