Open
Description
Is your feature request related to a problem? Please describe.
Currently, admiral syncs all the k8s resources using controllers for different clusters. This takes time depending on the size of the clusters etc. During this time no updates are made to the resources. As a result there is a time window when a subsequent admiral version is deployed when no updates are made to the endpoints if resources like GTPs are edited during this time.
Describe the solution you'd like
Currently admiral has readiness probe, if we can fail the readiness probe until after the caches are refreshed then that would result in previous version of admiral still running until this instance comes up.