On Thu, 25 Jun 1998, Jeffrey Goldberg wrote:
> Sounds good, but I am a bit confused by remote_max_parallel.
>
> It sets the maximum number of simultaneous delivery processes as I see
> it for a particular source message. That is, if I set it to 5 then
> I will get list deliveries broken up into at most 5 "sessions"
> irrespective of the size of the list. So, if I have a list with
> 10 addresses I will get 5 processes with 2 addresses each and if I
> have a list of 1000 I will get 5 processes with 200 addresses each.
It doesn't work quite like that. The main process picks off all the
addresses that are to be sent to the same host (assuming you have not
set max_rcpt in the smtp transport, or if you have, up to that many at
once). Then it fires off a process to do that delivery. (Strictly, it
will do this only if there are other addresses left. No point in firing
off a process if it's the only one.) Then it picks off the next set of
addresses and fires off another process. When it has fired off 5
processes, it waits till one of them finishes before firing off the next
one.
So if your list of 10 addresses are all to go to different hosts, it
will ultimately run 10 processes, but no more than 5 at a time.
> I would much prefer what bulk_mailer does, and sets the maximum number
> of deliveries per process and will fork another process only if the list
> contains more addresses then the max per process.
I'm afraid Exim doesn't work like that for single messages. If you want
that behaviour, then you should leave remote_max_parallel unset, and
arrange to submit separate messages to Exim, each with your maximum
number of addresses. (I accept that this costs disk space, because there
are multiple copies of the message.)
--
Philip Hazel University Computing Service,
P.Hazel@??? New Museums Site, Cambridge CB2 3QG,
ph10@??? (sic) England. Phone: +44 1223 334714
--
*** Exim information can be found at
http://www.exim.org/ ***
