Hi, Neil,
My confusion lies in the fact that the email client campaign software is
actually initiating a connection to the virtual domain, and sending the
mail to it -- all on the same server, mind you -- then IT sends it out
through the interface with it's virtualized IP address.
I don't seem to grasp how the receiving smtp server for the recipient
would not see the original IP address as that of the originating email
campaign software IP address since IT originated the email.
On 5/23/2011 12:12 PM, Noel Jones wrote:
On 5/23/2011 10:19 AM, Jeffs wrote:
On 5/23/2011 11:06 AM, Noel Jones wrote:
The Received: header is added by the receiving system, not
by your server, and always shows the IP of the connection.
If you want each client to have their own identity, each
client will need its own IP. You can't virtualize this.
Understood. Thank you. I do understand virtualization to a
degree although I haven't actually instituted it yet. So you
are saying each sender, as per example, above, would have
their own static IP address and that would show as the
originating IP sender?
If that's what works I will institute that.
Thank you and correct me if my thinking on this is inaccurate.
Cheers.
Yes, if each client needs their own identity, they each need their own
IP. Note this is a limitation of the SMTP protocol, not postfix.
With SMTP, the connection is identified by the IP, then the name is
derived from the PTR hostname of that IP. That's why you need
multiple IPs to support multiple names.
(While it's technically possible to set multiple PTR hostnames for an
IP, there is no mechanism tell the remote client which hostname
applies to "this" connection. So as a practical matter, it makes no
sense to use more than one PTR hostname).
Also note that the accepted practice is for the source IP to identify
itself as the mail service provider, not the client.
-- Noel Jones
