Autor: David S. Madole Data: Para: Jerry Nicholls, exim-users CC: Assunto: Re: [exim] Pre-emptive fallback hosts ?
From: "Jerry Nicholls" <Jerry.Nicholls@???> >
> On Tue, 2004-09-14 at 12:34, Chris Edwards wrote:
>
> > First of all you'll need a way to flag when too many emails fail or defer. > > I'd be inclined to use a log-watcher script, but no idea if that's
> > practical in your environment. Whatever, I guess you'll end up with a > > table of maybe-compromised customer IP addresses.
>
> I'd prefer not to have any external process with this. I already do a
> 'reverse' grey list on senders, ie if they send to too many recipients
> too fast they get 4xx errors on subsequent RCPTs until a timeout occurs. > All that's done with a MySQL database and ACL logic. I was hoping I
> could do the same for this instance. I can't currently figure out how to > update the database when a router has deferred or failed tho'.
How about using an expansion in a message condition in the ACL to look at
the text in $acl_verify_message and doing the update to your database
there? You could make the expansion a "noop" as far as the actual text is
concerned:
It's a little hackish, but not more so than a log-watcher script, in my
opinion.
Just an idea. It's untested, so there may be {} or other errors in the
example above. You would need to tune the text match to whatever defer
reason you were interested in, the above is just an example.