On 2/23/2012 3:55 PM, rg86...@airpost.net wrote: > Hello Noel, > > On Thu, Feb 23, 2012, at 03:31 PM, Noel Jones wrote: >> >> (sorry, had to work for a while) > > No worries. Nice to have someone reasonable to chat with. Bit of a dry > spell on that. It's appreciated. > >> On 2/23/2012 2:58 PM, rg86...@airpost.net wrote: >> Ah, a different server. Important infomation (or maybe I missed it...). > > Easy to miss. Fwiw, as originally @ > > http://marc.info/?l=postfix-users&m=133002076514425&w=2 > > my sendmail 'send' was sent from 192.168.1.13 to postfix listening @ > "inet_interfaces = 192.168.1.10 192.168.1.11". > > Clearly mistakenly having presumed that that's clear enough, I called it > out specifically only recently in response to the lashings. > >>> Feb 23 11:51:52 desk postfix/smtp[20627]: 986C040083: >>> to=<s...@mail.rogermail.lan>, >>> relay=mail.rogermail.lan[192.168.1.10]:25, delay=0.3, >>> delays=0.21/0/0.03/0.06, dsn=5.1.1, status=bounced (host >>> mail.rogermail.lan[192.168.1.10] said: 550 5.1.1 >>> <s...@mail.rogermail.lan>: Recipient address rejected: User >>> unknown in virtual mailbox table (in reply to RCPT TO command)) >> >> This is the error you need to get rid of. The spamtrap user must be >> accepted for the downstream listener to use it to trigger the DISCARD. >> >> If you add the spam@ user to your local storage all will be well. > > Aha, I think ... > > So IIUC, I need to RE-add the list of spamtraps in the has table as > valid users (removed as valid after known to be compromised ...) so > that, rather than being off-hand rejected as an unkonw/non-existent > user@domain, the multiple-recipient message gets carried through far > enought to get to where that DISCARD check+action is valid? > > Is that approaching correct?
Yes, that is the intent. The recipient must be accepted to get to the post-filter DISCARD. > > If so, then I assume that I leave only postscreen listening at the > 'real' external address, and the spamtrap address check shoud remaing > associated with the 127.0.0.1:10026 reinjection listener, as you'd > previously advised? Yes, the check for the spamtrap recipient must be in the 127.0.0.1:10026 listener for DISCARD to work. -- Noel Jones
