December 8, 2003 at 4:00 am
I have a table with 54 million rows. When I run this query:
select count(*) from search_results
where sr_cs_id='40865-TN'
and sr_prole_id is null
I get 6066 rows - which is incorrect
But when I run this query:
select count(*) from search_results
where sr_cs_id='40865-TN'
and sr_prole_id is NOT null
I get 5543 - which is correct
On further querying the total rows with an sr_cs_id = '40865-TN' is 6066, 5543 do have an entry in the sr_prole_id column.
I have dropped and recreated the indexes on this table, but it has not made a difference. When I transfer the 6066 result set into a new table the query works fine.
Does anyone have any ideas?
December 8, 2003 at 5:10 am
I'm going to guess that the query against the larger table exceeds the cost threshold for parallelism (while the smaller does not), causing this problem. Try using the MAXDOP 1 query hint.
--Jonathan
--Jonathan
December 8, 2003 at 9:14 am
From what that Links says I think you are right - Now I need to get the patch
Cheers
Geoff
Viewing 3 posts - 1 through 2 (of 2 total)
You must be logged in to reply to this topic. Login to reply