Looks like a mistake. What are the limitations?
Logically, if all lines in MASTERSOURCE_FUNCTION had a NON-OSDA function, then this would not exclude any lines (or if none of them had this value, then all lines would be excluded).
Going further, if each line in MASTERSOURCE has one or zero NON-OSDA line in MASTERSOURCE_FUNCTION, then it should be an exception candidate. But between identifiers MASTERSOURCE and NAME should also be one-to-one.
I would pull the ROWID from ACCOUNTSOURCE for 48 rows, then track the MASTERSOURCE and NAME identifier and see on what basis these rows are duplicated or not excluded. That is, there are 12 duplicate names in MASTERSOURCE, where it is expected to be unique using the NOVALIDATE constraint.
source share