It's working!

Vernon Schryver
Tue Jan 8 16:58:18 UTC 2002

> From: That Funky Chick <>

> Woo hoo!  Just checked the rejected mail log from yesterday--DCC bounced 
> about half a dozen that got through SPEWS (and I'm pretty sure that 
> something from an address of "tonsofcash" is spam).

Thats good news.  Note that SPEWS, the RBL, and so forth are good for
detecting spam based on its source, while the DCC is for detecting
spam based on (the number of) its targets.

If you're using SPEWS, it is good to arrange to report SPEWS hits to the
DCC.  That is done by using FEATURE(dccdnsbl) instead of FEATURE(dnsbl).
Copy (or better, sym-link) misc/dccdnsbl.m4 to your sendmail cf/feature
directory and change to .mc file.
If you're interested and have any questions, please let me know.

> So I'm here with more questions:
> Is it possible to have it give a message when it rejects mail, a la the 
> Sendmail 5xx error message?

I don't understand.  All sendmail+dccm hits get either success or
5xx rejections, unless you use DCCOM_ARGS="-a DISCARD" or similar.
To test this, report a test message as spam with `dccproc -t many`.
Then send the same message to yourself from a non-white-listed
SMTP client (it's good to whitelist your own IP addresses)
or wait perhaps 30 seconds and send the same message some other
user of the same DCC network.

The DCC rejections happen in response to the SMTP DATA command,
which some lame and broken SMTP clients do not notice.

> And: if you're using sendmail's "spamfriend" feature, does it skip the DCC 
> check?  (I believe it would, as it does skip any RBL checks.)
> ...

I've throught of writing the ugly hacks that would be needed to
connect sendmail's spamfriend mechanism to the DCC, but keep deciding
it's not needed and so punting.  I think it is not needed because
"spamfriend's" can be listed in the dccm client whitelist.  For example
I use these entries for postmaster:
    ok      env-to  postmaster

Vernon Schryver

More information about the DCC mailing list

Contact by mail or use the form.