>>>>> "km" == Karl Morgan <ksm at dogbrain.com> writes:

km> Nullclient is a very useful configuration. But I think what bob is
km> asking is if there is a way to change the sendmail behaviour
km> regarding how a Received header is inserted into the message for
km> each hop that an email might take.

You can change the way that a sendmail process will insert a
"Received" header.  In your M4 master config file, you could use
something like, for instance:

define(`confRECEIVED_HEADER', `$?sfrom $s $.$?_($?s$|from $.$_)
           $.by $j ($v/$Z)$?r with $r$. id $i
           SMTP "MAIL FROM" = $f$?u
           SMTP "RCPT TO" = $?s$u$|{local origin}$.$|$.
           at $b')

You *are* using M4 to maintain your sendmail config file, aren't you?
:-)

NOTE: altering the "Received" header may put your MTA out of
compliance with the RFC 821.  Beware of email cops.

km> Other posters have asked this on comp.mail.sendmail and the answer
km> has been that you would need to modify source code to do
km> this. Realize that you could modify the received header in the M4,
km> or resulting cf file but then you would most likely lose real (and
km> perhaps valuable) header info by doing that. The inability to
km> trace a message because you decided not to include host, ident,
km> time info in an email header etc.

Note that RFC 1123 (section 5.2.8) prohibits anyone from mucking with
a "Received" header that "... was previously added to the message
header."

If you do muck with the confRECEIVED_HEADER variable, it's unwise to
rip out so much info that you can't track what happened to a message.
If you need to explain to your boss, your CEO, or the FBI how a
message passed through your system, you probably ought to be able to
do explain.

If you're in a "to hell with the RFCs, the IETF, law enforcement, and
my boss" kind of mood ... you can use sendmail's "milter", the message
filter, to modify messages on the fly, including the headers, AFAIK.
You could even apply this filter selectively (e.g. only to messages
leaving the company) and have much more flexibility in how sensitive
information could be obfuscated.

-Scott