Why Prometheus couldn’t see Cilium metrics at 2 a.m.

A production Kubernetes team found blank Grafana panels because Prometheus was not scraping Cilium agent and operator pods.

The issue was not broken software, but a missing ServiceMonitor integration between otherwise healthy CNCF projects.

The example highlights a broader cloud-native risk: outages and blind spots often come from poor platform wiring, not component failure.

It underscores the operational cost of day-2 integration work across tools like Prometheus, Cilium, Grafana, and Hubble.