On Wed, 20 Sep 2000, Melanie Lampert wrote:
> Is there any way to make Exim retry X number of times before giving up
> and bouncing the message as "unresolvable"?
As I read the chapter on Retry Rules, there is not a direct way of doing
this. (You can have a separate retry rule for a DNS_timeout, but this is
something else).
I do think that the correct solution to your problem is to set up a local
caching name server (following the link in Nigel's message for such a name
server designed to sit along side an MTA). But I can also see why it
might be nice to have some more choices of errors for the second field of
a retry rules, and so I will second your implicit request to have this
added to the wish-list (although I don't actually anticipate using it
myself)[1].
Now, there might be an indirect way of doing this, which I started to look
at before I fixed my routers for multiple ISPs.
Routers are tried in order, so if a router could be added as the last
router which simply says something like "defer this" then all unroutable
domains will get deferred and will then be subject to the normal retry
rules. I never figured out how to get a router that said "defer this",
but it shouldn't be hard.
Maybe if you add a final router that says
force_defer:
driver = domainlist
route_list * host.which.never.runs.smtp byname
And place the IP address of host.which.never.runs.smtp in /etc/hosts
(and configure your resolver to check /etc/hosts before using DNS).
This is just an artificial way of forcing an error that will cause a
deferal. IMPORTANT: If the names host ever does start running an SMTP
daemon (even by accident, and you'd be surprised at how often that
happens) you will be in deep doo-doo.
It would be nice to have an option that forced a deferel in routing, or a
more robust way of doing this. Maybe others have ideas. But in the mean
time, you should sort out your DNS problems indendently of exim.
Notes:
[1] I think that I have lost count of the wish-list requests I've made
that have been honored, but I ended up not using. So maybe others
who would like to see more of this in retry rules will succeed.
-j
--
Jeffrey Goldberg
I have recently moved, see
http://www.goldmark.org/jeff/contact.html
Relativism is the triumph of authority over truth, convention over justice
