> If I do this though, doesn't it mean that I would have to go through each
> email one-by-one?  I'm guessing that RBL & delaying will definitely help
> cut down on the pounding I'm getting, but over the weekend alone, my own
> accounts accumulated more than 1500 pieces of email, of which, probably
> 1400 were spam.  I can't even begin to imagine how long it would take me
> to go through them all.  If I cc both Bayesian spam & nonspam (setting
> both log settings to 7), then I would basically be receiving all emails,
> wouldn't I?
>
> Would I then be able to select all spam and submit to the asspspam &
> asspnotspam address as a single email with hundreds of attachments?
>
> There are privacy issues at stake here too; I may end up reading someone's
> legit email, which is definitely a concern.

I did/do this.  I ran ASSP for a long time with none of the users any wiser.
Leave everything in testmode.  Don't prefix the subject with [SPAM].  Do
turn on all connection tests but have them score only.  In your mta have a
system wide rule that looks for a header line that includes:

X-Assp-Spam-Reason:

but does not include:

X-Assp-Spam-Reason: Bayesian spam

and route those emails to the users individual spamboxes or a central 
spambox you control as those are most certainly spam.  The users will still 
receive their Bayesian spam but will see a nice drop in total received spam. 
Then occasionally search for text in files in your spam folder for your 
company name, street address and such.  Also scan your notspam folder for 
text in files with known spam words and sex words.  Do that every so often 
and as your users use the system and addresses get autowhitelisted the 
system will smooth out on it's own.  Then after a few weeks turn on the 
[SPAM] prefix for spam mail and let the users know that it has been 
identified as Bayesian spam, and they can either notify you of errors or you 
can educate them on how to use the email interface.  I have done this over 
the years and have lowered the false positives to an almost acceptable 
level, but I will never take Bayesian out of testmode.  Some of our users 
who have had great success with no false positives create a rule that 
deletes anything with [SPAM] in the header, some merely move them to a spam 
mailbox in their email program for review and deletion.  Even after all this 
time though, legitimate HTML emails from companies hawking their wares 
usually get marked as spam until they get whitelisted.  And I have had to 
add either domains to the whitelist or their servers if known to the accept 
all mail list to keep poorly configured servers from customers and 
associates from being marked as spam.

The image spam everyone is talking about is usually caught by the connection 
test or spambomb regex, but if it isn't, it is always marked as Bayesian 
spam.

BTW, delaying makes a huge difference right off the bat.  It also will 
almost completely eliminate the virus laden emails, almost.

Doug Traylor



-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys - and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Assp-user mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/assp-user

Reply via email to