RE: Help interpreting 10046 trace file anomaly
- From: "Aponte, Tony" <AponteT@xxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Mon, 24 May 2004 10:48:58 -0400
How big/small is the difference in your trace files? I'm asking because
the differences I'm used to seeing are in a few centi-seconds and this
case was almost 46 seconds.
Thanks for taking the time.=20
Tony Aponte
-----Original Message-----
From: Daniel W. Fink [mailto:optimaldba@xxxxxxxxx]
Sent: Thursday, May 06, 2004 7:48 PM
To: oracle-l@xxxxxxxxxxxxx
Subject: Re: Help interpreting 10046 trace file anomaly
Tony,
It is important to remember that not all code is instrumented for=20
timing, so the session could have performed some of that code. Another=20
clue is that there is a timestamp emitted after the FETCH. I have see=20
this occur if an entry is not emitted into the time stamp for a certain=20
period of time (seems to be around 3 seconds). In looking at a couple of
trace files from past tests, I see that the TIM values don't seem to=20
true up with the e=3D values. So far, it has not had any impact in terms =
of mis-diagnosing a performance issue.
Daniel Fink
Aponte, Tony wrote:
>I need help understanding the time difference between the to "tim=3D"=20
>values. I calculated the difference to be (1007496616 - 1007492040)=20
>4576. But the time clocked in "SQL*Net message from client" is
reported=20
>as 1 centisecond. The "***" line shows that there was a long pause
but=20
>why doesn't the wait event account for the elapsed time? This is the=20
>first time I observe a discrepancy while walking the trace file. Any=20
>thoughts would be greatly appreciated.
>
>FETCH =
#4:c=3D0,e=3D0,p=3D0,cr=3D20,cu=3D0,mis=3D0,r=3D1,dep=3D0,og=3D3,tim=3D10=
07492040
>*** 2004-05-06 00:13:10.168
>WAIT #4: nam=3D'SQL*Net message from client' ela=3D 1 p1=3D1413697536 =
p2=3D1
p3=3D0
>=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
>PARSING IN CURSOR #4 len=3D422 dep=3D0 uid=3D22564 oct=3D3 lid=3D22564
tim=3D1007496616 hv=3D557190862 ad=3D'eb942db8'
>SELECT
>
>Tony Aponte
>Home Shopping Network
----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to: oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------
Other related posts:
