SQL Server create primary key constraint duplicate key error - sql

Question

I have been experiencing some strange behaviour with one of my SQL commands taken from one of our stored procedures.
This command follows the below order of execution:
1) Drop table
2) Select * into table name from live server
3) Alter table to apply PK - this step fails once out of 4 daily executions
My SQL statement:
IF EXISTS (SELECT * FROM sys.objects WHERE object_id = OBJECT_ID(N'
[inf].[tblBase_MyTable]') AND type in (N'U'))
DROP TABLE [inf].[tblBase_MyTable]
SELECT * INTO [inf].[tblBase_MyTable]
FROM LiveServer.KMS_ALLOCATION WITH (NOLOCK)
ALTER TABLE [inf].[tblBase_MyTable] ADD
CONSTRAINT [PK_KMS_ALLOCATION] PRIMARY KEY NONCLUSTERED
(
[ID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY =
OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
GRANT SELECT ON [inf].[tblBase_MyTable] TO ourGroup
This is very strange considering the table is dropped, and I thought the indexes / keys would also be dropped. However I get this error at the same time every day. Any advice would be very much appreciated.
Error:
The CREATE UNIQUE INDEX statement terminated because a duplicate key was found for the object name 'inf.tblBase_MyTable' and the index name 'PK_KMS_ALLOCATION'.

Answer 1

Duplicate keys in [inf].[tblBase_MyTable] table are actually possible thanks to the WITH (NOLOCK) hint which allows "dirty reads". Have a look at blog which describes this in detail: SQL Server NOLOCK Hint & other poor ideas:
What many people think NOLOCK is doing
Most people think the NOLOCK hint just reads rows & doesn’t have to
wait till others have committed their updates or selects. If someone
is updating, that is OK. If they’ve changed a value then 99.999% of
the time they will commit, so it’s OK to read it before they commit.
If they haven’t changed the record yet then it saves me waiting, its
like my transaction happened before theirs did.
The Problem
The issue is that transactions do more than just update the row. Often
they require an index to be updated OR they run out of space on the
data page. This may require new pages to be allocated & existing rows
on that page to be moved, called a PageSplit. It is possible for your
select to completely miss a number of rows &/or count other rows
twice.

Answer 2

Well... you might have to repeat creating the new table and filling it until the check-query from #DarkoMartinovic does not return duplicates. Only then you can continue to add the PK. But this solution might cause heavy load on your live system. And you nave no guarantee that you have a 1:1 copy of the data as well.

Answer 3

Having reviewed various helpful comments here, I have decided against (for now) implementing SNAPSHOT isolation as this interface does not make use of a proper staging environment.
To move to this would mean either creating a staging area and setting that database to READ COMMITTED SNAPSHOT isolation, and a rebuild of the entire interface.
To that end and on the basis of saving development time, we have opted for ensuring that any ghost reads where dupes could be brought across from the source are handled before applying the PK.
This is by no means an ideal solution in terms of performance on the target server but will provide some headroom for now and certainly remove the previous error.
SQL approach below:
DECLARE #ALLOCTABLE TABLE
(SEQ INT, ID NVARCHAR(1000), CLASSID NVARCHAR(1000), [VERSION] NVARCHAR(25), [TYPE]
NVARCHAR(100), VERSIONSEQUENCE NVARCHAR(100), VERSIONSEQUENCE_TO NVARCHAR(100),
BRANCHID NVARCHAR(100), ISDELETED INT, RESOURCE_CLASS NVARCHAR(25), RESOURCE_ID
NVARCHAR(100), WARD_ID NVARCHAR(100), ISCOMPLETE INT, TASK_ID NVARCHAR(100));
------- ALLOCATION
IF EXISTS (SELECT * FROM sys.objects WHERE object_id = OBJECT_ID(N'[inf].
[tblBase_MyTable]') AND type in (N'U'))
DROP TABLE [inf].[tblBase_MyTable]
SELECT * INTO [inf].[tblBase_MyTable]
FROM LiveServer.KMS_ALLOCATION WITH (NOLOCK)
INSERT INTO #ALLOCTABLE
SELECT *
FROM
(SELECT
ROW_NUMBER() OVER (PARTITION BY ID ORDER BY ISCOMPLETE DESC) SEQ, AL.*
FROM [inf].[tblBase_MyTable] AL
)DUPS
WHERE SEQ >1
DELETE FROM [inf].[tblBase_MyTable]
WHERE ID IN (SELECT ID FROM #ALLOCTABLE)
AND ISCOMPLETE = 0
ALTER TABLE [inf].[tblBase_MyTable] ADD CONSTRAINT
[PK_KMS_ALLOCATION] PRIMARY KEY NONCLUSTERED
(
[ID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON)
GRANT SELECT ON [inf].[tblBase_MyTable] TO OurGroup