On 2017-07-20 09:18, Andrew C Aitchison wrote:
> On Thu, 20 Jul 2017, admin@??? wrote:
>
> > https://bugs.exim.org/show_bug.cgi?id=2137
> >
> > Dreas van Donselaar <dreas@???> changed:
> >
> >
> > --- Comment #5 from Dreas van Donselaar <dreas@???> ---
> > One more attempt from my side to get this resolved upstream (we prefer
avoiding
> > patches where possible).
>
> > If the above is not convincing (which I presume so as I'm not an expert
and
> > likely am just repeating the others ;)), my other argument would be
that if
> > this change from previous Exim versions was intentional you would have
expected
> > it to be shown in the changelog?
>
>
>
Hi Andrew,
Unfortunately no, it's not the same problem as in #2147. Just to clear up
the differences :), in #2137 we've seen the following scenario:
- We attempt a verification of the recipient
- The random catch all check is being made, but the destination returns a
4XX response instead of 5XX
- Exim doesn't continue with the the verification of the actual recipient,
which is in fact valid.
As Dreas said there are quite some cases (e.g. migrating users) were a 4XX
response for an invalid recipient will be returned, and with the change in
the behavior in 4.89, valid recipients would have seen sudden temp
rejections.
In #2147 we've seen a different scenario:
- We attempt a verification of the recipient (cold cache)
- The random catch all check is being made, destination returns a 2XX
- Exim defers the callout verification, temporarily rejecting the
recipient, but still writes random=accept in the cache
- We attempt a verification of the recipient (hot cache)
- Recipient is accepted
Regards,
Alex
