Re: whitelist_from_rcvd / trusted_networks

2014-11-11 Thread RW
On Mon, 10 Nov 2014 23:30:28 -0500 (EST)
Derek Diget wrote:

 
 We have a department that has subscribed to a service in the cloud
 product that is sending email to us via our MX record.  The problem
 is that they appear to be using shared servers/IPs and thus every
 once in a while mail will source from an IP address that will cause
 it to score above 5. ...

 I would like to use whitelist_from_rcvd as the envelope from 
 (RFC5321.MailFrom) and sending system is not exactly static, but
 close enough that the globing should work.  The issue is that SA is
 running on our MXes via a milter and since SA (and these boxes) only
 see MX traffic, trusted_networks and/or internal_networks are empty.
 This causes the whitelist_from_rcvd to never fire.
 
 Our MTA does construct a synthetic Received header
 
 My question is how can I make this Received header trusted 


What makes you think it isn't? Most blocklists run on the last-external
IP address, the fact that it's being flagged as spam based on IP
address suggests it is.

Try adding the following to local.cf:

add_header all Relays-External  _RELAYSEXTERNAL_

The first section of this header should have the parsed information from
the MX server. Check that the ip and rdns fields are correct.


Re: whitelist_from_rcvd / trusted_networks

2014-11-10 Thread David B Funk

Even in that configuration (which is -very- much like ours) you must have
your MXs (at least their IP addrs) in your internal_networks.
All kinds of things break if your MXs aren't listed as trusted/internal.

Just be sure that synthetic Received header is constructed correctly
(the one Achilles-heel of milters).

Are the messages DKIM authenticated? (Either DK signed or SPF listed)?
IE can you use whitelist_auth ? It's more reliable than whitelist_from_rcvd
which depends upon finding the correct DNS names of all the SMTP exit points.
It also depends upon the Envelope From address being available to SA.

On Mon, 10 Nov 2014, Derek Diget wrote:



We have a department that has subscribed to a service in the cloud product 
that is sending email to us via our MX record.  The problem is that they 
appear to be using shared servers/IPs and thus every once in a while mail 
will source from an IP address that will cause it to score above 5.


I would like to use whitelist_from_rcvd as the envelope from 
(RFC5321.MailFrom) and sending system is not exactly static, but close enough 
that the globing should work.  The issue is that SA is running on our MXes 
via a milter and since SA (and these boxes) only see MX traffic, 
trusted_networks and/or internal_networks are empty.  This causes the 
whitelist_from_rcvd to never fire.


Our MTA does construct a synthetic Received header as it passes the message 
to SA via the milter interface.  The message is passed to SA before the MTA 
accepts/rejects the message (scanned before the reply to DATA command).  The 
Received header it creates and adds before sending to SA is what the Received 
header would look like if the message had been accepted, queued and then 
handed off to SA via the LDA.  Therefore, the from clause is whatever 
system is relaying the message (HELO, DNS and IP), and the by clause is our 
system's name.


My question is how can I make this Received header trusted or how can I 
force whitelist_from_rcvd to fire (or some other way to whitelist a sending 
pair - envelope from, sending IP/host).  I don't want to add the IPs of the 
cloud provider to the trusted_networks.  I know that the first/top Received 
header can be trusted.






--
Dave Funk  University of Iowa
dbfunk (at) engineering.uiowa.eduCollege of Engineering
319/335-5751   FAX: 319/384-0549   1256 Seamans Center
Sys_admin/Postmaster/cell_adminIowa City, IA 52242-1527
#include std_disclaimer.h
Better is not better, 'standard' is better. B{