Sean Wilkerson wrote, on 16. mar 2007 15:45:
I have looked through the list-archives and although I have seen several
threads on the following issue, I don't feel like they were fully
concluded, so would like to get an opinion now if possible.
Dspam as a spam filter is working great, and we are seeing good results in
our testing. However, we are struggling a bit with getting the accounts
to work the way we want them to.
My goal is to avoid a massive amount of unnecessary users from being
created by spammers who mail to hundreds of made-up [EMAIL PROTECTED]
1: You mention later in your post that you're using Postfix (what
version you don't mention). If you're using Postfix, any version apart
from prehistoric vmailer, there is absolutely no reason why your MTA
should ever accept mail to *any* "made-up [EMAIL PROTECTED]":
local_recipient_maps = whatever
or
virtual addresses in ADDRESS_CLASS_README.html
smtpd_recipient_restrictions =
<load of stuff>
reject_unlisted_recipient
<more stuff>
I also
want to maintain the ability of users to create special addresses for
individual online accounts, such as [EMAIL PROTECTED], but have those map
back to the user in dspam so only a single account is necessary for each
user.
2: My sites don't let the users themselves do that, mailadmin does it
for them, but it would be possible using a CGI interface to LDAP. All
sites use LDAP as backend for users and it's possible to give an LDAP
user with a mail attribute as many mail aliases as one wants. Obviously
all go through dspam ... ("obviously?") and get mapped to a single UID.
Fix your initial premise in 1: above (don't accept mail for non-existent
users) and reformulate your aim ;)
Best,
--Tonni
--
Tony Earnshaw
Email: tonni at hetnet dot nl
