Author: Lorens Kockum Date: To: exim-users CC: Subject: [Exim] How do I avoid freezing messages routed to an unavailable host?
OK, the reason the mail was frozen is that localhost was
incapable of finding the IP address of the remote mail gateway
(*my* host was cut off from the 'net). But I don't want the
mail frozen, I want it sent out without manual intervention when
communication is back up. How do I get that behaviour? I love
the concept of freezing messages when there's nothing else to be
done, but it doesn't seem to me that it was justified in this
case.
Scene: mail generated locally by user X for root, .forwarded
from root to user Y, and from user Y to user X (don't ask) and to
a remote address.
The remote address goes through a router because localhost is
the primary MX for the remote address, which is then statically
routed to its destination. Using a router is the way I found/was
told was best to emulate sendmail's mailertable.
So, the remote destination is in relay_domains, the primary MX
of the remote host is localhost, and the first router in the
router configurations is
[router name]:
driver = domainlist
transport = remote_smtp
route_list = "[remote destination] [name of remote mail gateway] byname ; [another remote destination] [same remote mail gateway] byname"
2000-09-24 06:17:32 13d3EK-0000Ay-00 <= [local user X recipient address] U=[local recipient user X] P=local S=642
2000-09-24 06:18:47 13d3EK-0000Ay-00 == [remote recipient address] <[local user Y]> R=[router name] defer (-1): lookup of host "[name of remote mail gateway]" failed in [router name] router
2000-09-24 06:18:47 13d3EK-0000Ay-00 => [local recipient user] <root@[localhost]> D=localuser T=local_delivery
2000-09-24 06:18:47 13d3EK-0000Ay-00 Frozen
Basically, I'm disappointed that the IP lookup in the router is
treated as a fatal error. Is this unavoidable? I see that the
message cannot be placed in an outgoing queue since we don't yet
know the remote IP to send it to, but isn't it possible to leave
the message where it is, *before* going through the router,
without freezing it, and letting it retry the *router* every so
often?
If I replaced the name of the remote mail gateway in the router
config with its IP address, or if I put the name in /etc/hosts,
would that solve my problem?
