I've found that the eventual consistency provided by orchestrators like k8s solve this problem rather well. If the services are written using the right paradigms to handle such a situation, they will also be much more resilient to platform disruptions.

I like to view k8s a lot like erlang's OTP, if something isn't right with the state of a service, I advocate calling 'exit()' and letting the restart with exponential backoff handle the transient.