Yes, that perfmon counter is only showing how many deadlocks are encountered
in the last second. The alert fires for every deadlock encountered. In your
case if you are seeing a deadlock per second, then it is not good.
[quoted text, click to view] >> trace on 1204 cause error log file increase quickly, this
>> is dangerous.
Well, what is your priority? You want to identify and fix all the deadlocks
or rather keep the error logs small and clean?
--
HTH,
Vyas, MVP (SQL Server)
http://vyaskn.tripod.com/ What hardware is your SQL Server running on?
http://vyaskn.tripod.com/poll.htm [quoted text, click to view] "hb_zhang" <hb_zhang@smics.com> wrote in message
news:06ee01c3557a$6f99fe40$a301280a@phx.gbl...
It's the same monitor item
SQL Server:Locks --- Number of Deadlocks/sec
Are you sure this item on SQL Alert become a encountered
item?
trace on 1204 cause error log file increase quickly, this
is dangerous.
Thanks.
[quoted text, click to view] >-----Original Message-----
>I think the system monitor counter is showing average
over a period of time,
>which is different from the total number of deadlocks you
encountered.
>
>Best way to track down deadlocks is to enable traceflags
1204 and 3605 to
>capture detailed deadlock informatin in the errorlog.
Then start
>troubleshooting the deadlocking procedures.
>
>--
>HTH,
>Vyas, MVP (SQL Server)
>
http://vyaskn.tripod.com/ >What hardware is your SQL Server running on?
>
http://vyaskn.tripod.com/poll.htm >
>
>
>
>"hb_zhang" <hb_zhang@smics.com> wrote in message
>news:04a101c354e3$7b438dc0$a301280a@phx.gbl...
>we monitor "Number of Deadlocks/sec" use system monitor
>and SQL alert, but we get the very different result,
>
>system monitor only get 1 or 0
>
>SQL alert get mor than 100.
>
>Why the value is so fidderent? which value can we use for
>performance tuning?
>
>Thanks.
>
>
>.
>
