During insertion of clustered index, InnoDB does the check for foreign key
constraints. Problem is that it uses the clustered index entry to search
indexes of referenced tables and it could lead to unexpected result
when there is no foreign index.

Solution:
========
  Rebuild the tuple based on foreign column names before searching it
on reference index when there is no foreign index.