On Tue, Jan 15, 2008 at 14:51:09 +0100, Mark Martinec wrote: > TimeElapsedReceiving is the elapsed time (wall clock) between > the reception of a MAIL FROM smtp command from a MTA (this is where > a new smtp transaction starts) and the time when content checking > begins (when check_mail() is called), so this time interval consists > mostly of a data transfer of a message from MTA to amavisd > (RCPT TO and DATA smtp commands).
Ah, OK. So basically this a value that describes more the behaviour of the SMTP clients than the performance of the content scanner. > The TimeElapsed* counters are collected at a top level of a program > flow and are of rather course granularity. The fine granularity timing > measurements are more finely intertwined in the code and end up > in log level 2 logging. In principle there could be more TimeElapsed* > counters, but then one could always say: why this one and not also > that one, so I have to draw a line somewhere. I understand. log_level=2 is not really an option for us, however, because we don't want to log that much for every mail. I can live with it though :-) > For occasional assessment you could run a one-liner Perl command > across a log to collect times of interesting tasks, and perhaps > plot a scattergram (gnuplot) for each such task. The diagram > could be more revealing than a cumulative value. Yep, that's what I am going to do if we have performance issues and I want to find out where the bottleneck is. The timing report is really cool. Thanks again for all the answers David ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2008. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ AMaViS-user mailing list AMaViS-user@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/amavis-user AMaViS-FAQ:http://www.amavis.org/amavis-faq.php3 AMaViS-HowTos:http://www.amavis.org/howto/
