Steve:
> Hi,
>
> I'm sure I read a post here a while ago saying it was possible to set up
> multiple queue directories and transports. I thought I had saved the
> link therein, but I'll be darned if I can find it.
>
> What I'm ideally looking to do is something like this;
>
>
> [THE WILD] -> POSTFIX 'INBOUND' ?FINAL DESTINATION? YES -> VIRTUAL DIRS
> [OUTBOUND USERS] -> POSTFIX 'OUTBOUND' -> [THE WILD]
>
> Finally if the mail in the 'POSTFIX INBOUND' is for a final destination
> that we are just proxying/holding for, put it in the 'outbound'.
>
> Not sure if that makes sense ?
>
> This way I could apply different sets of rules to 'inbound' mail that I
> don't want to apply to 'outbound' mail, and, hold inbound mail in the
> 'outbound' queue if a final destination server is not reachable.
>
> Apologies for the verbosity, but can someone point me to some
> explanation on multiple transports/queues.
I think that the "walk-through" example in the new MULTI_INSTANCE_README
document covers a lot of this.
http://www.postfix.org/MULTI_INSTANCE_README.html
By using separate Postfixen, you can set different rules for
inside and outside sources with less confusion than when you
would try to build it into one configuration.
Multi-instance support may appear daunting at first sight, but
the mechanics are really simple.
> I have a couple of other questions I'm struggling on I think I've solved
> them but I'll say it out loud so to speak - one is regarding By default
> my Postfix is using the defacto sylog on the box. This dumps
> to /var/log/mail.info|err I'd really like to change this so it logs
> to /home/mail/mail.info. My guess is I would need to do this with
> something like syslog-ng and filters as opposed to something I can
> configure in Postfix?
The logfile is configurable (man syslogd). I don't think it is
necessarily a good idea to mix system logs among user files.
> The other is giving me more trouble. I want to control the amount of
> information given to the connecting client when it is blocked by an rbl.
> I appreciate the default is;
>
> default_rbl_reply = $rbl_code Rejected; $rbl_class [$rbl_what] blocked
> using $rbl_domain${rbl_reason?; $rbl_reason
>
> But I understand you can set up different patterns for different RBL's,
> that is a different layout for zen.spamhaus to sorbs for sake of
> example. I just can't seem to find how I lay it out?
>
> I guess I add this to main.cf:
>
> rbl_reply_maps = /etc/postfix/maps/rbl_reply (or a hash of it)
>
> Where:
> /etc/postfix/maps/rbl_reply =
>
> 554 Rejected; Client IP $client_address listed at $rbl_domain
>
> Where I'm not clear is the syntax that links each of these entries to a
> specific rbl service. I can't seem to uncover it in the DOCS. It's
> probably staring me in the face. I can see a list of options and parts I
> can use in the reply, but not where I define the actual RBL the answer
> should be used for. Anyone offer any guidance/real world examples?
>
> Thanks to all and have a great weekend.
rbl_reply_maps is indexed by the name of the provider. Thus
the left-hand side would ne zen.spamhaus.org and the right-hand
side would be your custom template.
Wietse
