Re: [exim] fallback_hosts issue - is it an Exim bug?

On 2008-06-01 at 09:31 -0700, Marc Perkel wrote:
> The command works correctly if it fails to deliver to the main host it
> will send the message to the fallback host, unless the problem sending
> to the mail host is because of dns resolution. Then it just retries
> rather that sending to the fallback host.

Fallback hosts are used when there's a problem connecting to the
original hosts, at the transport level. It's for "where you're
connecting to a remote host anyway, if there's a problem, fall back to
this backup list".

Manualroute is trying to find IP addresses; the deferring is happening
at the Router level, the address hasn't yet been accepted and passed to
a Transport which might try and then fallback.

Saying exactly what is happening is hard because you're not including
log entries or debug logs, the first basic step towards problem
diagnosis and resolution.

If the DNS lookup is purely timing out, then "pass_on_timeout" combined
with a second Router which routes for the same domains but routes to the
fallback hosts.

Alternatively, if the hostname definitively does not exist in DNS then
"host_find_failed = decline", again followed by the identical second
Router.

Whilst you're doing this, "route_data = " instead of "route_list = * "
is perhaps clearer.

-Phil