What about adding periodic task to conductor service that will iterate over all provisioned instances and check compute instance status (guest status can be defined only directly asking guest or adding new guest API call that will be used by conductor to perform force update for guest statuses) and if compute instance is Stopped/Deleted - mark guest as SHUTDOWN/FAILED/ERROR ?
What about adding periodic task to conductor service that will iterate over all provisioned instances and check compute instance status (guest status can be defined only directly asking guest or adding new guest API call that will be used by conductor to perform force update for guest statuses) and if compute instance is Stopped/Deleted - mark guest as SHUTDOWN/ FAILED/ ERROR ?