Is there any plan to improve the create / update / recreate / destroy reasoning for azure-native by the way? I keep running into issues forcing me to destroy resources before updating them, for instance:

azure-native:compute:VirtualMachineScaleSet (xxxxx-scaleset-linuxagents):
    error: autorest/azure: Service returned an error. Status=<nil> Code="PropertyChangeNotAllowed" Message="Changing property 'osDisk.managedDisk' is not allowed." Target="osDisk.managedDisk"

In this case, pulumi shouldn't propose to

update

but suggest to

recreate

right?

this is a long running problem whereby upstream do not indicate that a recreate is required so the provider cannot provide the right feedback see this issue Manual overrides are managed here and I guess us community should submit a PR 🙂