Patrick von der Hagen wrote:
> The exchange-log says:
> 12:47:25 15 - 220 smtp-greeting
> 12:47:25 15 EHLO Exchange
> 12:47:25 15 - 250-smtp
> 12:47:25 15 MAIL FROM:<somone@somewhere> SIZE=1934
> 12:47:25 15 - 250 OK
> 12:47:25 15 RCPT TO:<destination@???>
> 12:47:25 15 - 250 Accepted
> 12:47:25 15 DATA -
> 12:47:25 15 - 354 Enter message
> 12:47:25 640 - 250 OK id=1DM1sb-0001i6-Ae
> 12:47:25 703 RSET -
> 12:47:25 703 - 250 Reset OK
> 12:47:25 703 MAIL FROM:<someone.else@somwhere> SIZE=6211
> 12:47:25 703 - 250 OK
> 12:47:25 703 RCPT TO:<one@???>
> 12:47:25 703 RCPT TO:<two@???>
> 12:47:25 703 RCPT TO:<three@???>
> 12:47:25 703 RCPT TO:<four@???>
> 12:47:25 703 - 250 Accepted
> 12:47:25 703 DATA -
> 12:47:25 703 - 354 Enter message
> 12:52:25 300703 - 421 smtp
> 12:52:25 300718 QUIT -
>
> (first colum timestamp, second column seems to be ms since start of
> connection, third column the commands sent by exchange, 4th column the
> replies).
>
> The second message is probably transfered completely, but either
> exchange fails to terminate it correctly or Exim fails to react to the
> completion... The exchange-log won't tell us.
>
> However, there is still an exim logfile:
> 2005-04-14 12:47:25 1DM1sb-0001i6-Ae <= someone@somewhere
> H=rzms-ex3.rz.uni-karlsruhe.de [172.21.63.166] P=esmtp S=2074
> id=F540278AB4DF944D9263CD0F2D9EE13A0BD7FF@???
> from <Andre.Richter@???> for destination@???
> 2005-04-14 12:47:27 1DM1sb-0001i6-Ae fastreklama.ru [216.74.90.3]:
> Connection reset by peer
> 2005-04-14 12:47:28 1DM1sb-0001i6-Ae == destination@???
> R=domain_dns T=smtp_out defer (104): Connection reset by peer:
> fastreklama.ru [216.74.90.3]
>
> I don't find ANY reference for the second email in my exim-logs. I did
> not try to run a tcpdump (yet), because there is quite some traffic on
> my smarthost, but since exim does not write any indication to its
> logfiles, that it as ever seen a message or that a connection has been
> reset or a connection suffered a timeout, I suspect there might be a bug
> in exim somehow concerning the RSET.
It certainly never *accepted* the second mail, hence the lack of
a "<=" line for it.
Ideas:
- add smtp_protocol_error and smtp_syntax_error to your log_selector
- can you get a packet-level trace of such a conversation? Tcpdump
lets you filter by IP.
- is pipelineing offered by exim? Try turning it off
(pipelining_advertise_hosts)
- can you run a conversation like this into exim in debug-mode,
and capture the console output?
- Jeremy