Looking up email addresses (was: Identifying spamhosts)
My domain names are being flooded with bounces these days. After
talking with others it seems to be their largest source of bad mail,
I think what's happening is a virus is attached to email that turns a
machine into a mail proxy. Then mail is sent out from that machine, but
with a from address as a stolen domain name (<random name>@hank.org).
Some of that spam goes to bad addresses and then bounces back to, eh,
A few weeks ago I sat down for a couple of hours and whet through the
bounces and (ignoring biggies like AOL) entered the RFC822 attached
messages into Spamcop to get a email address (based on the Received:
headers in the original message).
I then sent mail to about 20 of these people explaining that their
machine was being used as a proxy to send spam.
I was amazed that I got back somewhere between 10 and 15 messages from
real people saying thanks and the machine was fixed. I assumed my
messages would be ignored, but most mail/sys admins seem interested to
know they have a broken machine.
I asked Spamcop author if any of their code was available to do the
lookups locally (when the bounces arrive at my machine), but, no, that's
only available via their interface.
I'd like to see an open source tool like Spamcop for looking up an email
address to contact the admin in charge of the netblock where the spam
came from based on a list of Received headers.
In many cases I'm able to track down addresses manually, but I have not
attempted to automate it yet because of time (and skill/experience in
determining the correct addresses). I also believe spamcop has a
database of email address which would be needed for such a system -- I
assume you can't always rely on the host record to give the best or
complete contact information.
Hard to know if mail admins would appreciate getting automated mail of
this type, though. This is probably "virus->send spam->go on to next
host". So by the time the admin gets the notification and tracks down
the problem it may be too late to make a difference. Probably more
effective to get people to stop using systems that are so easily