Re: [exim] Time based conditionals in ACLs

Página superior
Eliminar este mensaje
Responder a este mensaje
Autor: Marc Perkel
Fecha:  
A: W B Hacker
Cc: exim users
Asunto: Re: [exim] Time based conditionals in ACLs


W B Hacker wrote:
> Marc Perkel wrote:
>
>
>>>> The idea here is I return a temp error 1 in 5 times. Not enough to
>>>> block them. But enough to maybe get people's attention when they look
>>>> at their logs. Hopefully someone will notice it and fix it.
>>>>
>>>> I recommend that everyone do this and if they did it would improve
>>>> things in general.
>>>>
>>>>
>>>>
>>> If you do at least a host -v, dig, or whois on a sampling of those
>>> arrivals, you will probably find yourself trying to modify the
>>> behaviour of zombified WinBoxen on dynamic IP.
>>>
>>> Hardly likely to 'improve things in general' - unless you own stock in
>>> the local power grid.
>>>
>>> Bill
>>>
>>>
>> Yes - I have it at the end of a lot of other tests so it's not doing it
>> to zombie bots.
>>
>>
>
> Well then you might consider the *other* thing you have overlooked:
>
> No one, human, animal, or computer - will pay much attention to a log entry that
> may not even appear, or at least not with the details you sent - for a delivery
> attempt that *eventually* gets through.
>
> 'Defer' won't cut that, and even 'drop/deny' may be ignored.
>
> If you intend to put something 'in your face' for that remote sysadmin, then you
> will want to use:
>
> accept
>      control = fakereject/<message>

>
> So that the *senders* harass their sysadmin to fix the problem.
>
> Otherwise, that individual owes you Jack Squat, will do SQRT-Future Activities
> about his PTR, but *will* show up on this list wingeing about how rude we are to
> actually enforce RFC 822 and subsequent in the face of his budget limitations...
>
> Bill
>


That's an interesting idea.