Hello -
I don't know if this is a bug, a mis-configuration, or just a usaged I
didn't understand. Let me start by describing the configuration. I've
got an SGI IRIX 6.2 box that is a backup mail server, running Exim 1.92.
The MX records specify an IP address that is an alias address on that
server (why? I don't know. No good reason). Whenever the primary mail
server would go down, messages would come into the backup server and then
start looping immediately with mail going back into the secondary server.
This would take a few seconds and then the message would get returned
with a too many headers, mail-loop message.
After playing around with various options, I tried adding in the aliased
IP addresses to a local_interfaces configuration line and then it started
doing the right thing of recognizing the loop and holding the mail for
later attemted delivery (when the primary came back up).
First, the documentation says that "it also gets a complete list of
available interfaces and treats them all as local when routing mail." It
doesn't seem like it is picking up the aliased addresses on an interface.
Is this correct behavior, an oddity of an IRIX implementation, or a bug?
If it is not a bug, it should be added in to the docs/upcoming FAQ so
people can be aware of this limitation. Or is there some other
configuration I've gotten wrong and I'm just masking the problem by
adding in the interfaces here?
-Jim straus
--
*** Exim information can be found at
http://www.exim.org/ ***
