On Thu, Aug 03, 2006 at 06:19:10PM +0100, Ian Eiloart wrote:
>
> --On 3 August 2006 17:20:46 +0100 Dave Evans <davide-20060629@???>
> wrote:
>
> > Should the per-domain callout cache records be consulted for recipient
> > verification when use_sender is true? I suspect that they should not be
> > consulted, but in fact they are consulted.
> >
> > For example: my server has "require verify=sender/callout", and someone
> > tries to email me from <someaddress@???>. coscon.co.uk's mail
> > servers are RFC-ignorant, and reject the null sender; therefore, all
> > sender verification for that domain will fail, and a callout cache record
> > is written (as I understand it) stating that verification for
> > "coscon.co.uk" fails for all local parts.
> >
> > My server also has "require verify=recipient/callout,use_sender". When
> > someone now tries to send email *to* <someotheraddress@???> (or
> > indeed to the same "someaddress"), verification seems to fail, using the
> > cached callout record. However if the callout had actually been
> > attempted, it probably would not have failed, because the sender address
> > was non-null (therefore coscon's servers would accept the MAIL FROM
> > command, etc).
> >
> > My server's behaviour, the 4.62 source code (as far as I understand it),
> > and the spec, all seem to agree that this is what happens, but it doesn't
> > seem like a Good Thing. Is it a bug?
>
> It's not a bug - it's a fault in coscon's configuration. Rejecting NULL
> senders breaks lots of things. Have you complained to their postmaster?
>
But from what I understand the OP is questioning the Exims behaviour when
there is a previously cached result using different parameters.
1. Mail server at coscon rejects null sender - ok, that's bad and they
shouldn't do it.
2. Exim doesn't even check realuser@??? because the previous
check using the null sender failed.
IMO that's wrong.
Steven.
--
Ass, n.:
The masculine of "lass".
