[Mailman-Users] SMTP error handling? Something that should be in the FAQ...

[J C Lawrence]

What happens when a post is broadcast by MailMan but the MTA refuses
to accept the connections?

MailMail runs on the same machine which accepts all incoming mail for
the domain.  The MTA on that machine is configured to throttle system
load by limiting the total number of inbound connections and (if
system load is high enough) refusing all connections to port 25 carte
blanche until system load falls below X (there are secondary MX'es to
take the load).

This morning, due to extended downtime from brownouts, when the system
came back up and ETRN'ed my secondary MX'es I got several hundred
messages within a few minutes.  A fair percentage of those were posts
to lists with reasonable numbers of subscribers (500 - 1000).  As a
result I got entries in my exim logs like:

--<cut>--
1999-03-13 03:58:11 Connection from 127.0.0.1 refused: load average = 26.61
--<cut>--

and in ~mailman/logs/smtp-failures for the same time (and I suspect
the same message) I have log entries ala:

--<cut>--
Mar 13 03:58:11 1999 TrySMTPDelivery: To ['coder at kanga.nu']:
Mar 13 03:58:11 1999 TrySMTPDelivery:    Mailman.pythonlib.smtplib.SMTPServerDis
connected  / Server not connected  (dequeued)
--<cut>--

Does Mailman try and re-sub that message later?  Manually running
run_queue did not seem to reveal any queued messages.

-- 
J C Lawrence                              Internet: claw at kanga.nu
----------(*)                            Internet: coder at kanga.nu
...Honorary Member of Clan McFud -- Teamer's Avenging Monolith...