Marco Fretz schrieb am Thursday, den 21. April 2011:

> 
> On Apr 19, 2011, at 9:26 AM, Alexander Wirt wrote:
> 
> > Marco Fretz schrieb am Tuesday, den 19. April 2011:
> > 
> >> 
> >> On Apr 1, 2011, at 5:53 PM, Mark Martinec wrote:
> >> 
> >>> Kshitij,
> >>> 
> >>>> Hello everybody from this i know very well what i am doing rather than
> >>>> guiding me for experience.
> >>>> 
> >>>> Please any can tell be the copnfiguration for amavisd-new to enable
> >>>> virussendernotification .
> >>> 
> >>> $final_virus_destiny  = D_BOUNCE;
> >>> @viruses_that_fake_sender_maps = ();
> >>> 
> >>> But ... don't do that!
> >>> 
> >>> The only sensible way to warn a potentially valid sender of
> >>> infected mail is to use a pre-queue setup and to D_REJECT viruses,
> >>> *not* to bounce them.
> >> 
> >> btw. That is the best way to handle virus and spam mails! Reject in 
> >> smtp-dialoge with sending mailserver. That way the sending mailserver will 
> >> create the bound back to the original sender (if there is one) and not 
> >> you, you won't have any problems with getting blacklisted and you don't 
> >> have to handle the junk mails.
> > That is indeed the best way to get unsubscribed from mailinglists as this
> > behaviour triggers bounce handling. Unless you are able to discard mails of
> > severity junk or lists - don't do this. 
> 
> I never thought about that. It's an interesting point: The question is why 
> should a mailing list forward spam to its members when my spamassassin can 
> reject them with a clear score above 6.31 (the best proven reject level for 
> SA). and if the blacklist does forward such spam (thus is not using content 
> filtering) why should I as a mail provider care? even for a workaround our 
> users are able to whitelist senders via a Web user interface for their 
> accounts, domains, etc. 
we are running a listserver for ~200k users in several languages. It is
impossible to have spamfilters that get every spam without false positives.
As a public service we usually are not able to do spamfiltering as hard a
user would do it for its private mail. 
 
> You should really read "Das Postfix Buch" by "Peer Heinlein" if you haven't 
> yet. He makes a really good point about spamhandling, false-positives and 
> lawful background. 
I usually don't follow his points. 

> As I said, we haven't had any problems so far. If more and more providers and 
> enterprises start using this spam handling strategy (what's currently 
> happening) the mailinglists will have to think again about their bound 
> handling and spam filtering solution. 
Ehm, please read the relevant SMTP RFCs. Its not really possible to
differate a bounce if a user don't exists from a spam reject. (Yeah you could
start parsing dsn's but this is not really an option). 

> another question you could ask yourself: why should I fight with my customers 
> about false-positves laying around in their junk-boxes? why not let the 
> sender fight with his provider why his mail was marked as spam somewhere in 
> the world? 
If you subscribe to a mailinglist accept the mail - or don't subscribe. If
you discard it afterwards I/we don't care. 

Alex

Reply via email to