Re: [exim] PRDR config example

Αρχική Σελίδα
Delete this message
Reply to this message
Συντάκτης: Jeremy Harris
Ημερομηνία:  
Προς: exim-users
Αντικείμενο: Re: [exim] PRDR config example
On 28/09/2020 20:52, Yves Goergen via Exim-users wrote:
> I'm trying to get the PRDR extension to work but can't find any
> documentation about its ACL. All I see is the global option
> 'prdr_enable' and the ACL 'acl_smtp_data_prdr'. But I don't know how to
> use it.


It gives you a chance to accept/reject individual recipients of
a multi-recipient message. But only when the sending system
has flagged that they can handle the feature.

It gets called once for each recipient, and $local_part
and $domain will have the values for the recicpient.
(docs ACL chapter, section 9).


> It's also unclear how the PRDR and DATA ACLs interoperate. I don't
> understand what the documentation says about it, it's very vague.


The DATA ACL will still be called (after); it could reject the message
even if some single recipients are still non-rejected after
the set of PRDR ACL calls.

You need to write suitable ACL coding to ensure you don't do
no-longer-needed operations in DATA ACL for a message where PRDR
is being done.


> I want to set this up so that multiple recipients in a single message
> can be delivered in one go. Previously, I had to defer all from the
> second recipient so that content filtering works properly. This delays
> additional recipients by many minutes.


Yup, that's why PRDR exists.

> Would this check in the RCPT ACL be correct then?


ACLs chapter, section 26 on ACL conditions.

>
>>     defer
>>         message = only one recipient at a time, or use PRDR
>>         condition = $prdr_requested
>>         condition = ${if def:acl_m_rcpt_addr {1}{0}}
>


You're using a "condition" general-purpose condition. It takes
a possibly-empty string. Generate a suitable string for your
needs using a string-expansion (chapter 11).


Use the "concept index" to navigate the docs
(
http://exim.org/exim-html-current/doc/html/spec_html/ch-concept_index.html )
--
Cheers,
Jeremy