We occasionally have issues during maintenance windows that a large number of pods are being restarted. This causes the apiserver to get overloaded with requests and it begins to become very slow at responding. One issue that exasperates this slowness is etcd-operator rescheduling an etcd cluster that is still coming up. Is this startup timeout configurable anywhere? It would help our situation to be able to increase the timeout, it appears to be around 30 seconds today.
- Drew
