Why Kubernetes EndpointSlice Delays Break Service Discovery

Episode notes

In episode 52 of DevOps Daily with Fexingo, Lucas and Luna dig into a hidden performance gotcha in Kubernetes service discovery: the EndpointSlice controller's default sync interval. Lucas walks through a real incident at a mid-size e-commerce company where a deployment rollout caused five minutes of traffic blackholing because EndpointSlices lagged behind pod readiness. They compare EndpointSlice to the older Endpoints API, explain the kube-controller-manager flags that control sync timing, and discuss the trade-offs between faster propagation and API server load. Luna challenges Lucas on whether most teams should even tweak these defaults, and they land on a practical recommendation: monitor your service propagation latency before you tune anything. If you manage Kubernetes at scale and have ever seen 'no endpoints available' for longer than you expected, this episode explains why.