At 12:30 AM 20/12/2000 -0800, Rob Tanner wrote:
>Probably not a good idea since sending to postmaster is a common practice
>when some experiences problems mailing to a user (I don't get many, but I
>do get maybe one or two a month). Postmaster is typically aliases to
>root, and is a required alias. You could try aliasing it so that delivery
>is done to /dev/null -- but again, I don't recommend it. Also, since
>postmaster is a required alias it might not work anyway.
Having the postmaster alias is not the problem. The problem is getting
these error emails coming up about 8 a week or so.
I just got another one and what is actually going on has just clicked into
place.
1) Spammer send spam to invalid address on our mail domain.
2) Mail system sends rejection message because of said invalid address.
3) However, spammer has spoofed their origin address, so the return address
is also an invalid email address, usually not even a valid host or domain, so
4) I (as postmaster) get an error message saying my system couldn't return
an invalid mailbox message to some bad address spoofed from some scum of
the earth spammer.
5) Being the diligent system administrator, I have to actually check though
the message to make sure it isn't anything more pertinent then yet another
bounced spam error message.
I think I'm getting spammed by osmosis.
This email address hasn't been valid for at least the two years I've worked
here. Can anybody explain why I've started getting these messages in the
last month since I switched to a cyrus server?
I believe that I can coerce sendmail into rejecting emails to certain
addresses at the time the remote server tries to send it to us (ie before
it even gets to deliver). So I can probably block that specific addresses,
which appears to account for the bulk of the dodgy emails. That's one way
around it, and given that it's only been about three addresses so far that
are causing issues, its probably a reasonable solution.
Curious as to why Cyrus/deliver does this were sendmail to a more standard
mailbox (via procmail presumably) doesn't, and if this highlights a
bug^H^H^H^H unexpected feature in Cyrus.
E.
