Re: [exim] local_user section with check_local_user removed …

Top Page
Delete this message
Reply to this message
Author: W B Hacker
Date:  
To: exim users
Subject: Re: [exim] local_user section with check_local_user removed (as advised by most cyrus HOWTOs) causes problems
Luke Kenneth Casson Leighton wrote:
> On Thu, Feb 01, 2007 at 05:55:16PM +0800, W B Hacker wrote:
>> Luke Kenneth Casson Leighton wrote:


*snip*

>> Look in your config for some variation of:
>>
>> deny !verify = recipient
>
> ok - the verifications: are these verifications using the 'router'
> sections?
>


When so used, yes - in 'verification mode'. Per the docs.

A router verification run is the natural and 'automagic' way. It inherently
insures that what is accepted is what is deliverable.

But if you have even ONE 'leaky' router (i.e. catchall, accept
everything/anything) then you need to either exempt that router from being used
for verification (look for 'no_verify')


> because, as far as those acl tests are concerned, lllllllll@???
> _is_ a valid recipient (i think...) at least it is by the 'front-end'
> part of exim4, until it's attempted to be delivered over lmtp to cyrus,
> and then it all goes horribly wrong.


Why would you want Exim - which can reject bogus traffic *during* the smtp
session and avoid bounce backscatter - to onpass garbage to Cyrus - which cannot?

>
> i am beginning to suspect a bug in exim4: it's not responsive at all to
> any of these acl changes.
>
> i even put MAIL FROM:<> RCPT TO:lllll@localhost and _that_ gets
> accepted, and that's _bad_ right?
>


'Bu ci' (Chinese. 'Not So').

You may be more familiar with the Americanized version:

Bullshit

;-)

All the information you need to clean up your configuration is in (both) the
Exim 'main' documents, and the Debian-specific ones.

Poets can read, nicht var?

Your misuse is not a software bug.

A Gadfly, maybe, but not an *Exim* bug.

Bill