Skip to content

Introduction

KubeMeridian is a Grafana app plugin that turns Grafana into a turnkey Kubernetes observability product. You install one plugin, link the observability backends you already run, and get a coherent set of views across the four pillars:

PillarSourceWhat you get
Inventory / topologyKubernetes API (read-only)Cluster status, namespace → workload → pod maps, node health, events
MetricsPrometheus / Mimir / ThanosCluster, namespace, workload, storage, control-plane and networking dashboards
LogsLokiInline pod logs (click a pod) and a logs explorer
TracesTempoTraceQL search and the service graph

On top of that it adds cost & right-sizing, RED & SLOs for any ingress/mesh stack, and a curated Prometheus alert pack.

How it's built

  • App plugin (devopstech-kubemeridian-app) — the pages, dashboards and config.
  • Bundled datasource (devopstech-kubemeridian-datasource) — a thin, frontend proxy to the Kubernetes API. It injects a read-only ServiceAccount bearer token server-side (the token never reaches the browser) and only ever issues GET requests.
  • No backend binary, Apache-2.0, works on Grafana ≥ 12.3.

Each "cluster" you add is one datasource instance. A cluster links a metrics, logs and traces datasource plus its cost and traffic settings — see Connect your stack.

What it is not

  • It does not collect data — it reads the Kubernetes API and queries the datasources you already operate (kube-prometheus-stack, Loki, Tempo, …).
  • It does not write to your cluster — everything is read-only.

Next: Install.

Apache-2.0 licensed.