> But if you intend to use Spews as a blocking list then you (and your
> management, and your users) need to understand the consequences, and
> to support the resulting actions. No-one has an unalienable right to
> deposit mail in your server.
Where I work, upper management wanted spews removed because it (after a year
of usage) blocked 1 person.
> 2. the postmaster address is not blocked (except in the worst of
> cases, where we won't even accept SMTP calls from the offenders) (and
> anyone who is dumb enough to spam the postmaster address goes straight
> into the blocking list without appeal, of course)
Not a bad idea, IMO. But as a a single user with a single mailserver, I am
everyone thus, spam me, there's no appeal to email me on my server anymore.
(I have an ACL setup to blacklist any IP which attempts to relay through me
and there's no appeal for that either)
> On the whole, this strategy works. It sometimes causes awkwardness
> with mailing lists to which the recipient has consented, when a
> blacklisted spammer distributes mail to the list.
I hate when that happens. lkml comes to mind, but that doesn't happen
often.
So, what do you think about making auto whitelists?
I setup a router (basically because I inject messages using sendmail -oi
-oem from mutt and I want my outbound to be whitelisted).
Something like this:
whitelister:
driver = accept
senders = *@+local_domains
domains = !+local_domains
condition = ${lookup mysql{insert into ...}{no}{no}}
transport = <some random transport>
According to philip, the senders line above works. I didn't know about this
and patched exim to have sender_domains in routers (like acls)
--
Lab tests show that use of micro$oft causes cancer in lab animals
