Thanks all for the confirmation, it's just as I suspected.
I would have loved this as a feature request, because logically the guaranteed restore point requires space and would halt operations to do so, whereas the normal restore point is a 'nice to have':
- save pre-upgrade database state with a guaranteed restore point pre_deploy
- do "X" transaction (upgrade, deploy, application release) successfully
- convert pre_deploy from guaranteed to normal restore
the pre_deploy save point becomes non-critical but a good thing to have in case we want to flashback to prior snapshots of specific tables or something as time goes on. As the flashback points age-off, I wouldn't want to hold the instance hostage just to keep an old flashback point which we technically wouldn't need, but may be.
My workaround solution will likely be to create TWO flashback points:
CREATE RESTORE POINT pre_deploy_guaranteed guarantee flashback database;
CREATE RESTORE POINT pre_deploy;
After the success, I delete the guaranteed one:
drop restore point pre_deploy_guaranteed;