aggregate query reporting different results

forgive the absense of DDL and data to reproduce this problem, but it
is 8 tables and hundreds of thousands of rows of data. instead, let me
just post the query and the symptom and we'll see how far that can get
me.
select r.description, sum(t.amount), count(*)
from salesregion as r
join metroarea as w on r.salesregionid = w.salesregionid
join saleitems as a on a.website = w.id
join transactions as t on t.adid = a.adid
join orders as o on t.orderid = o.orderid
join products as p on t.productid = p.productid
where o.orderpaid = 1
and datepart(mm, o.datecreated) = 2
and datepart(year, o.datecreated) = 2006
and p.type in (1, 3)
group by r.description
order by r.description asc
when this query is run in our production support environment, which has
almost no load, it produces completely predictable results every time.
however, when this same query is run in our production environment,
which has identical data, but a much higher load, the aggregate
functions fluctuate very slightly (one to three rows).
i have read on this issue and found two possible remedies:
1) "option (maxdop 1)" (a response i've seen in many threads on this
ng) - i applied this, it had no effect at all on the stability of the
results
2) apply sql service pack 4 (known parallelism bug described at
http://support.microsoft.com/kb/814509) - my question here is: does
load affect parallelism? because production and production support
environments both have service pack 3 applied, but it's happening in
one and not the other. i will apply the service pack if necessary, but
i want to make sure that the parallelism bug is known to occur only
under high load.
thanks in advance for any help,
jasonoh, an additional note:
when i further narrow the search results to just calculate the
aggregates for a single salesregion, like so:
select r.description, sum(t.amount), count(*)
from salesregion as r
join metroarea as w on r.salesregionid = w.salesregionid
join saleitems as a on a.website = w.id
join transactions as t on t.adid = a.adid
join orders as o on t.orderid = o.orderid
join products as p on t.productid = p.productid
where o.orderpaid = 1
and datepart(mm, o.datecreated) = 2
and datepart(year, o.datecreated) = 2006
and p.type in (1, 3)
AND r.description = 'Arizona' -- NEW LINE
group by r.description
order by r.description asc
then it becomes deterministic again!

after loop wants to start for top with out closing cursor

Hi All,

I need your help urgently. I am using SQL Server. First i try to explain what i want to do. Imagine a warehouse having thousands of books or whatever. Sometimes multiple copies of same books can be on the same shelf but its also possible that 10 copies of a book are placed in 10 different shelf means one book per shelf. So i am writing program to pick location for orders received. My progarm works fantastic if a customer order single copy of any number of differnt books. But if a buyer order a 5 copies of the same book and if first location my progarm picked has only 1 or any number less than actual quantity ordered then my program fails to pick second or third location.
I am Using a cursor to fetch and process every single order and execute a stored procedures to pick location and update inventory. I want that after it has picked first location program should run last step without with same values in cursor but one updated value: (just want to redirect to last step) In VB we can use rst.Requery but what about SQL server

Following is a my code

CREATE PROCEDURE OPENCRSR2

--parametes needed to birng orders info in to program
@.var_orid varchar(25),
@.var_itemid varchar(25),
@.var_sku Varchar(25),
@.var_qtsold numeric(9)

AS

--variables to store fetched info from tables

Declare @.var_shelf numeric(9) ,
@.var_dateadded smalldatetime,
@.var_qtonhand numeric(9),
@.var_sporder varchar(25),
@.var_supplier varchar(25),
@.var_qtstore numeric(9)


DECLARE CrsrInside CURSOR FOR

SELECT Min(A.Shelf), A.DateAdded, Sum(A.QuantityOnHand) FROM UniversalBooks.dbo.tblInventory AS A
WHERE ((A.ISBN) = @.var_sku AND (A.QuantityOnHand>0)
AND (A.DateAdded=(SELECT Min(B.DateAdded)
From UniversalBooks.dbo.tblInventory as B
WHERE ((A.ISBN =B.ISBN) AND (B.QuantityOnHand>0)))))
GROUP BY A.DateAdded, A.QuantityOnHand

OPEN CrsrInside

FETCH NEXT FROM CrsrInside
INTO @.var_shelf, @.var_dateadded, @.var_qtonhand
--SELECT @.var_sku as ISBN, @.var_shelf as Shelf, @.var_dateadded as Dateadded
EXEC de_fetchstage2 @.var_sku, @.var_shelf, @.var_dateadded, @.var_qtonhand OUTPUT, @.var_sporder OUTPUT, @.var_supplier OUTPUT

--SELECT @.var_qtonhand as QtAvialable



IF @.var_qtsold > @.var_qtonhand
BEGIN
SET @.var_qtstore = @.var_qtsold - @.var_qtonhand
SET @.var_qtsold = @.var_qtonhand

--Following excute statement will insert picked shelf and other info into different table and subtarct quantity sold from quantity on hand but still i have few books to be picked for same order and @.var_qtstore has that number i want cursor should use same varibles but new value from @.var_qtsold which has the number of books to be picked(after execute Statement)


Exec de_insertstage3 @.var_sku, @.var_shelf, @.var_dateadded, @.var_qtonhand, @.var_sporder, @.var_supplier, @.var_orid, @.var_itemid, @.var_qtsold

SET @.var_qtsold = @.var_qtstore

--**** HERE I want to redirect my code to start from the current if statement instead of starting from next fetched cursor
END

ELSE
--REST OF THE CODE

--SELECT @.var_sku as ISBN, @.var_shelf as Shelf, @.var_dateadded as Dateadded, @.var_qtsold as Qtsold,
@.var_qtonhand as QuantityAvailable, @.var_sporder as OrderNumber, @.var_supplier as Supplier
END

CLOSE CrsrInside

DEALLOCATE CrsrInside
GO

Any help will be appericiated

Thanks in advance
DevWould a WHILE statement work for you?:

WHILE @.var_qtsold > @.var_qtonhand
BEGIN
SET @.var_qtstore = @.var_qtsold - @.var_qtonhand
SET @.var_qtsold = @.var_qtonhand
Exec de_insertstage3 @.var_sku, @.var_shelf, @.var_dateadded, @.var_qtonhand, @.var_sporder, @.var_supplier, @.var_orid, @.var_itemid, @.var_qtsold
SET @.var_qtsold = @.var_qtstore
END

blindman