On Thu, 19 Jan 2006, John W. Baxter wrote:
> On 1/19/06 11:21 AM, "Chris Knadle" <Chris.Knadle@???> wrote:
> > On Thursday 19 January 2006 13:55, Tony Finch wrote:
> >
> >> Alternatively, you could just not bother to maintain a copy of the
> >> userlist and use recipient callout verification instead.
> >
> > Yes, I did think of that, and it is a good suggestion. I haven't suggested
> > that to the other admin yet, as I'm trying to avoid doing that as it
> > generates more traffic and this admin is on a somewhat slow static DSL line.
There is a callout cache which should minimize this effect.
> I'm not sure recipient callout verification is really appropriate here.
> Aside from spammers trying to sneak in (lots of activity), the reason a
> message is presented to the backup MX is that the primary MX is unavailable.
> It's likely still to be unavailable when the callout is done. So then you
> generate a 4xx response, and the sender queues just as it would if there had
> been no backup.
You can use the defer_ok option to avoid this problem. If the outage is
short, the callout cache will still be useful. You can increase the
lifetime of negative cache entries to minimise the number of messages to
invalid recipients that you accept during longer outages.
Tony.
--
<fanf@???> <dot@???>
http://dotat.at/ ${sg{\N${sg{\
N\}{([^N]*)(.)(.)(.*)}{\$1\$3\$2\$1\$3\n\$2\$3\$4\$3\n\$3\$2\$4}}\
\N}{([^N]*)(.)(.)(.*)}{\$1\$3\$2\$1\$3\n\$2\$3\$4\$3\n\$3\$2\$4}}
