Potential Edge Cases Preventing GUID Deallocation? #201
Description
Hi I have a question about a potential issue that could arise with GUID management. We are reliant on Kubernetes watchers to deallocate a GUID PKey association in UFM. Is there risk in relying on the Kubernetes API server to be available to successfully deallocate GUID in all scenarios? For example if the API server is down the watcher might not get the delete events for the Pod therefore the GUIDs attached to that pod might forever live in UFM until manually deleted. I can imagine a couple ways to solve this but I am just confirming that this is a potential problem.
Thank you!