lmtpd -a and unified murder

[Andrew Morgan]

On Wed, 25 Apr 2007, Janne Peltonen wrote:

> On Tue, Apr 24, 2007 at 02:01:08PM -0700, Andrew Morgan wrote:
>>> valid authentication mechanisms available and gives up. So. Is there a
>>> way to configure lmtpd so that it won't expect being able to
>>> authenticate to the remote backend (sic) and just proxies through
>>> everything it gets?
>> Nope.  I beat my head against this for a while myself and finally
>> configured my MTA (postfix) to use authentication when talking to lmtp.
>> Previously, I was using pre-authentication and tcp-wrappers.
>
> OK, thanks, that's what I thought. *sigh*
>
> Now, the Murder here will receive mail from two different
> systems: mail that's addressed directly to users' personal mailboxes
> comes from our spam-scanner machines, and list-mail comes from the list
> server (which receives the mail through the spam-scanners, so that we
> don't have to scan mail for spam for every member of every list
> separately). The spam-scanners are modern Linux machines, and we have no
> trouble configuring the Sendmails there to use authentication, but the
> list server is a Tru64 system - and whatever we do, we don't seem to get
> the Sendmail there to authenticate succesfully; as far as we understand,
> the Sendmail config there is identical to the Sendmail config on the
> Linux machines that work. Our best guess is that the SASL library is
> just too old (and newer versions don't seem to be available for Tru64).
>
> So, dear all: might it be possible that Cyrus SASL 1.5.28 is 'just too
> old'? Or are there some Catch-22's in its configuration with a
> Sendmail that wants to use authentication (the version we've here is
> 8.14.1)? Or are there some people who've succesfully compiled a newer
> SASL on a Tru64?

I can't answer the SASL question.  How about relaying the mail from the 
list server to your spam-scanner machines?

 	Andy