Question

I tried to delete duplicate records in my DB. The only difference is the PrimaryKey which is a uniqueidentifier. I have about 1500 entries that have been duped so in all I’m looking at around 3000 entries. So I sectioned off about 60 entries (based on the receivedOn date) and executed my code to pare them down to 30 and OH CRAP the 30 disappeared! Here is the code I tried:

DELETE dupes
FROM [emailTable] dupes, [emailTable] fullTable
WHERE (dupes.ReceivedOn > '2009-08-18 23:59:59.999' AND dupes.ReceivedOn < '2009-08-20 00:00:00.000')
      AND (dupes.emlPath = fullTable.emlPath)
      AND NOT (dupes.GUID = fullTable.GUID)

My goal is to delete the duplicate. I dont care which one… but I need ONE of the two entries to stay on the server… Can anybody shed some light on what I did wrong?

Answer 1

You can do this without a second table. Something like this:

SELECT * FROM emailTable
WHERE EXISTS (
    SELECT * FROM emailTable AS t2
    WHERE t2.emlPath = emailTable.emlPath AND
    t2.GUID > emailTable.GUID)

That will show you which records are about to get deleted. If that’s okay, change it to:

DELETE FROM emailTable
WHERE EXISTS (
    SELECT * FROM emailTable AS t2
    WHERE t2.emlPath = emailTable.emlPath AND
    t2.GUID > emailTable.GUID)

The t2.GUID > emailTable.GUID will make sure that one record with that emlPath will remain in the table.