-----Original Message-----
From: Marvin Pierce
Sent: Tuesday, February 10, 2004 8:37 PM
Spamassassin is using a lot of processing time due to the number of
procmail recipes I have.
I think it is highly unlikely that procmail is taking even 1% of the
total time taken by Spamasassin, using all the recipes that you have
now.
Try a test:
1. Save a largish message to a text file, including all the headers.
Let's call the file msg.txt. Make sure that msg.txt is selected as
a message that is not on a mailing list, is _not_ spam, and that
is ultimately delivered to your inbox.
2. Run a simple test with the recipes that you have now, including
spamassassin. Then do this:
time procmail < msg.txt
3. Comment out the spamassassin lines, and try again.
4. Compare the times.
I just did this, on a lightly loaded, P4 running at 2Ghz:
% wc msg.txt
1010 9729 63861 msg.txt
With spamassassin:
% time procmail < msg.txt
User=79.360 System=0.330 Wall=1:44.16 (U+S)/W=76.5% I/O=0/0
Without spamassassin:
gary(_at_)screamer[9]% time procmail < msg.txt
User=0.050 System=0.000 Wall=0:00.04 (U+S)/W=125.0%
Doing the same thing with spamc (which calls the spamd deamon) cut the
wall clock time to 20 secs. rather than 1 min. 44 secs., and although
the cpu time isn't recorded in the log, we might assume that the spamd
processing took several seconds of cpu time, and probably 100 times
more cpu time than the procmail script. Maybe a small message should
be tested as well since SA's runtime is proportional to the message
length, but I'm fairly certain that the conclusion will be the
same: the overhead of spamassassin swamps the miniscule procmail overhead.
_______________________________________________
procmail mailing list
procmail(_at_)lists(_dot_)RWTH-Aachen(_dot_)DE
http://MailMan.RWTH-Aachen.DE/mailman/listinfo/procmail