In this lesson, you will learn how to bridge the gap between raw Prometheus metrics and actionable insights by connecting Grafana to your cluster’s telemetry data.
While Prometheus is the engine that collects and stores your time-series data, Grafana acts as the "control room" where you actually see what is happening. By connecting these two, you turn abstract numbers—like CPU percentages or request latency—into visual, interactive dashboards that make cluster health immediately obvious.
Configuring the Data Source
To start, you must point Grafana to your Prometheus instance. In a Kubernetes environment, if both are running in the same cluster, you can use the internal service URL (e.g., http://prometheus-server.monitoring.svc.cluster.local:80).
Within the Grafana UI, navigate to Connections > Data Sources > Add data source and select Prometheus. Once you input the URL, Grafana verifies the connection, allowing you to use the Prometheus Query Language (PromQL) directly in your dashboard panels.
Building Your First Panel
Once connected, you can create a dashboard and add a "Panel." This is where you write your PromQL queries. For example, to visualize the average CPU usage across your containers, you would enter a query like this:
sum(rate(container_cpu_usage_seconds_total[5m])) by (pod)
Grafana then executes this query against your Prometheus data source in real-time. Because it is interactive, you can:
- Change Time Ranges: Zoom into a specific spike during a deployment.
- Variable Filtering: Use "Variables" (or Template variables) to create dropdown menus, allowing you to switch between different Namespaces or Nodes without editing the queries manually.
- Alerting: Configure thresholds directly on the panel so that if a metric crosses a critical line, Grafana sends an alert to your notification channel.
By visualizing your metrics, you transition from "reactive troubleshooting" (logging into nodes to check top) to "proactive observability" (monitoring the health of your entire stack from a single glass pane).
Key Takeaway: Grafana transforms your Prometheus data into interactive, human-readable dashboards, enabling you to detect performance bottlenecks and cluster anomalies at a glance.