Re: [exim] Simple(?) greylisting with exim?

Top Page
Delete this message
Reply to this message
Author: W B Hacker
Date:  
To: exim users
Subject: Re: [exim] Simple(?) greylisting with exim?
Heiko Schlittermann wrote:
> Hello Bill & others,
>
> I've updated the simple greylisting to be much more system independent.
> (Now using Perl linked into Exim). Please see http://www.schlittermann.de/doc/grey
>
> Now it doesn't eat all inodes, uses DBM files and should be *really*
> easy to use and configure, no external program/daemon is needed...
>
>     Best regards from Dresden
>     Viele Grüße aus Dresden
>     Heiko Schlittermann

>


Well... with all due respect, once perl-ized, I no longer see any particular
'simplicity' advantage over other perl (et al) implementations - even if the
perl'ishness is being handled by Exim.

As far as inode consumption, when the issue came up recently I did a 'du -i' on
what should be a worst-case scenario, an 'exired' variant that originally had no
time-outs or chroned clean-ups in many, many weeks of running and was reporting
several GB of fs storage used.

The inode use, however, was paltry, and a comparison of storage and inode
consumption indicated that it was not likely to ever become a factor on any
system I would run for an mx (no mounts with less than ~ 200 GB, as smaller HDD
are just too poor a value even if available).

Mind - that was on a mount-point with over 190 GB free, under FreeBSD 6.1, UFS2,
Gmirror RAID1, with more than a quarter-terabyte still in reserve on the other
two RAID1 arrays, (2U server, 6 HDD).

So... I'll stick with the simplest-of-the-simplest - a slightly slimmed-down
version of your original script.

NB: IF I needed a more sophisticated one (such as other-than-source-IP-only), I
have the tables in place within the PostgreSQL that we use to drive Exim to
support a very simple SQL one that Tollef Fog Heen wrote a coule of years back:

http://raw.no/personal/blog/tech/Debian/2004-03-14-15-55_greylisting

Other folks have since modified that for MySQL as well, and it seems easy to
alter/adapt, so a good starting point for SQL 'departures' in general.

That said, the real 'KISS" method w/r greylisting seems to be just to use it
very, very 'selectively' ...

Best,

Bill