Re: [Exim] Port number / RBL not dynamic fields?

Pàgina inicial
Delete this message
Reply to this message
Autor: exim
Data:  
A: exim-users
Assumpte: Re: [Exim] Port number / RBL not dynamic fields?
>
> > i.e.: the port number of the 'smtp' transport (port = <portnumber>)
>
> That's on the wish list, but it isn't as easy as it appears.

Currently,
> when an SMTP connection fails, the retry information is keyed only IP
> address (assuming a fixed port number). The first change will have to

be
> to add the port number to that key. Then we can allow port numbers to
> vary each time the transport is called. I was close to doing this, but
> have been overtaken by other events which have pushed it down the

stack
> again. :-(
>
> [The mistake I made in the beginning was not to realize that lots of
> different ports would be used. I should have always used address+port
> combinations in all places.]


Ok, no problem. Take your time.

Maybe others have the same 'problem', so our workaround:

Create an extra router AND transport. The router is an DNS-lookup that
only 'runs' with domains which are in an special table (MySQL database).
It uses a standalone transport with a *fixed* port number. So we have 2
remote_smtp transports.

We use this extra router (and transport) to relay to other mailserver on
a different port than 25. In The Nederlands some providers decided to
close port 25, so their customers can't run a mail server at home. With
our relaying we can still deliver mail to them, so they don't have to
POP their mail.

It's no problem, but it was nice if we could use an extra column in our
database which described the port to use. Untill the time you integrated
that, we use our 'solution'.

>
> > i.e.: the list of RBL servers you can use (deny rbllist = <rbl's>)
>
> That *is* expanded. All arguments of conditions and modifiers are
> expanded. At least, that's what the manual says. I'd better
> doublecheck... goes and reads code ... well, the code seems to think
> it is expanded.


Damn, you have certainly right. I'd created a table with multiple
records (one record for each RBL server). It failed, because it only
used the first record and the others didn't work.

The solution is simple: 1 record with all RBL servers separated (with a
colon).

>
> > Reason: my configuration uses a lot of information which is

retrieved
> > from a MySQL database. It works great, especially with a hand made
> > webfrontend. Everything is dynamic and on the fly changeable. When I

can
> > change i.e. the list of active RBL servers, this would be great and
> > quick, but it doesn't seem to work....
>
> Please test with -d and send me evidence of failure.


See above. I can't send you the evidence because maybe I haven't read
the FM enough =)
(maybe I had a bad day, because I have read almost the complete PDF,
FAQ's and examples....)

Thank you for your answers and hopefully others can use the information
above to make some nice configurations to.


Michael
The Netherlands