On Thu, 4 Jan 2001 brian.wilkinson@??? wrote:
> Could the 'Secondary queue' suggestion be extended to improve Exim's
> handling of mail for dial-up hosts.
Same reaction - I want to keep Exim clean and simple. I don't like
complicated multi-level queueing schemes.
> Currently mail for dial-up domains is stored externally to Exim. When an
> ETRN command is received, a non-Exim delivery mechanism is triggered to
> deliver the mail stored for the specified domain direct to its mail host.
> This avoids a complete queue run every time an ETRN command is received.
> Big disadvantage is that the mail is no longer in the Exim queue and thus
> delivery is totally dependent on the non-Exim delivery mechanism.
There was a posting on this list not very long ago that had a very
cunning way of getting round this. The mail is stored externally, in
BSMTP format, but only after it's been on the queue for a certain time.
(Once a store file exists, new messages go there immediately.) Then the
ETRN triggers a call of exim -bS to read the messages back into Exim,
which tries to deliver, and now succeeds. Incoming mail while the client
is online gets delivered immediately. Thus, no non-Exim delivery
mechanism is required. In effect, you build your own secondary queue.
(You need to build your own mechanism for inspecting such queues.)
I think I saved the posting for adding to the FAQ at the next update. If
you can't find it, I may be able to dig it up.
> If Exim supported multiple secondary queues, mail for dial-up domains could
> be moved to dedicated queues (ie one queue per dial-up domain). When an
> ETRN command is received, a queue run would be performed on the queue for
> the specified domain. The main queue would not be impacted. No external
> storage or delivery mechanisms would be required, and queued mail would
> still be subject to normal Exim retries rather than delivery being totally
> dependent on receiving an ETRN command.
There certainly seems to be a need for an MTA that operates in that way.
Exim is not it, and I don't want to go that way, I'm afraid. (One could
also argue that, in 5 years time, dial-up will be a thing of the past,
and everybody will be "always online", but I don't want to get into a
futurology discussion.)
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
