Chris Wilson, 2010-06-16 15:55:
>>>> connected via the same physical network. Even in the very unlike case
>>>> that there is a transient problem at the time of the connection
>>>> attempt to the first address, which does not exist at the time of
>>>> second attempt, the delivery will be retried again, multiple times.
>>> How do you know whether or not they tried and failed to connect to the
>>> primary MX before hitting the backup?
>>
>> I don't, but that doesn't matter.
> I suspect that we are just misunderstanding each other, but what I
I guess so.
> understood from your email was that this could happen:
>
> client -> primary MX (fails due to temporary network error, never seen by
> primary)
> client -> secondary MX (succeeds and then client is blocked, mail refused)
>
> and later for another message:
>
> client -> primary MX (mail refused due to client still being blocked)
No. By "block", I mean "reject with '4xx use primary MX'", not "reject
forever". The latter would be really dumb setup, because it is more
complex and violates fundamental rules of system design (resilience,
robustness).
> Could this be what happened to relay.rzone.de?
You mean what happened to the OP? I can't think of any reason why they
would do such a thing.
