Grafana for Kubernetes-based platforms: every cluster, every region, one dashboard

A platform team doesn’t run one Kubernetes cluster. It runs its customers’ clusters: a fleet of them, spread across regions, each one provisioned and reconciled by an operator on a management cluster. When a customer’s cluster gets stuck, the questions come fast. Which cluster? Which region? Stuck at which stage? And whose is it?

The operator already knows. Every reconcile writes status conditions (InfrastructureReady, ArgoAppReady, Synced) back onto the cluster resource. This talk exports those conditions as Prometheus metrics (one gauge, the condition as a label, min to aggregate) and turns them into a single Grafana dashboard that answers those questions for every customer cluster, in every region.