When a Kubernetes LoadBalancer VIP suddenly became unreachable, the team traced the issue to Cilium's L2 announcement mechanism. This case study details the investigation: checking Cilium agent logs, analyzing network traffic with tcpdump, and reviewing the L2 announcement configuration. The root cause was a misconfiguration in the L2 announcement policy that caused conflicting ARP responses. The fix involved adjusting the announcement scope and verifying with end-to-end connectivity tests. For DevOps teams running Cilium, this is a practical lesson in understanding L2 announcement behavior and its impact on service availability.
A real-world debugging story of how Cilium's L2 announcement feature led to LoadBalancer VIP unavailability, with step-by-step root cause analysis.