Browsing resources¶

Pick a kind from the nav drawer and Kubus shows you a live list of every resource of that kind, across all your selected clusters. Lists stay in sync over a WebSocket watch. There's no refresh button because there's nothing to refresh.

The Pods list — Live, sortable and filterable, with the right columns for each kind.

Live updates¶

Every list is backed by an informer-style watch. New objects appear, status changes ripple through and deletions drop out instantly, without flicker. If a watch is interrupted (the classic Kubernetes 410 Gone), Kubus reconnects and resyncs on its own, so you can leave a list open all day and trust what it shows.

Columns that fit the kind¶

Each built-in kind has hand-picked columns. Pods show readiness, status, restarts, CPU, memory, node and age; Deployments show ready/up-to-date/available; Services show type and ports. When several clusters are selected, a Cluster column is added automatically.

Sort by clicking a column header.
Filter with the search box. Plain text works, or start with / for smart filters: structured clauses like /status:crash ns:prod restarts>3, with autocomplete.
Labels get their own column. Each row shows its first labels as chips with a +N overflow; hover to see them all, or click a chip to filter by that label.
Secret values are redacted by default. Kubus never shows secret data in a list. Reveal them deliberately in the details drawer.
Warnings find you. A small amber marker next to the name flags any object with warning events or container restarts in the last hour; hover it for the reasons. Sort by that column and the noisiest objects come first. The same signal puts a count on the drawer's Events tab and a dot on a page tab whose object turned unhealthy while you were looking elsewhere.

Kubus remembers where you were¶

Nothing resets behind your back. Each kind remembers the filter, label selector and scroll position you left it with, and opening the kind again from the nav brings them back. Clearing the filter forgets it, and the empty state offers a Clear filters button when a filter hides everything. Saved views and deep links carry their own query and are never overridden.

The namespace filter is remembered per cluster too, so switching to the prod cluster lands on the namespace you use there.

Custom resources, first-class¶

CRDs aren't an afterthought. Kubus discovers every CustomResourceDefinition in your selected clusters and lists them under Custom Resources, grouped by API group. It also renders each CRD's own additionalPrinterColumns (the same extra columns you get from kubectl get) as real, sortable columns.

A custom resource list with printer columns — Your operators' CRDs, with their printer columns rendered as columns.

Pick the right version automatically

When a CRD serves multiple versions, Kubus prefers the most stable, newest one (v1 over v1beta1 over v1alpha1) so you land on the version you almost certainly want.

Filtering by namespace¶

The namespace filter in the top bar narrows every list. Leave it empty for all namespaces; it applies across all selected clusters at once.

Saved views¶

If there's a list you keep coming back to, such as failing pods in prod or all Ingresses in team-a, save it. The current kind plus its namespace and cluster filters become a saved view that appears in the nav drawer right under its kind, one click away. Delete a saved view from the same spot when you're done with it.

Acting on a row¶

Every row has a ⋮ menu with the actions that make sense for that kind: logs, shell, scale, restart, port-forward, delete and more. That's covered in Quick actions. To inspect instead, click the resource's name to open the details drawer.