Navigate a Kubernetes cluster with k9s

Tools used

• 

  k9s

  Terminal UI for managing and observing Kubernetes clusters with real-time monitoring and resource navigation

  34.9kGo

Before you start

• -A working kubeconfig: kubectl get pods succeeds against the cluster
• -kubectl on your PATH (k9s calls it for shells and edits)
• -A terminal with 256-color support

Steps

1. 1Step 1: Install k9s

   Install from a package manager or download a release tarball from GitHub. k9s version confirms the install.

   sh

   # macOS or Linux with Homebrew
   $ brew install derailed/k9s/k9s
   
   # Arch Linux
   $ sudo pacman -S k9s
   
   # Windows
   $ winget install k9s

   sh

   $ k9s version

2. 2Step 2: Start it against the right cluster and namespace

   With no flags, k9s uses the current context from your kubeconfig. --context picks another context, -n a namespace, -A all namespaces, and --kubeconfig a different kubeconfig file. -c sets the first view, for example -c deploy.

   Inside k9s, :ctx lists contexts and :ns lists namespaces; select one and press Enter to switch.

   sh

   $ k9s --context staging -n payments

   sh

   $ k9s --kubeconfig ~/.kube/prod-config -A -c deploy

3. 3Step 3: Switch views with :

   Press : and type a resource name, then Enter. Singular, plural and short names all work: :pod, :deploy, :svc, :ing, :cm, :secret, :node. Add a namespace after the name to scope it, as in :pod kube-system. Ctrl-A lists every resource and alias the cluster offers.

   In a table, Enter drills down (deployment to its pods, pod to its containers), d describes, y shows YAML, and Esc goes back. ? shows all key bindings for the current view.

   text

   :deploy
   :pod kube-system
   :svc

4. 4Step 4: Filter with /

   Press / and type a regular expression to filter the current table by name, for example api|worker. Prefix with ! to invert the match, use -l to filter by labels, and -f for a fuzzy match. Esc clears the filter.

   Since k9s 0.30.0, the command prompt also takes filters directly: :pod /api or :pod app=web,env=prod.

   text

   /api|worker
   /! Running
   /-l app=web
   /-f chkout

5. 5Step 5: Read logs

   Select a pod, container or deployment and press l to stream its logs; p shows the previous container's logs after a restart. In the log view, 0 follows the tail, 2 to 6 show the last 1m, 5m, 15m, 30m or 1h, w toggles line wrap, t toggles timestamps, s toggles autoscroll and f toggles full screen. Press / to filter log lines.

   How many lines k9s loads by default is set under k9s.logger in config.yaml (tail, buffer, sinceSeconds).

6. 6Step 6: Open a shell and port-forward

   Select a pod and press s to open a shell. If the pod has several containers, k9s asks which one. It runs kubectl exec with bash, falling back to sh, so images without a shell (such as distroless images) cannot be entered this way. Exit the shell to return to k9s.

   Shift-F on a pod, container or service opens a port-forward dialog for the local and container ports. :pf lists active forwards, and Ctrl-D there stops one. Port-forwards end when k9s exits.

7. 7Step 7: Use read-only mode on production clusters

   --readonly disables commands that change the cluster: delete, kill, edit, and shell or attach into pods. Viewing, logs and port-forwards still work. To make it the default, set readOnly: true in config.yaml, or in a single context's config file to protect only that cluster. --write overrides the setting for one session.

   sh

   $ k9s --context prod --readonly

   yaml

   # $XDG_CONFIG_HOME/k9s/config.yaml (~/.config/k9s/config.yaml on Linux)
   k9s:
     readOnly: true

8. 8Step 8: Find the config files and set a skin

   Since k9s 0.30.0, the main config is config.yaml (not config.yml) in the k9s config directory: ~/.config/k9s on Linux, ~/Library/Application Support/k9s on macOS and %LOCALAPPDATA%\k9s on Windows, unless XDG_CONFIG_HOME or K9S_CONFIG_DIR points elsewhere. aliases.yaml, plugins.yaml, hotkeys.yaml, views.yaml and the skins/ directory live there too. Per-context settings live under clusters/ in the data directory (~/.local/share/k9s on Linux). k9s info prints the exact paths on your machine.

   To apply a skin, copy a YAML file from the skins directory of the k9s repository into skins/ and set its name, without the extension, in k9s.ui.skin. K9S_SKIN overrides it for one run.

   sh

   $ k9s info

   yaml

   # config.yaml
   k9s:
     ui:
       skin: dracula   # loads skins/dracula.yaml