k9s

k9s is a terminal-based UI that provides a curses-style interface to interact with your Kubernetes clusters. It gives you a real-time view of your cluster, with keyboard-driven navigation and built-in actions for the most common kubectl operations.

Installation

Homebrew (macOS/Linux)

brew install derailed/k9s/k9s

# Verify
k9s version

Package Managers

# macOS via MacPorts
sudo port install k9s

# Windows via Scoop
scoop install k9s

# Windows via Chocolatey
choco install k9s

# Arch Linux
pacman -S k9s

# Nix
nix-env -iA nixpkgs.k9s

Manual Installation

# Linux — download latest release
VERSION=$(curl -s https://api.github.com/repos/derailed/k9s/releases/latest | jq -r .tag_name)
curl -L "https://github.com/derailed/k9s/releases/download/${VERSION}/k9s_Linux_amd64.tar.gz" | \
  tar xz -C /usr/local/bin k9s

# macOS (arm64)
curl -L "https://github.com/derailed/k9s/releases/download/${VERSION}/k9s_Darwin_arm64.tar.gz" | \
  tar xz -C /usr/local/bin k9s

Run with Docker

docker run --rm -it \
  -v ~/.kube/config:/root/.kube/config \
  quay.io/derailed/k9s

Configuration

Launch Options

# Start k9s (uses current kubeconfig context)
k9s

# Start in a specific namespace
k9s -n production

# Start with a specific context
k9s --context staging

# Start focused on a specific resource type
k9s --command pods

# Read-only mode — prevents destructive actions
k9s --readonly

# Logging verbosity
k9s --logLevel debug

Config File

# ~/.config/k9s/config.yaml
k9s:
  liveViewAutoScroll: true         # Auto-scroll log view
  maxConnRetry: 5
  readOnly: false
  noExitOnCtrlC: false
  ui:
    skin: monokai                  # Skin name (see ~/.config/k9s/skins/)
    enableMouse: true
    headless: false
    logoless: false
    crumbsless: false
    reactive: false
    noIcons: false
  refreshRate: 2                   # Seconds between refreshes
  skipLatestRevCheck: false
  disablePodCounting: false
  logger:
    tail: 100
    buffer: 5000
    sinceSeconds: -1               # -1 = all logs
    textWrap: false
    showTime: false
  thresholds:
    cpu:
      critical: 90
      warn: 70
    memory:
      critical: 90
      warn: 70

Custom Aliases

# ~/.config/k9s/aliases.yaml
aliases:
  dp: deployments
  svc: services
  ns: namespaces
  no: nodes
  ing: ingresses
  cm: configmaps
  sec: secrets
  pvc: persistentvolumeclaims
  po: pods
  rs: replicasets
  sts: statefulsets
  ds: daemonsets
  cj: cronjobs
  hpa: horizontalpodautoscalers

Core Commands

Global Keyboard Shortcuts

Shortcut	Description
:	Open command mode to enter resource type
/	Filter resources by name or label
?	Show help / all available shortcuts
Ctrl+a	Show all available resource aliases
Ctrl+d	Delete selected resource
Ctrl+k	Kill (force delete) selected resource
Ctrl+e	Edit resource in your $EDITOR
Ctrl+s	Save current view to a file
Ctrl+u	Clear the current filter
Ctrl+w	Toggle wide output columns
Ctrl+r	Refresh current view
Ctrl+f	Toggle fullscreen
Ctrl+c	Exit k9s
q	Quit current view / go back
Esc	Clear filter or go back
g	Jump to top of list
G	Jump to bottom of list
h or ←	Go back in breadcrumb
Enter	Drill into resource
d	Describe selected resource
e	Edit selected resource YAML
l	View logs for selected pod
s	Shell into selected pod
f	Port-forward selected pod/service
y	Show resource YAML
u	User impersonation

Resource Navigation Commands

Command	Description
:pods or :po	Navigate to Pods view
:deployments or :dp	Navigate to Deployments
:services or :svc	Navigate to Services
:namespaces or :ns	Navigate to Namespaces
:nodes or :no	Navigate to Nodes
:ingresses or :ing	Navigate to Ingresses
:configmaps or :cm	Navigate to ConfigMaps
:secrets or :sec	Navigate to Secrets
:pvcs	Navigate to PersistentVolumeClaims
:events or :ev	Navigate to Events
:contexts or :ctx	Switch kubeconfig context
:all	Show all resource types
:namespaced-resources	List namespaced resource types
:cluster-resources	List cluster-scoped resource types
:crds	Navigate to CustomResourceDefinitions
:helm	Navigate to Helm releases
:popeye or :pop	Run Popeye linter
:xrays	Show X-ray tree view
:pulse	Show cluster pulse dashboard
:benchmarks	Run HTTP benchmarks

