Quick start

1. Start it

   kubeglass

   KubeGlass reads ~/.kube/config (or $KUBECONFIG), prints its address and the contexts it found, and opens localhost:8090 in your browser. It starts on the current context; every other context in the file is available too.

   With the default local auth there is no sign-in: KubeGlass acts with your kubeconfig credentials. Press Ctrl+C to stop it.

2. Run it again any time

   Running kubeglass while it is already running opens the running one. If port 8090 is taken by another program, KubeGlass uses the next free port and prints it. A port you chose, with --port, KUBEGLASS_PORT or port in kubeglass.yaml, stays: KubeGlass stops with an error when it is taken.

   Like k9s, it can open a list or a cluster straight away: kubeglass deploy -n shop, kubeglass --context prod pods. See Command line.

3. Look at the Overview

   The Overview lists what needs attention, errors first and then the newest: crash-looping or unschedulable pods grouped by workload with the reason in plain words, nodes that aren’t ready, stalled rollouts and failing jobs. Each entry links to the logs or events that explain it, and some have a fix, such as reverting the change made just before the problem or giving a container more memory.

4. Open a list

   Open Resources › Workloads › Pods in the sidebar, or press : and type po. Move with j and k, press Enter to open a pod, l for its logs or s for a shell. Press / to filter, for example by app=web. On a wide screen, clicking a row opens the pod beside the list, and the pane follows as you move.

5. Search with the command palette

   Press ⌘K (Ctrl+K on Linux and Windows) and type an object’s name, a kind or a page. :deploy kube-system opens the Deployments in kube-system, @ switches cluster and # switches namespace.

6. Open a terminal

   Press ⌘` (Ctrl+`) for the dock, the panel at the bottom with shells, logs and YAML in tabs. Open a shell in a pod from its Shell tab or the dock. To get a shell on your own computer for kubectl and helm, start KubeGlass with kubeglass --local-shell; it opens on the cluster you’re viewing.

Switch clusters

Open Clusters to see every context in your kubeconfig with its problem count, and press Enter on one to switch to it. The Overview lists the other clusters too, and the cluster switcher in the top bar and @ in the command palette do the same. In a shared deployment each person’s selection is their own; switching doesn’t affect anyone else.

Add a cluster

Open Clusters › Add cluster and pick how you reach the cluster: a kubeconfig, your cloud’s CLI (EKS, GKE or AKS), an API server and token, several kubeconfigs or a list of clusters, a folder of kubeconfigs, single sign-on, or installing KubeGlass in the cluster with Helm. KubeGlass merges new entries into your kubeconfig and leaves the existing ones alone. See Add cluster.

No cluster?

kubeglass --demo

Demo mode runs on sample data and changes nothing. The online demo is the same UI.

Where things are

Every page is also reachable from the command palette. A few to start with:

To	Open
See what is broken and why	Overview, then the object’s page
Find out why one workload fails	Its page’s Overview tab, which runs Checks
See what changed and who changed it	Timeline
Undo a change	Timeline, open the change, then Revert change
Find pods that keep restarting	Monitoring › Restarts
Check the cluster against best practices	Security › Checks
See who can do what	Security › Access
Manage Helm releases	Applications › Helm releases
Check Argo CD and Flux	Applications › GitOps
Compare clusters	Clusters › Compare
Prepare a Kubernetes upgrade	Clusters › Upgrade check
Size requests and limits	Monitoring › Rightsizing
Choose the theme, notifications and terminal font	Settings

Next

Overview and resources

Terminal

Keyboard shortcuts

Configuration