AAAARRRRGGGHHH!!!!!
On 8 Aug 2007, at 11:42, Graeme Fowler wrote:
> Can you change the logic in your first router to:
>
> send_local_origin_to_smart_host:
> driver = manualroute
> route_list = * mta.york.ac.uk
> condition = ${if eq{$sender_host_address}{}{yes}{no}}
> transport = remote_smtp
>
> The way you have it seems to want to generate a null route_data
> value if
> the message is non-local, which looks wrong to me (but might not be!).
>
> The way above should only trigger the router if the message is locally
> submitted.
Many thanks for the suggestion Graeme, but it wasn't that...
According to the book if route_data's value is an empty string then
the router declines. I assume this is so you don't have to jump
through hoops when using it to look entries up in a database and find
no entry there.
The scream of pain at the start is because I've just discovered what
the cause was: the hints database!
When a message can't be delivered because the recipient's mailbox is
full an entry gets added to the hints database. When a subsequent
message arrives delivery is attempted and, when it fails, Exim
appears to consult the hints database and then decides it should
immediately REJECT (as against defer) the incoming message.
If I sneakily delete the hints database and then send another test
message it gets deferred and held in the queue as I expect.
I assume this is to try and keep the size of the queue small when a
recipient is known to have a full mailbox and messages keep pouring in.
[Aside: Thank heavens for the Exim course the other week... it was
remembering Philip explaining about the hints database and its use
that gave me the clue!]
But many thanks again for the suggestion!
Cheers,
Mike B-)
--
The Computing Service, University of York, Heslington, York Yo10 5DD, UK
Tel:+44-1904-433811 FAX:+44-1904-433740
* Unsolicited commercial e-mail is NOT welcome at this e-mail address. *
