Re: [exim] Usage of dnsdb primary mx

Top Page
Delete this message
Reply to this message
Author: W B Hacker
Date:  
To: exim users
Subject: Re: [exim] Usage of dnsdb primary mx
Marc Perkel wrote:
>
> Phil Pennock wrote:
>> On 2009-05-26 at 15:08 -0700, Phil Pennock wrote:
>>
>>> On 2009-05-25 at 11:47 +0100, Mike Cardwell wrote:
>>>
>>>> Thomas Klaver wrote:
>>>>
>>>>
>>>>> Is it possible to retrieve the highest prioritized MX record using a
>>>>> dnsdb lookup? If so, how?
>>>>>
>>>> The results returned by dnsdb when doing an mx or mxh lookup are not
>>>> ordered by priority:
>>>>
>>>> root@haven:~# exim4 -be '${lookup dnsdb{mx=googlemail.com}}'
>>>> 30 alt3.gmail-smtp-in.l.google.com
>>>> 40 alt4.gmail-smtp-in.l.google.com
>>>> 5 gmail-smtp-in.l.google.com
>>>> 10 alt1.gmail-smtp-in.l.google.com
>>>> 20 alt2.gmail-smtp-in.l.google.com
>>>> root@haven:~#
>>>>
>>>> I'm not sure how you could pick out the item with the lowest priority
>>>> using Exim expansions.
>>>>
>>> ${reduce{<\n ${lookup dnsdb{mx=googlemail.com}}}{65535 .}{${if <{${extract{1}{ }{$item}}}{${extract{1}{ }{$value}}}{$item}{$value}}}}
>>>
>> Note that this will reject a legitimate singleton MX at priority 65535.
>> It's better to just use a higher-than-16bit-unsigned start value. Eg,
>> 70000. It will also result in "65535 ." being returned for the NOERROR
>> empty response case; eg, a lookup where there's an A record but no MX.
>>
>> So the fix is to use 70000 as the initial seed and then check for
>> "70000 ." for errors. In an ACL, you might have:
>>
>> warn set acl_m_lowest_mx = ${reduce{<\n ${lookup dnsdb{mx=$domain}}}{70000 .}{${if <{${extract{1}{ }{$item}}}{${extract{1}{ }{$value}}}{$item}{$value}}}}
>>
>>   deny condition = ${if eq{$acl_m_lowest_mx}{70000 .}}
>>        message = "No MX records found for $domain"

>>
>> Otherwise, because comparators like eq{} don't store the two tested
>> values in variables for use later in the expansion, you're stuck
>> re-evaluating -- which should be fast, because the results will be
>> cached, but not pretty. Even if we regard my first response as pretty,
>> this one does not qualify.
>>
>> ${if !eq{${reduce{<\n ${lookup dnsdb{mx=$domain}}}{70000 .}{${if <{${extract{1}{ }{$item}}}{${extract{1}{ }{$value}}}{$item}{$value}}}}}{70000 .}{${reduce{<\n ${lookup dnsdb{mx=$domain}}}{70000 .}{${if <{${extract{1}{ }{$item}}}{${extract{1}{ }{$value}}}{$item}{$value}}}}}fail}
>>
>> However, just because it's not pretty, that doesn't mean it doesn't
>> work. Exim provides enough flexibility to let you do whatever you want,
>> rather than constraining you to the solutions which are provided by the
>> more normal convenience features.
>>
>> Provided that you built Exim with dnsdb support, of course.
>>
>> -Phil
>>
>>
>
> My opinion - for what it's worth, is that this functionality should be
> built into Exim as a feature.
>
>


It is.

Been there for years. Used for determining where to send the messages.

You didn't think Exim just sent mail to random IP's did you?

;-)

It is also already 'wired' to elevate the relevant results 'report style' when
flagged as in standalone / debug mode.

Have a look at the last part of ~/src/host.c

For those who don't trust Exim to JFDI, 'twould be child's play to trick that
into doing the same 'on demand'.

Now - there'll be them as wants pony's, perl, dnsdb, and pizza ovens built into
the 'default', so perhaps the 'C' code will be taken out and Exim *will* then
send to random IP's unless one remembers to reinvent that wheel each time it is
installed....

"... and the wisdom to tell the difference."

Bill