If you have to engineer a solution due to problems caused by the failures of those you depended on to deliver the expected level of service then the entire solution is over-engineered.
Over-engineering is when you make something to be more robust than is necessary for it to work successfully. Evidently depending on the registrar actually failed here (just as it has for many in the past), and this will surely happen again in the future. I don't see how making your system robust to failures that actually come up is over-engineering. It sounds more like just plain old engineering. (And I honestly also fail to see what is productive about pressing on with this conversation.)