Failed To Resume Change Stream - Recovery

I’m currently trying to understand to limitation of the source connector thus having read this post:

I wonder if I understood correctly that if we apply the strategy: Temporarily Tolerate Errors
this would mean that we have data loss by skipping what was missing.

and if we apply the strategy: Reset Stored Offsets
it would perform another snapshot before consuming from the oplog.

Thus having a big collection and having the oplog size set too small, we could end up in an endless loop because by the time the snapshot finishes, we’re out of the oplog messages already. Thus we’d have to tune the connector to be faster etc. to not run into this problem - hard to predict.

Is that really what could happen or am I missing something?

Kind regards