hi
we have some troubles with a object that kills our SQL2000SP4. CPU spikes to
100% and stay there. We are searching for a solution, but currently don't
have one! So it comes to situation where a object named
"company.dbo.fn_example" have more then one ProcessID. If this occour -
"all" other processes are zombies and i only know how to kill them by hand.
this makes me *stressed*, while i must kill 10 every 20 minutes.
is there any way to kill the object for e.g. with all processes inside - by
command line?
Regards
MarcSounds like a very poorly written function that is monopolizing the CPU's.
Try setting MAXDOP to 1 and see if it stays contained to just one CPU. But
I would look at other ways to optimize the statements that use this
function.
--
Andrew J. Kelly SQL MVP
"Marc Bauer" <marc.bau@.gmx.net> wrote in message
news:%23IxhtuvLGHA.2300@.TK2MSFTNGP15.phx.gbl...
> hi
> we have some troubles with a object that kills our SQL2000SP4. CPU spikes
> to 100% and stay there. We are searching for a solution, but currently
> don't have one! So it comes to situation where a object named
> "company.dbo.fn_example" have more then one ProcessID. If this occour -
> "all" other processes are zombies and i only know how to kill them by
> hand. this makes me *stressed*, while i must kill 10 every 20 minutes.
> is there any way to kill the object for e.g. with all processes inside -
> by command line?
>
> Regards
> Marc
>|||hi
> Sounds like a very poorly written function that is monopolizing the CPU's.
> Try setting MAXDOP to 1 and see if it stays contained to just one CPU.
> But I would look at other ways to optimize the statements that use this
> function.
our database specialist, reviewed the function and said - that this one is
correct and should work very fast. but sometimes - if this function gets
executed very very often it comes to something like a deadlock situation.
then all tables are locked... i have had a look to the function today, and
there is everywhere in the selects a with(nolock) used. so it cannot be a
read deadlock.
so - today - we don't know what's wrong. maybe a bug in SQL and we need to
find a workaround... this should be partly a con job with a kill - until we
found a solution.
Regards
Marc|||Can you post the function?
--
Andrew J. Kelly SQL MVP
"Marc Bauer" <marc.bau@.gmx.net> wrote in message
news:unyGffyLGHA.420@.tk2msftngp13.phx.gbl...
> hi
>> Sounds like a very poorly written function that is monopolizing the
>> CPU's. Try setting MAXDOP to 1 and see if it stays contained to just one
>> CPU. But I would look at other ways to optimize the statements that use
>> this function.
> our database specialist, reviewed the function and said - that this one is
> correct and should work very fast. but sometimes - if this function gets
> executed very very often it comes to something like a deadlock situation.
> then all tables are locked... i have had a look to the function today, and
> there is everywhere in the selects a with(nolock) used. so it cannot be a
> read deadlock.
> so - today - we don't know what's wrong. maybe a bug in SQL and we need to
> find a workaround... this should be partly a con job with a kill - until
> we found a solution.
>
> Regards
> Marc
>|||On Sat, 11 Feb 2006 17:26:09 +0100, "Marc Bauer" <marc.bau@.gmx.net>
wrote:
> but sometimes - if this function gets
>executed very very often it comes to something like a deadlock situation.
>then all tables are locked...
What do you mean, "very very often"?
Does the calling code happen to use #temp tables?
If you post source code, please also post what you can of the
invocation.
J.|||we found a endless loop in the function... :-(
Marc
订阅:
博文评论 (Atom)
没有评论:
发表评论