KubernetesEnvoy Proxy

Single Ingress Container Overload During Traffic Spikes

critical
scalingUpdated Jan 31, 2022

Configuring Ingress with wildcard host directs all traffic to one container, overwhelming it during spikes and potentially taking down the entire cluster. Load balancing capabilities are unused, creating single point of failure.

How to detect:

Check Ingress resource definitions for wildcard host configurations. Monitor for single pod handling disproportionate traffic volume. Watch for pod resource exhaustion, increased latency, and gateway timeouts during traffic bursts. Track uneven request distribution across available pods.

Recommended action:

Configure specific Ingress rules instead of wildcard hosts to enable proper load balancing. Distribute traffic across multiple pod replicas using Service resources. Implement HorizontalPodAutoscaler to dynamically scale pods based on traffic. Monitor per-pod request distribution to verify balanced load.