Coldfusion unexplainable duplicate issues, cannot figure this out!

Question

    So I've been fighting this for days.  I dont have the slightest idea where I can go from here.  I have your standard coldfusion insert queries set up to insert new data into the database.  Something along the lines of:
Select * from table where tableuniqueID = Form.UniqueID  
Cfif databasecheck.recordcount EQ 0, then insert new row with form.uniqueID as the uniqueID.  Well recently, its been throwing the generic Violation of PRIMARY KEY constraint 'PK__OLMS_Dat__9C04DA075F0A3F61'. Cannot insert duplicate key in object 'dbo.table'.
I dont understand that in the slightest, the only way it inserts is if the query right above it is defined and equal to 0 recordcount.
I did some tooling around, and found someone suggesting to insert into my SQL "IF NOT EXISTS" to have SQL do the actual check to see if data exists.  Well we put that in place, and just recently, ANOTHER duplicate happened, same error, cannot insert duplicate showing the new SQL structure.
This is beyond me.  I have two statements checking to see if it exists and, yet, Coldfusion still tries to execute a database insert.  What is going on??  Any ideas?  You would figure this would be the basics.

Answer 1

Hello.  I know it's been awhile but did you figure this out?  I have a very similar thing happening. Something along the lines of
INSERT INTO WorkSheet (CODE_KEY)
SELECT TOP (1) EI.CODE_KEY
FROM EmpInfo AS EI  LEFT OUTER JOIN
WorkSheet AS WS ON WS.CODE_KEY = EI.CODE_KEY
WHERE (EI.CODE_KEY = 'param1') AND (WS.CODE_KEY IS NULL)
where the field "CODE_KEY" is a the primary key in SQL Server 2008.  This is similar to what you're doing except that the check for an existing value happens as part of the insert query.  Just like you every once in awhile I get the "[Macromedia][SQLServer JDBC Driver][SQLServer]Violation of PRIMARY KEY constraint" error. I'd say I get one a week.
The only hint I have is that there is an insert that succeeds.  This insert happens at the same time as the one that fails. That seems that the code is executing twice at the exact same time which seems impossible. In fact it seems to me that a second query would have to happen so fast that the join shows no value (otherwise I'd get records updated = 0 rather than the error) but then the first (successful) record is created prior to the 2nd one being inserted.