On 29 Dec 1998, Rask Ingemann Lambertsen wrote:
> I'm currently logging (because of
> <URL:http://www.gbar.dtu.dk/~c948374/qmail/qmail-antispam4.diff>) ...
I generally log the SMTP command transactions, and any failure on their
part (ie. what the remote end sees, I see too, except for successes), the
recorded DATA receipt (which helps track down the transaction to
subsequent deliveries), and the Message-ID (for Support's benefit for when
people phone up about a message - they're much more likely to know their
message's id than some obscure DATA response which hardly any submission
agent (qmail-inject included) will tell you anyway!).
I did also add some code to record the entire message transaction too with
the appropriate flag set, which we used on occasion with some customers to
try and find out which messages of theirs were being resent continuously,
etc. This was before I became aware of Dan's recordio thingummy, but then
I prefer something that can be turned on per connection origin (or indeed
mailfrom/rcptto combination in our case).
Yes, there are tools to plug in to perform this task, but at the end of
the day qmail-smtpd will log what I want it to log in the way I want it
when I patch it directly myself, rather than try to interpret output from
another process (smtpd knows whether MAIL FROM:<> is the envelope or the
message text, the external process would need to work it out!).
IMHO!
C.
