Comments on: Wait Events 3 – What is Wrong with this Quote?

By: Charles Hooper

Charles Hooper — Fri, 10 Dec 2010 18:50:52 +0000

Flado,

Thank you for the research. When I read this section of the book I immediately thought that there must be something missing from the trace file, after all my 10g (10.2.0.2) trace file explains the meaning of the p1, p2, and p3 parameters as well as including tim= values on those wait event lines. This book section, which was probably written by Robert Freeman when Oracle Database 10.1 was the latest release, probably should have been updated to reflect changes to the 10046 trace file that were made since 2004 or 2005. The Oracle-l thread that you linked to contained part of a trace file that was captured by Oracle Database 10.1.0.3, and I tried to fix the formatting of a section of that file below:

EXEC #25:c=0,e=627,p=0,cr=0,cu=8,mis=0,r=1,dep=0,og=4,tim=208054417197
WAIT #25: nam='SQL*Net message to client' ela= 5 p1=1229996800 p2=1 p3=0
WAIT #25: nam='SQL*Net message from client' ela= 862 p1=1229996800 p2=1 p3=0

From the output posted in the book, it could just as likely be from Oracle 8i as it could be from 10g R1, which would mean a significant wait of 0.1 seconds.

Good catch stating that the ‘SQL*Net message to client’ wait event probably is not the best wait event to use in an example directly below the heading “Those Evil Wait Events in the 10046 Trace File”. A couple of other articles that might be of interest:
* http://blog.tanelpoder.com/2008/02/10/sqlnet-message-to-client-vs-sqlnet-more-data-to-client/
* http://blog.tanelpoder.com/2008/02/07/sqlnet-message-to-client-wait-gotcha/
* http://sites.google.com/site/embtdbo/wait-event-documentation/oracle-network-waits

By: Flado

Flado — Fri, 10 Dec 2010 10:41:01 +0000

Found a better link:
https://portal.hotsos.com/newsletters/volume-i-issue-2/tips/sql-net-message-to-client-isn2019t-what-you-might-think-it-is/

By: Flado

Flado — Fri, 10 Dec 2010 08:18:34 +0000

Yeah… For a 10g instance, the waits are missing some features (parameter names, obj#, tim). Indeed, they look suspiciously like 8i waits, in which case ela would be in centiseconds.
And ‘SQL*Net message to client’ is really a poor choice for an example, as it is one of the few “unreliable” events in the sense that neither the byte count nor the elapsed time are correct (http://www.freelists.org/post/oracle-l/SQLNet-Message-to-client-wait-on-batch-job,4).

Cheers,
Flado

By: Charles Hooper

Charles Hooper — Thu, 09 Dec 2010 23:32:35 +0000

Maybe I should not have provided a trace file section that was captured on April Fool’s Day?

To help you out, take a look at these lines that were captured to a trace file on December 11, 2005:

...
Oracle8i Release 8.1.7.3.0 - Production
...
WAIT #5: nam='SQL*Net message to client' ela= 0 p1=1413697536 p2=1 p3=0
WAIT #5: nam='SQL*Net message from client' ela= 0 p1=1413697536 p2=1 p3=0
WAIT #5: nam='SQL*Net message to client' ela= 0 p1=1413697536 p2=1 p3=0
FETCH #5:c=0,e=0,p=0,cr=1,cu=4,mis=0,r=1,dep=0,og=4,tim=369800029
...

By: Charles Hooper

Charles Hooper — Thu, 09 Dec 2010 18:03:41 +0000

This one might be more difficult than I originally thought. To help you out, take a look at these lines that were captured to a trace file on April 1, 2006:

...
Oracle Database 10g Release 10.2.0.2.0 - 64bit Production
...
WAIT #3: nam='SQL*Net message to client' ela= 1 driver id=1413697536 #bytes=1 p3=0 obj#=-1 tim=4518311899
WAIT #3: nam='SQL*Net message from client' ela= 141 driver id=1413697536 #bytes=1 p3=0 obj#=-1 tim=4518312106
WAIT #3: nam='SQL*Net message to client' ela= 1 driver id=1413697536 #bytes=1 p3=0 obj#=-1 tim=4518312250
FETCH #3:c=0,e=130,p=0,cr=2,cu=0,mis=0,r=1,dep=0,og=2,tim=4518312326
...