Log Viewing Shortcuts

Shortcut	Description
l	Open logs for selected pod
0	Show logs for all containers
1-9	Show logs for container N
t	Toggle timestamps in logs
w	Toggle log line wrapping
s	Toggle auto-scroll
/	Filter log output
Ctrl+s	Save logs to file
f	Full-screen log view

Advanced Usage

Plugins

# ~/.config/k9s/plugins.yaml — add custom menu actions
plugins:
  # kubectl neat — clean up managed fields from YAML
  neat:
    shortCut: Ctrl-N
    description: Neat dump
    scopes:
      - all
    command: bash
    background: false
    args:
      - -c
      - kubectl neat get --context $CONTEXT -n $NAMESPACE $RESOURCE_NAME $NAME -o yaml | less

  # Get pod's resource requests/limits
  get-all:
    shortCut: g
    confirm: false
    description: get all
    scopes:
      - namespaces
    command: kubectl
    background: false
    args:
      - get
      - all
      - -n
      - $NAME
      - --context
      - $CONTEXT

Custom Skins

# ~/.config/k9s/skins/monokai.yaml
k9s:
  body:
    fgColor: "#f8f8f2"
    bgColor: "#272822"
    logoColor: "#f92672"
  prompt:
    fgColor: "#f8f8f2"
    bgColor: "#272822"
    suggestColor: "#66d9ef"
  info:
    fgColor: "#66d9ef"
    sectionColor: "#a6e22e"
  frame:
    border:
      fgColor: "#75715e"
      focusColor: "#f92672"
    menu:
      fgColor: "#f8f8f2"
      keyColor: "#f92672"
    crumbs:
      fgColor: "#f8f8f2"
      bgColor: "#75715e"
      activeColor: "#f92672"
    status:
      newColor: "#a6e22e"
      modifyColor: "#66d9ef"
      addColor: "#a6e22e"
      errorColor: "#f92672"
      highlightColor: "#e6db74"
      killColor: "#f92672"
      completedColor: "#75715e"

Benchmarks (HTTP Load Test)

# Port-forward a service, then run a benchmark from within k9s:
# 1. Navigate to :services
# 2. Select a service and press b to benchmark
# Benchmark results appear in :benchmarks

X-Ray View

# Navigate to :xrays for a hierarchical view
# Shows: Deployment -> ReplicaSet -> Pod -> Container relationships
# Useful for diagnosing ownership chains

# In xray view:
# x        — toggle X-ray
# Enter    — drill into resource

Common Workflows

Stream Logs from Multiple Pods

# 1. Navigate to :pods
# 2. Filter by label: /app=backend
# 3. Press l on any pod to view logs
# Shift+l to view logs for all selected pods simultaneously

Port-Forward and Test

# 1. Navigate to :services or :pods
# 2. Select the resource
# 3. Press f to start port-forward
# 4. k9s shows the local port in the footer
# 5. Test in another terminal: curl localhost:<local-port>/health

Exec into a Container

# 1. Navigate to :pods
# 2. Select a pod
# 3. Press s (shell) — defaults to /bin/sh
# 4. For multi-container pods: press s and select container

# Pro tip: if /bin/sh is unavailable, set K9S_EXEC_SHELL=bash
export K9S_EXEC_SHELL=/bin/bash

Rolling Restart a Deployment

# 1. Navigate to :deployments
# 2. Select the deployment
# 3. Press r to trigger a rolling restart
# 4. Watch the Pods view update in real time

Investigate a CrashLoopBackOff

# 1. Navigate to :pods
# 2. Find the pod in CrashLoopBackOff (red row)
# 3. Press d to describe — check Events section
# 4. Press l to view logs (previous container logs with p)
# 5. Press y to inspect the YAML spec for resource limits

Tips and Best Practices

• Use :ctx to switch contexts without leaving k9s — no need to run kubectl config use-context in a separate terminal.
• Filter with / accepts label selectors like app=frontend,env=prod as well as plain name substrings.
• Press ? on any view — keybindings are context-sensitive; the help overlay shows only what’s available for the current resource type.
• Use --readonly in production — prevents accidental deletions when you just need to inspect a cluster.
• Customize aliases.yaml — short aliases like :dp for deployments speed up navigation significantly.
• Popeye integration (:popeye) runs a linter against your cluster config, surfacing misconfigurations and resource waste.
• The pulse view (:pulse) shows a live dashboard of pod counts, CPU/memory, and event rates — good for a quick health check.
• Install the neat plugin — it strips managedFields from YAML output, making resource specs much easier to read.
• Ctrl+w toggles wide output — reveals additional columns like node affinity, QoS class, and nominated node.
• Set liveViewAutoScroll: true in config — log views will tail automatically without pressing s each time.