Mailing List Archive

remote_max_parallel logic
I assumed that for a long recipient list involving multiple target
hosts (comming from dozens of RCPT TO commands in the SMTP dialog) are
split and distributed across multiple delivery processes. However,
that doesn't seem to be the case, there are only extremely few
delivery processes. Or are MX lookups needed during the split and
stalling delivery?

--
Florian Weimer Weimer@CERT.Uni-Stuttgart.DE
University of Stuttgart http://CERT.Uni-Stuttgart.DE/people/fw/
RUS-CERT +49-711-685-5973/fax +49-711-685-5898
Re: remote_max_parallel logic [ In reply to ]
On Mon, 27 May 2002, Florian Weimer wrote:

> I assumed that for a long recipient list involving multiple target
> hosts (comming from dozens of RCPT TO commands in the SMTP dialog) are
> split and distributed across multiple delivery processes.

No, it doesn't work like that.

> However,
> that doesn't seem to be the case, there are only extremely few
> delivery processes. Or are MX lookups needed during the split and
> stalling delivery?

All the routing is done before any deliveries. Routing is done serially.
Remember that routing *might* cause rewriting (e.g. if an abbreviated
domain gets expanded). In theory, no deliveries should be done until all
the routing is complete, so that all such rewriting has happened. In
practice, Exim does not hold up delivery if it cannot complete the
routing of one address.

It is only during remote delivery that one message gets handled by
multiple processes.



--
Philip Hazel University of Cambridge Computing Service,
ph10@cus.cam.ac.uk Cambridge, England. Phone: +44 1223 334714.