Le mercredi 18 août 2010 à 11:27 -0700, Marc Perkel a écrit : > > On 8/18/2010 7:53 AM, Kris Deugau wrote: > > Alexandre Chapellon wrote: > >> When other well known DNSBL (I have always heard spamhaus sbl and xbl > >> are trust worthy) list less at most 50 entries , barrcuda lists > >> almost 8000!!!! > > > > That's not a problem all by itself, but when combined with this: > > > >> Finally there is a special feature that barrcuda folks call "deep > >> scanning" which makes the appliance scans the 'Received' headers and > >> reject the mails if an IP found in that headers, is listed in the > >> DNSBL... a feature that should obviously be called: 'even increase my > >> false positive rate' > > > > ... it makes life difficult. (In fact, if you provide Internet access > > for residential customers, a big chunk of your IP address space > > *should* be listed on Spamhaus' PBL - these IPs should be using your > > SMTP relay, or submitting mail via SMTP AUTH to another relay, not > > contacting recipient MXes directly.) > > > > I've had far too many incidents in the last ~6 months of having tech > > support ask me to dig into why a certain customer of ours is suddenly > > getting postmaster rejections on their mail to certain recipients - > > usually "important business contacts". > > > > All of them have proven to be recipients behind a Barracuda filter > > appliance that's deep-scanning headers and rejecting the message based > > on our customer's connection IP on our network - an IP behind our > > standard block for SMTP to anywhere but our own SMTP relay... and the > > rejected message was properly relayed through that system. Or worse, > > an IP on some other provider's network, where our mail customer is > > using SMTP AUTH on port 587 to relay through our server. > > > > I usually tell tech support to tell the customer that they'll have to > > contact the recipient by eg phone to let them know they're missing > > legitimate mail. > > > > -kgd > > > > I also scan IPs in received headers. I don't reject on that by itself > but it is a factor when combined with other conditions. >
I have no problem with this, this a normal behaviour (but personnally i would avoid using barracudaBL for this).
