Maybe this is the desired behavior, however: Currently, if reconciling fails, the neonvm-controller does not update the VM status, even if it has changed.
This means that it's possible for more up-to-date status information to be arbitrarily delayed due to reconcile failures (either due to transient issues or bugs).
I think this is probably not desirable as it impacts liveness, and as we move towards greater dependence on what's reported by the controller, we may want to tighten this up.
Feature idea(s) / DoD
Don't skip status update if there was some earlier failure.
Problem description / Motivation
Maybe this is the desired behavior, however: Currently, if reconciling fails, the neonvm-controller does not update the VM status, even if it has changed.
This means that it's possible for more up-to-date status information to be arbitrarily delayed due to reconcile failures (either due to transient issues or bugs).
I think this is probably not desirable as it impacts liveness, and as we move towards greater dependence on what's reported by the controller, we may want to tighten this up.
Feature idea(s) / DoD
Don't skip status update if there was some earlier failure.