On 2017-03-11 16:51, Jeremy Harris wrote:
> If I recall correctly that's a massively helpful error coming from
> GnuTLS - is that what you run?
>
> For workaround you could avoid advertising STARTTLS to that IP.
>
> Or, to investigate, set up for a packet capture of connections
> from that IP - and ask the gnutls-help mailinglist. They'll
> want to know what library version you're running;
> "exim -d -bV" should tell you.
It turns out the TLS messages just mask a different underlying problem.
With tls_advertise_hosts set to exclude this problematic host, the log
changes to:
2017-03-11 23:27:59 SMTP connection from [12.34.56.78]:55572 (TCP/IP connection count = 1)
2017-03-11 23:27:59 acl_check_mail: mail from hol-info-bounces@???
2017-03-11 23:30:57 SMTP connection from disorder.example.com (acedia.example.com) [12.34.56.78]:55572 lost while reading message data (after header)
Any sudden enlightement or should I reach for tcpdump?
I do some expansions with $bh_ and friends. Should I be worried that my
broken config code causes this? In particular: does the log above mean
the error happened _just after_ reading the headers, or simply that
reading headers was successful and the drop happened at any point while
reading the body?
--
Please *no* private Cc: on mailing lists and newsgroups
Personal signed mail: please _encrypt_ and sign
Don't clear-text sign:
http://cr.yp.to/smtp/8bitmime.html
