Author: W B Hacker Date: To: exim users Subject: Re: [exim] Ha: Re: How many times does router run for multiple
recipients?
vitas1@??? wrote: > wbh@??? wrote on 06.10.2008 19:12:04:
>
> ...
>
>> Unless / until you turn up your logging detail or add some tracking
>> tools (see also 'debug_print') you can only be sure what took place with
>
>> those routers that do what was expected of them and leave footprints in
>> the logs. Or somewhere else recoverable/visible.
>>
>> So the 'NOT offered' may or may not be correct, and may or may not be
>> for the reason you believe it to be such.
>>
>> Bill
>
> Well I've looked at the source code. Looks like Exim does the following.
>
> Assume a message has 2 recipients at the same domain. First of them *is
> handled* by some router
> that has a "same_domain_copy_routing" option set.
>
> Exim "marks" processed address and then checks if there are more
> recipients at this domain.
> If they are then they will be routed with *the same* router as the first
> recipient. In this case Exim
> *never* returns to routers list (maybe "unseen" option could help - I
> don't check).
Well.. (I haven't looked at the code), nor used this combination, but..
.. to the extent there are multiple recipients for a given message who
are in the same domain, you've effectively asked Exim to scan the
recipients list and handle any further ones for that domain at the same
time.
.. it does so, hence has no need to re-enter the routers. Message &
carbons have been onpassed to a transport (and hopefully, delivered..)
>
> Well it is not a bug - but someone *should be careful* with this option if
> he tries to use a router with per-recipient condition...
>
> Andrew, Bill - thanks again for help.
>
>
> Vitas.
'Shoudl be careful* (with routers in generall...)
In a manner of speaking, all routers are 'per recipient'.
This is most apparent for those applied to local delivery where the
recipient-specific location in the mailstore will ordinarily be unique
to a given recipient.
So I take your point another way - that a router that attempts to
*group* or consolidate recipients, by domain or otherwise, thereby not
further testing each one on its own... is the type that needs more care
and testing.
