[Pkg-exim4-users] selecting smarthost based on sender address

Marc Haber mh+pkg-exim4-users at zugschlus.de
Mon Jun 1 15:10:41 UTC 2015


On Sun, May 31, 2015 at 07:39:10PM +0200, J G Miller wrote:
> At 18:56h, on Sunday, May 31, 2015,
> in message <20150531165607.GU9568 at torres.zugschlus.de>,
> on the subject of "Re: [Pkg-exim4-users] selecting smarthost based on sender address",
> Marc Haber wrote -
> 
>  > I don't think so. It's fine to simplify things if you know what you're
>  > doing.
> 
> Okay then ;+)
> 
> I do sympathize with the maintainer

You can look in the changelog for my name ;-)

>  > I would, however, suggest only simplifying files that need
>  > local changes. That will greatly ease updates and reduce the number of
>  > dpkg-conffile questions asked during updates.
> 
> Indeed so.  What I do with all the config files (almost all from /etc) is
> to create a tree /usr/usr/local/archive/debian/etc and copy all of the original
> files which I change there for future reference.
> 
> When an update comes around, I can then do a diff (even fldiff for an
> easy visual comparison) to see if the file has altered and if so what changes
> I need to make to my customized version, and then copy the new official updated
> file (dpkg-new) if different to the archive for the next time.

Doesn't the .dpkg-{new|dist|old} mechanism already do that?

Actually, this is once again a point where ucf outperforms plain dpkg,
since ucf brings around a convenience copy of the package's file. We
should have ditched dpkg's conffile handling in favor of ucf long time
ago.

>  > Unfortunately, exim -bV does not check for such errors, so we cannot
>  > flag them at daemon start or configuration build time.
> 
> Well yes, obviously one would have to write a configuration file syntax
> checker which would be a an awful lot of work with a rather low value
> to work hours ratio.

Thinking about it, one could probably do exim -bt to find out whether
a message would be deliverable.

On a satellite system:
[3/503]mh at swivel:~$ sudo exim -bt foo at bar.baz
R: smarthost for foo at bar.baz
foo at bar.baz
  router = smarthost, transport = remote_smtp_smarthost
  host smtp.zugschlus.de [2a01:238:4071:3201::1]  port=587
  host smtp.zugschlus.de [85.214.131.164]         port=587
[4/504]mh at swivel:~$ echo $?
0
[5/505]mh at swivel:~$

On a satellite system with deliberately broken smarthost setting:
[9/509]mh at swivel:~$ sudo exim -bt foo at bar.baz
R: smarthost for foo at bar.baz
foo at bar.baz cannot be resolved at this time: host lookup for empty did not complete (DNS timeout?)
[10/510]mh at swivel:~$ echo $?
1
[11/511]mh at swivel:~$


On an Internet system:
[2/501]mh at torres:~$ sudo exim -bt foo at bar.baz
R: domain_literal for foo at bar.baz
R: dnslookup for foo at bar.baz
foo at bar.baz is undeliverable: Unrouteable address
[3/502]mh at torres:~$ echo $?
2
[4/503]mh at torres:~$ sudo exim -bt submit at bugs.debian.org
R: domain_literal for submit at bugs.debian.org
R: dnslookup for submit at bugs.debian.org
submit at bugs.debian.org
  router = dnslookup, transport = remote_smtp
  host buxtehude.debian.org [140.211.166.26] MX=10
[5/504]mh at torres:~$ echo $?
0
[6/505]mh at torres:~$

So this issue is actually solvable. I'll open a bug report to that
regard.

Greetings
Marc

-- 
-----------------------------------------------------------------------------
Marc Haber         | "I don't trust Computers. They | Mailadresse im Header
Leimen, Germany    |  lose things."    Winona Ryder | Fon: *49 6224 1600402
Nordisch by Nature |  How to make an American Quilt | Fax: *49 6224 1600421



More information about the Pkg-exim4-users mailing list