February 11, 2009 at 3:01 am
Hi all,
I have the following configuration on a test environment server:
AMD 4-way dual core, so 8 procs
10 GB (8GB reserverd to SQL Server)
Microsoft Windows Enterprise Edition R2 X64
MSCS
SQL Server 2005 X64 failover cluster
performance flags:
number of connections: ~25
paging: ~0
disk queue: ~0
cpu utilization: in general is below 10%, but I from time to time: processor X is running at 100%
I have 4-10 runnable spids with a wait type of SOS_SCHEDULER_YIELD.
I tried everything:
- max degree of parallelism - 0,8,4 - no results
- query hints
I can't get rid of SOS_SCHEDULER_YIELD wait type.
Any clues?
February 12, 2009 at 3:16 pm
slavao should be able to help you out
February 13, 2009 at 4:35 am
Hi,
Have you identified what spid is utilizing CPU during 100% utilization?
Thanks,
Phillip Cox
February 13, 2009 at 5:03 am
Dan Puiu (2/11/2009)
I tried everything:I can't get rid of SOS_SCHEDULER_YIELD wait type.
Any clues?
Have you tried optimising the queries involved? (and I don't mean applying query hints)
Gail Shaw
Microsoft Certified Master: SQL Server, MVP, M.Sc (Comp Sci)
SQL In The Wild: Discussions on DB performance with occasional diversions into recoverability
February 13, 2009 at 5:09 pm
Have you tried 1 for the MAXDOP? 0,4 and 8 won't buy you much if you have queries running concurrently that SQL server wants to run parallel.
Anything else on the box that may be utilizing the CPUs?
February 14, 2009 at 2:03 pm
Kinda unrelated but you may want to check out the boot.ini parameter /USEPMTIMER (http://support.microsoft.com/kb/895980) if your on AMD x64.
February 16, 2009 at 2:03 am
Is the o/s fully patched and up to date?
ditto sql server?
[font="Comic Sans MS"]The GrumpyOldDBA[/font]
www.grumpyolddba.co.uk
http://sqlblogcasts.com/blogs/grumpyolddba/
February 16, 2009 at 2:04 am
btw. I'd have 2gb for o/s as totally inadequate - 4Gb maybe, ideally 6GB
[font="Comic Sans MS"]The GrumpyOldDBA[/font]
www.grumpyolddba.co.uk
http://sqlblogcasts.com/blogs/grumpyolddba/
February 18, 2009 at 11:45 am
Hi all,
So on another machine the same query that has that wait type runs in 2 mins and on this new machine is running for aprox. 2 hours;
The query is not complicated, it has an agregation and a subquery(if it's necessary I can post here);
The ideea is that after some optimization on that query i get results in 10 secs;
So it's realy strange
February 19, 2009 at 6:06 am
colin Leversuch-Roberts (2/16/2009)
btw. I'd have 2gb for o/s as totally inadequate - 4Gb maybe, ideally 6GB
why do you sai that? in my case(2GBs reserved for OS) I have total of 300MB free all the time, no matter the workload on the server
Viewing 10 posts - 1 through 9 (of 9 total)
You must be logged in to reply to this topic. Login to reply