Running Exim 3.22 and fetchmail 5.8.3 on a Progeny Debian Linux 1.0
system (which is pretty close to Debian 2.2). Yesterday, I changed Exim
to listen to port 26 instead of port 25, since my ISP now seems to block
port 25 for me (thanks, guys!).
Alas, I forgot that fetchmail was running in the background to fetch
mail from my ISP address. It was trying, now unsuccessfully, to connect
to port 25 every 5 minutes or so. So this morning, I killed that poor
fetchmail process, and added "smtphost localhost/26" to my .fetchmailrc.
According to the fetchmail man page, that's how you make it connect to a
different host or port (the default, quite sensibly, is localhost/25 --
not sure why they use / instead of : for the port delimiter, but that's
how it's documented).
Luckily, I ran "fetchmail -Nv" so I could see exactly what happened.
What happened is not good. Here's an extract from syslog (with -v,
fetchmail logs POP and SMTP sessions):
POP3> TOP 9 99999999
POP3< +OK
reading message 9 of 13 (9409 octets)
SMTP> MAIL FROM:<python-dev-admin@???> SIZE=9409
SMTP< 250 <python-dev-admin@???> is syntactically correct
SMTP> RCPT TO:<greg@localhost/26>
SMTP< 501 <greg@localhost/26>: malformed address: /26> may not follow <greg@localhost
SMTP> RSET
SMTP< 250 Reset OK
flushed
POP3> DELE 9
I can see at least two fetchmail bugs here: it gets the local address
wrong, and it (gasp!) deletes the message from the POP server after a
local SMTP failure. Yikes! It appears that all my mail from the last
12 hours has been irretrievably lost.
The Exim problem is that there is no sign, either in rejectlog or
mainlog, that this connection (13 of them actually, for the 13 lost
messages) ever occurred. Grumble.
Anyone have any idea what went wrong here?
Greg
--
Greg Ward - just another /P(erl|ython)/ hacker gward@???
http://starship.python.net/~gward/
I just read that 50% of the population has below median IQ!
