Comment by autophagian

Yeah. We had a brief window where everything resolved and worked and now we're running into really mysterious flakey networking issues where pods in our EKS clusters timeout talking to the k8s API.