I've been using Postfix w/postgrey + spamassassin (via amavisd) +
RBL lists + SMTP sender verification.
My email address is very old, and I use a few aliases, so I still
get ~40 spams per day. Thunderbird's Bayesian filtering catches ~98% of
those that get to my Inbox.
I've looked into doing server-side Bayes spam filtering, however,
any solution I looked at could not be considered "secure". (It was a
couple of years ago... I forget exactly what the problem was. I think
it was because the Bayesian whitelist would contain keywords (and names)
from valid emails, thus exposing information about the organization. If
you sent a spam containing specific acronyms or part numbers, and it
didn't get bounced, you'd know those terms were being used in valid
emails in that organization.)
I also use the Mailhop store-and-forward service from DynDNS. That
protects me in the event my mail server (or Internet connection) goes
down -- it caches emails until my server comes back online. The Mailhop
service includes spam filtering, too, but my guess is that they're using
the same OSS tools I am, so I doubt that it helps.
I've been tempted to use TMDA:
http://tmda.net/
With TMDA, new email senders get a "Are you really a human?"
auto-response, which they must then reply to. If they reply, their
original message (and all subsequent messages) get through. If they
don't reply, you never see the first email (which is probably a spam).
I've done some consulting work in Europe and Asia... and here are
some of the spams I got today:
Тренажеры от 1500р.
dereks采购流程优化及供应商评估
合同管理与风险解析及规避
Любовь - не картошка, не выбросишь в окошко
ДИПЛОМЫ, АТТЕСТАТЫ, АКАДЕМИЧЕСКИЕ СПРАВКИ И Т.П.
--Derek
On 12/17/2009 07:21 PM, Kurt Buff wrote:
On Thu, Dec 17, 2009 at 15:58, Jules Agee <[email protected]> wrote:
I'm getting more complaints lately from users on my network about spam.
Investigating the complaints reveals they get between five or ten per
day on average. This is about what I expect in my own inbox, and I don't
mind because it's a lot better than it used to be.
But it seems like people have higher expectations these days, and I'm
wondering if I should consider alternatives to what I'm doing now.
Right now I'm using postgrey and spamassassin. According to my server
logs, internet MXs make about 1.5 million connections to our server per
week. 90% of those are blocked by greylisting, about 4% are tagged by
SpamAssassin, and the remaining 6% are delivered to users' inboxes
unchanged.
Are any of you getting better results with another solution?
-Jules
I use Maia Mailguard, and have few complaints about it. It uses
spamassassin underneath the covers, but the web interface allows much
better control for the users.
Also, do you get regular updates of rules and such?
Kurt
