case expression plus outer join in ole db source

I'm trying to generate the data for a 2-column table, where both columns are defined as NOT NULL and the second column is a uniqueidentifier.

In SQL Server Management Studio, this works fine:

insert into table_3(column_a, column_b)

select table_1.column_a, (case when table_2.column_b is NULL then newid() else table_2.column_b end) as column_b

from table_1 left outer join table_2 on table_1.column_c = table_2.column_c

That is, column_b of the SELECT result has no NULL values, and all 35,986 rows are successfully inserted into a previously empty table_3. (If I comment out the INSERT INTO clause and project table_2.column_b instead of "(case ... end) as column_b", the SELECT result includes 380 rows with a NULL in column_b, so I know the case expression plus the outer join are working as expected.)

But when I use the SELECT query as the SQL command in an OLE DB Source component that is connected directly to the OLE DB Destination for the result table, I get this error:

There was an error with input column "column_b" (445) on input "OLE DB Destination Input" (420

The column status returned was: "The value violated the integrity constraints for the column.".

And sure enough, when I modify the result table to allow NULL in column_b, truncate it, and re-run the data flow, it inserts the exact same 380 rows with a NULL in column_b among the 35,986 rows.

So what is SSIS doing to screw up the results of the SELECT command?

Kevin,

Can you see the values of column_b when you use the 'preview' of the OLE DB source?

Did you check twice the mapping tab in your OLE DB destination to make sure nothing is missing?

Have you used a data view right before the OLE DB Destination to check the values of column_b are shown correctly?

If the answer to those 3 questions is yes, I am affraid I could not help you.

Rafael Salas

|||Could there be an issue in executing the newid() function?|||

Argh, the problem was due to operator error: somehow the source component SQL statement had gotten out of sync with the corresponding variable. How embarrassing!

But many thanks to Rafael and Phil, who got me to examine the data flow task closely enough to discover my error.