RE: Negative wait time
- From: "Cary Millsap" <cary.millsap@xxxxxxxxxx>
- To: <oracle-l@xxxxxxxxxxxxx>
- Date: Mon, 2 Feb 2004 10:32:51 -0600
Not exactly, but kind of. It's not a negative "wait event" duration,
after all; it's that the difference e - c is negative (that is, c > e).
It's possible that some process got a c=10000 recorded in its process
table when the process spent only, say, 100 microseconds on the CPU. So,
for a single dbcall, you really don't have enough data to determine
what's going on. But over a span of several dbcalls, your negative and
positive quantization errors will tend to offset each other. As long as
total "unaccounted-for" time is between -10% and +10% of your total
response time, it's safe to ignore it.
By the way, another cause of the c > e phenomenon is the fact that
Oracle double-counts some CPU time in both the 'c' and 'ela' statistics.
The problem is most prominent for applications that do large multi-block
read calls.
Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com
* Nullius in verba *
Upcoming events:
- Performance Diagnosis 101: 2/24 San Diego, 3/23 Park City, 4/6 Seattle
- SQL Optimization 101: 2/16 Dallas
- Hotsos Symposium 2004: March 7-10 Dallas
- Visit www.hotsos.com for schedule details...
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Jamadagni, Rajendra
Sent: Monday, February 02, 2004 9:54 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: RE: Negative wait time
Thanks Cary, so is it okay to ignore negative wait times?
Raj
------------------------------------------------------------------------
-=
-------
Rajendra dot Jamadagni at nospamespn dot com
All Views expressed in this email are strictly personal.
QOTD: Any clod can have facts, having an opinion is an art !
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx]On Behalf Of Cary Millsap
Sent: Monday, February 02, 2004 10:48 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: RE: Negative wait time
Quantization error.
The error is greater on the 'c' statistic than any other, because it is
obtained by polling in 10,000-microsecond increments.
Cary Millsap
Hotsos Enterprises, Ltd.
http://www.hotsos.com
* Nullius in verba *
Upcoming events:
- Performance Diagnosis 101: 2/24 San Diego, 3/23 Park City, 4/6 Seattle
- SQL Optimization 101: 2/16 Dallas
- Hotsos Symposium 2004: March 7-10 Dallas
- Visit www.hotsos.com for schedule details...
-----Original Message-----
From: oracle-l-bounce@xxxxxxxxxxxxx
[mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Jamadagni, Rajendra
Sent: Monday, February 02, 2004 8:03 AM
To: Oracle List (E-mail)
Subject: Negative wait time
Okay it is Monday and I seem to have difficulty remembering stuff ...
I wrote a pl/sql procedure that analyzes a given trace file and produces
output like this ... but look at wait time ... why is it < 0 ??=20
wait =3D elapsed-cpu right? What am I missing? In case you are wondering
... this procedure (and its output) is based on Cary's paper 'LIO Vs
PIO' ... not precise, but should be pretty close.
SQL Hash Value : 98525727
SELECT count(distinct(us.Ut_Id)) Res_Prmpt
From yyyy ms, zzzz us
Where ms.Mg_Unit_Id =3D us.Ut_Id
And us.Pob_Id =3D :b4
and ( :b3 IS NULL OR us.Line_Number =3D :b3 )
And us.Monday_Dt between :b2 AND :b1
And us.Makegood =3D 'A'
And us.Reason_Cd <> -19
And ms.Mg_Unit_Type =3D 'P'
And us.Ut_Id Not In (Select ms1.Mg_Unit_Id
From xxxx ms1, yyyy mh
Where ms1.Mg_Number =3D mh.Mg_Number
And ms1.Mg_Unit_Type =3D 'M'
And mh.Mg_Pob_Id =3D :b4
And mh.Mg_Line_Number =3D us.Line_Number)
-------- Response Time --------
Action Count Rows Ela Time CPU Time Wait Time LIO Blocks PIO
Blocks
------ ------ --------- ---------- ---------- ---------- ----------
----------
PARSE 1 0 0.000 0.000 0.000 0
0
EXEC 51 0 0.004 0.010 -0.006 0
0
FETCH 51 51 0.450 0.390 0.060 9047
11
------ ------ --------- ---------- ---------- ---------- ----------
----------
TOTAL 103 51 0.454 0.400 0.054 9047
11
Any pointers are welcome, and yes please excuse me while I go have
another coffee.
Raj
------------------------------------------------------------------------
--------
Rajendra dot Jamadagni at nospamespn dot com
All Views expressed in this email are strictly personal.
QOTD: Any clod can have facts, having an opinion is an art !
----------------------------------------------------------------
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
-----------------------------------------------------------------
----------------------------------------------------------------
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
-----------------------------------------------------------------
----------------------------------------------------------------
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
-----------------------------------------------------------------
----------------------------------------------------------------
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:
