
Give things time to start and probe only as needed. This avoids premature health warnings such as: Normal Created 14s kubelet, node-0 Created container Normal Started 14s kubelet, node-0 Started container Warning Unhealthy 10s kubelet, node-0 Readiness probe failed: Change-Id: I224aed14a03a3690f1ba2a07a2d9fad54b7dd550