post...@ptld.com:
> >> Verify(8) wont attempt to retry timeouts until the
> >> address_verify_negative_refresh_time has lapsed? Correct? Anyway
> >> around
> >> this so timeouts would be re-tried on each delivery attempt? Or would
> >> I
> >> have to disable the negative database? Is there a difference between a
> >> timeout vs a successful rcpt command that reports the address is
> >> invalid
> >> in this regard?
>
> > You can set address verify refresh times as short as you like, but
> > that disables a safety mechanism, so it is not the default.
>
> Yes, but my question is, is there a distinction between a mail server
> saying its an invalid address vs a network timeout? Or does the same
> refresh value apply equally to both?
There are multiple Address verification error scenarios:
- All 5XX server responses result in a hard verification error status.
- All 4XX server responses result in a soft verification error status.
- No reply (timeout, etc.) results in a soft verification error status.
Postfix does not have different refresh strategies for different
error scenarios. On the cost side, it would make Postfix more
complicated to operate (Different negative refresh times for soft
and hard errors? A regexp map with probe-result -> refresh-time?).
The effort is better spent on things that benefit more people.
Wietse
