[Nottingham] Bizzare fetchmail problem.

Andy Davidson nottingham at mailman.lug.org.uk
Thu Nov 21 14:41:01 2002

Hi, guys --

I've been silent recently; feel cheeky asking for help, but I've got into=
confuzzled mess with fetchmail, and I don't see what's happening.

Assume the following mail setup :

A company has foo.com, and mail addressed to users of this domain name is=
delivered to their primary MX as normal.  However, some customers still w=
to the legacy address@foo.myisp.com addresses, which ends up in a multidr=
pop3 mail spool on pop.myisp.com.

A man is having a bad Thursday, trying to migrate from the old Mailtraq s=
to foo's new exim server.  He wants to use fetchmail on the mail server, =
collect email from the pop3 mailbox at myisp.com and push it to the smtp=20
server on "localhost" for processing.  The 'username' part of the email=20
address should stay the same, so that it's routed to the correct mail spo=
on the new mail server.  The domain part can be 'foo.myisp.com', or 'foo.=
or 'localhost', the mail server is told to accept mail for all three.

However, with the following .fetchmailrc :

kylie:~# cat .fetchmailrc
poll pop.myisp.com no dns localdomains foo.myisp.com:
   protocol pop3 user foouser with pass foopass to * here

*all* mail gets sent to the MTA on localhost addressed to=20

man pages to the rescue -- suggesting "no rewrite" and "no envelope." =20
However, when *both* are tried the *same thing* happens.

fetchmail logs onto the pop3 server, grabs the mail, then connects to the=
server with the following conversation :

fetchmail: SMTP> MAIL FROM:<r@r.r> SIZE=3D18344
fetchmail: SMTP< 250 <r@r.r> is syntactically correct
fetchmail: SMTP> RCPT TO:<postmaster@localhost>
fetchmail: SMTP< 250 <postmaster@localhost> verified
fetchmail: SMTP> DATA

I'm stumped.  I've had it working before (a looong time ago), but can't g=
et it=20
going now.

Version numbers, for interest :

kylie:~# fetchmail --version
This is fetchmail release 5.9.11+NTLM+SDPS+SSL+NLS

kylie:~# exim -bV
Exim version 3.35 #1 built 15-Apr-2002 20:48:24
Copyright (c) University of Cambridge 2001

Regards, Andy Davidson, Systems Administrator
Comodo Research Labs,            01274 730505