[KLUG Members] Tuning qmail

[Adam Tauno Williams]

> > I've looked at other things and they are all either slow and shaky perl
> > crap, too complicated, or require too much user interaction.  If the
> > user actually has to *DO* something, they'll just delete the messages
> > and ignore them (well, except for an enlightened *FEW*).  And none of
> > these tools really help solve the problem;  the SPAM continues to
> > arrive, get miltered, stored - whether or not the user takes the time to
> > check/flag/mark/whatever... the message.  So what does the *SYSTEM*
> > gain? Absa friggin' lutely nothin'.  So it is a user problem, not one
> > for the system to solve (IMHO).
> I tend to agree with you, but my boss is getting bugged by users who are
> "offended" by their spam, and is telling me to try something else.  So,
> I'll try some of that flaky perl crap and see what happens ...

Yea, I know that feeling.  Since we are using Cyrus all our users have the
option to define their own filtering rules via SIEVE for whatever reasons they
like - I just point to that when someone whines and say "Here ya' go".  Most
then shut-up once presented which a tool (and if I go back and look almost none
of those whiners take the time to define anything).  And timsieved (a daemon,
thus no forking) is written in C and recent Cyrus releases actually compile the
SIEVE script - so it scales to the moon.

> > And we encourage our users to get a SPAMCop submission address and
> > forward SPAM messages.  The enlightened *FEW* get their head around this
> > pretty easy.
> I used to report all my spam, until the guy who runs spamcop told me to
> stop it.  :-)
> You see, our anti-virus software (RAV, milter), does something weird to
> the headers, causing spamcop's reporting system problems.  Hopefully
> that'll change sometime soon when I switch to clamav or something else.

Well thats what you get for using some proprietary package! :)