Your first incident

Ten minutes with your own cluster — break a deployment on purpose and watch Moonway catch it, explain it, and record it.

Screenshots of somebody else's outage prove nothing. The honest demo is your own cluster and a deliberate failure, so this quickstart breaks something, watches Moonway react, and puts it back.

Use a cluster you're allowed to break

A kind or minikube cluster is perfect. The failure below is a bad image tag on one deployment — reversible with a single rollback — but do it somewhere consequence-free.

Pick a context

Open Moonway. The sidebar lists every context from your kubeconfig; pick one and the daemon opens a watch stream. Within seconds the initial sync completes and the Overview fills — nodes, workloads, recent warnings. From here on, everything you see is computed from a live in-memory cache, not from polling your API server.

Break a deployment

Point a deployment at an image tag that doesn't exist (swap web for any deployment you have):

kubectl set image deploy/web app=web:no-such-tag

Watch Triage catch it

Within moments, Triage lists the incident: an image-pull failure on the new pods, and a rollout-stuck warning on the deployment as its progress deadline approaches. Each item carries what's wrong, since when, and how many pods are caught in it. A desktop notification fires if you've turned those on.

See it in the journal

Open What changed. The image change is already recorded — old tag, new tag, timestamp — followed by the crash-and-pull entries it caused. This is the page that answers next week's question: what changed right before things went wrong?

Inspect the pod

Click through to the failing pod. The inspector opens with Events telling you exactly why the pull failed, Logs streaming from the still-healthy containers, and Related walking the ownership chain — pod to ReplicaSet to Deployment to Service.

Put it back

kubectl rollout undo deploy/web

Triage clears on its own — items resolve when the condition disappears, not when you dismiss them — and the journal keeps both the break and the fix.

Where to next