On Tue, 9 Nov 2004, John Goerzen wrote:
> So, in the standard situation -- where most messages are successfully
> delivered immediately -- -q and -qq would be functionally equivolent,
> since failures are batched up anyway?
Yes. Though -qq would of course use more resources. "Are batched up" is
not a guarantee. It's a "best efforts / this normally happens" approach.
> That leads me to another question: let's say I have a mailing list
> delivering mail to 1000 people, and a default Exim configuration (send
> immediately). Would these deliveries happen serially -- that is,
> person 900 would have to wait until all the earlier people on the list
> got their mail?
Depends entirely on how you inject the message or messages (more than
100 recipients per message is not recommended, though Exim can handle
thousands). Mailing list handlers like Mailman can be configured to send
multiple messages. If you *do* have 1000 recipients on a single message,
and they are all on other hosts, the number of simultaneous deliveries
depends on the remote_max_parallel setting. Set it to 1 for serial
delivery; set it to 100 for huge parallelism. Local deliveries are done
serially.
> Would a queue runner process that got kicked off later
> be able to deliver the message to some people, even while the intial
> "immediate" delivery process is still running?
No.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
Get the Exim 4 book: http://www.uit.co.uk/exim-book
