Morning all,

I'm having problems with my FTS rankings on our SQL2005 database and I'm a
bit stumped as where to start looking at what the the problem might be.

The simplified SQL is as follows:
---------------------------------------------------------------------------------------------

DECLARE @SEARCH nvarchar(2000)

SET @search = '"touchpoints"'

SELECT TOP 250 [KEY], SUM(Rank) AS Rank, content_live.title

FROM

(

SELECT Rank, [KEY] from CONTAINSTABLE(content_live,(title), @search)

) AS ftt

INNER JOIN content_live ON content_live.id = [KEY]

GROUP BY

[KEY], content_live.title

ORDER BY Rank DESC

---------------------------------------------------------------------------------------------

This returns me the following example results:

---------------------------------------------------------------------------------------------

KEY    RANK  TITLE
3304    192    TouchPoints Site: Using TouchPoints
3324    96     TouchPoints Site: Contact Us
3267    64     TouchPoints Site: TouchPoints Home
2137    48     Vision vs. practice seminar - agencies, media owners and
clients to discuss their views on TouchPoints
3273    32     TouchPoints Site: How we do it

---------------------------------------------------------------------------------------------

My problem is that the 3rd result with a Rank of 64 should have a rank of
192 based upon the occurence of the word "TouchPoints".

If I rebuild the index then I get the correct results back, but with the
knowledge that it'll start going wrong again.

The table in question is set to Track Changes : Automatic.

Title field is VARCHAR(1000).

Version = Microsoft SQL Server 2005 - 9.00.3054.00 (Intel X86)   Mar 23 2007
16:28:52   Copyright (c) 1988-2005 Microsoft Corporation  Standard Edition
on Windows NT 5.2 (Build 3790: Service Pack 2)

Any suggestions of what might be causing this problem?

Are there some settings that I need to change to ensure that the Rankings
get updated correctly?

Do I need to go through the crawl logs to see if there is a problem that
stops population (even though I'm sure that the record in question hasn't
been changed).

Anyone else seen this issue?

Many thanks,

Larry.