Re: [exim-dev] Bug? Exim rewrites domain in response to CNAM…

トップ ページ
このメッセージを削除
このメッセージに返信
著者: Max Bowsher
日付:  
To: Graeme Fowler
CC: exim-dev@exim.org
題目: Re: [exim-dev] Bug? Exim rewrites domain in response to CNAME, in contradiction to what is written in Chapter 31
On 06/09/13 15:43, Graeme Fowler wrote:
> On 6 Sep 2013, at 15:40, Max Bowsher <_@maxb.eu> wrote:
>> Thanks for responding, but you appear to have ignored most of what I
>> wrote in my email, and missed the problem that I am describing entirely.
>
> You erroneously used the word "reroute" in your original email.


I called it that because that's the word the Exim source code uses to
describe what it is doing. If you look in src/routers/dnslookup.c in the
block which calls rf_change_domain(...), both comments and result
constants use the word "rerouted".

>> If you re-read my original email you will discover that I'm not saying
>> anything about the destination server being wrong at all - rather I'm
>> saying that Exim is rewriting the address incorrectly.
>
> Config, please. Without that, no help can be forthcoming.
>
> I would surmise that you're in submission mode without an appropriate fixup - or you simply have a duff rewrite section.


I understand that config is key to many problems, but in this case, the
only relevant point is having a dnslookup router configured in the
approximate manner than any internet-connected host does.

I can also understand that you might be reluctant to just take my word
for it, so I ask you to perform the following trivial test:

The problem can be reproduced just using the "exim -bt" address testing
mode. Please run "exim -bt nonesuch@???". The
behaviour I have described results in Exim clearly showing the address
rewrite as follows:

nonesuch@www.bcidahofoundation.com
    <-- nonesuch@???
  router = dnslookup, transport = remote_smtp
  host mail2.bcidaho.com [207.170.246.155] MX=10
  host mail1.bcidaho.com [207.170.246.146] MX=10


You can see that the domain has incorrectly been changed to
@www.bcidahofoundation.com - the correct behaviour would not include any
"<--" address change lines.

Regards,
Max.