[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Spam detection & rejection
- From: John Summerfied <debian herakles homelinux org>
- To: "Red Hat Enterprise Linux 4 (Nahant) Discussion List" <nahant-list redhat com>
- Subject: Spam detection & rejection
- Date: Wed, 13 Jul 2005 11:02:08 +0800
This Q isn't specific to Nahant, but otoh there are probably a few
thoughtful souls here with useful thoughts and opinions on the topic,
and the volume of email isn't so great as to cause those thoughtful ones
to skip over lots of stuff.
I see two main ways "chicken bone" spammers garner email addresses:
1. Scan usenet, web sites and email lists (how many are subscribed here
to harvest email addresses?) and collect addresses that way.
2. Enumerate likely addresses such as
{bob,paul,john,susan,suzanne} example com
The first is susceptible to harvesting some spambait addresses such as
those in my sig.
The second one can spot my manually pursuing logs (or better, logwatch
summaries) for bounces.
So far, most attention I've seen for blocking spam has centred on black
lists, some free some not. Maintaining these lists necessarily involves
some delay while reports are evaluated and updates created. Then there
are ideas such as grey-listing, teergrubbing (spelling?), sender-id and
such.
I have a new idea I'd like to bounce off the forementioned Thoughtful Ones.
On a Debian box or two that I maintain I use a packages called
pop-before-smtp which is basically a configurable Piece o Perl that
peruses mail logs looking foe POP an IMAP logins. When it finds one it
address an entry to a hash (which it expires half an hour or so later).
I have configured postfix to use this hash to help decide whether sites
external to ours (eg staff at home or roaming th world) can send mail
through our servers.
It seems to me that this same technique can be used to detect attempts
to send email to non-existent accounts (we don't have susan or suzanne)
or which fail spamassassin tests and block the source for a day or so.
I particularly do not want to hear from anyone who emails
1aaaaaaa computerdatasafe com au - one try and two days in the sin bin
seems fine to me, and I don't see a good reason to treat anyone who
emails john example com any better; I have several email addresses at
various domains and none of them is john so anyone trying that's clearly
fishing.
Thoughts anyone? In particular, is this a Bad Idea? Can the idea be
improved?
--
Cheers
John
-- spambait
1aaaaaaa computerdatasafe com au Z1aaaaaaa computerdatasafe com au
Tourist pics http://portgeographe.environmentaldisasters.cds.merseine.nu/
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]