Getting started
v8s reads your kubeconfig, watches the clusters you select, and shows them in one UI. This page takes you from nothing to a failing pod's logs in about five minutes.
The five-minute path
- Install the desktop app. Download the installer for your OS from the releases page and run it. Details and checksums are in Install.
- Open v8s. It starts its own agent in the background. Nothing is connected yet, on purpose.
- Pick clusters. Open the cluster selector in the top bar. v8s lists every context from your kubeconfig. Tick the ones you want, or click a group chip, then press Connect. See Clusters.
- Find a problem. Press
:and typepb, then Enter. The Problems view lists what is broken across every connected cluster, worst first. - Open it. Press Enter on a problem to expand its evidence and quick actions, then
o(or the open button) to open the object in the detail drawer. The drawer shows status, events, and YAML. Use the Logs tab (or presslon the row) to read the pod's output, and the Shell tab (ors) to get a terminal in it. See Operations.
You are not locked into one cluster
Everything you just did spanned every cluster you selected. Add another with :ctx +staging, drop one with :ctx -staging. See Navigation.
Prerequisites
- A kubeconfig with at least one context. v8s uses the same loading rules as
kubectl:$KUBECONFIG(several files are merged) or~/.kube/config. - Whatever your contexts already need to authenticate: client certificates, tokens, or an exec plugin such as
kubelogin,aws, orgke-gcloud-auth-plugin. Ifkubectl --context <name> get nsworks in a terminal, it works in v8s. - Kubernetes RBAC that lets you list and watch what you want to see. v8s disables actions you are not allowed to take and tells you why.
Run it in a browser instead: v8s serve
If you prefer a browser tab, or you are on a Linux box without a desktop shell, use the CLI.
- Download the CLI archive for your platform from the releases page and unpack it.
- Run it:
v8s servev8s listens on 127.0.0.1:7931, prints a URL, and opens your browser. The URL carries a one-time token in its fragment, so only that browser tab can talk to the agent. To pick a different port or skip opening the browser:
v8s serve --port 8080 --no-browserTo use a specific kubeconfig file, add --kubeconfig /path/to/config.
On a remote machine, bind stays on loopback. Forward the port over SSH and open the printed URL locally:
ssh -L 7931:127.0.0.1:7931 devboxNot for sharing
v8s serve runs as you, with your credentials, and has no login of its own. Do not expose its port to other people. For a team, use hosted mode.
Try it without a cluster: demo mode
The agent can serve synthetic clusters so you can explore the UI, or check how it feels at scale, without touching a real cluster. The fake clusters have names like prod-us, prod-eu, and staging, realistic pods, a stream of churn, and objects you can edit and act on.
v8s agent --fake clusters=3,pods=50000,churn=200 \
--listen 127.0.0.1:7931 --dev-token demoThen open:
http://localhost:7931/#agent=ws://localhost:7931/ws&token=demo| Option | Meaning | Default |
|---|---|---|
clusters=N | Number of fake clusters | 3 |
pods=N | Total pods across the clusters | 50000 |
churn=N | Pod events per second across the clusters | 200 |
lograte=N | Synthetic log lines per second per pod | 20 |
rbac=1 | Makes the second cluster read-only and adds a linked-able admin context | off |
Fake mode keeps its settings in memory unless you pass --settings <file>. Use rbac=1 to try linked read and write contexts.
Next steps
- Learn the command bar and shortcuts.
- Understand clusters, groups, and linked contexts.
- Choose which problem classes are on.
- See everything the CLI can do in the CLI reference.
- If something does not connect, see Troubleshooting.