For downstream clusters, we have (at least) 3 different scenarios, and each of them involve different upgrade procedures/mechanisms. AFAIK:
- For EIB provisioned clusters -> Fleet approach
- For Metal3 provisioned clusters -> Redeploy with a fresh image
- For phone-home provisioned clusters -> K8s upgrade via rancher + SUC for the OS and everything else
It would be nice to have a table/matrix/something to explicitly clarify each scenario.
cc @hardys
@ranjinimn is this something you can take care of?