The current implementation makes two queries: one to search for a chunk that would be broken by the lower boundary of the range (min), and another for a chunk that would be broken by the upper boundary (max). Since chunks cannot overlap, if follows that there can be only one chunk that could be broken by either boundary, so both queries can be written to return only one potential match. Doing that would prevent unnecessarily calling getMore() by the query, which by itself can cause problems (see BF-46040).