On 4 Jun Matthias Andree wrote:
To me, it looks like fetchmail intersperses the output of two separate SMTP
sessions it keeps:
reporting an obscure bug with Postfix that can be worked around by
configuring Postfix to "smtpd_delay_reject = no".
Matthias's bug report has been sitting in my mailbox for a month now,
making my brain hurt whenever I looked at it and tried to grok what
was actually going on there.
But I think Paul Howarth has now dispelled the mystery. Postfix
behaves by default roughly in the way fetchmail behaves with
FEATURE(delay_checks, friend). In both cases, we need to handle bad
SMTP listeners to RCPT TO without resetting the connection.
Matthias? Paul? Any comment?
--
>>esr>>