On Thu, 8 Feb 2001 02:09:07 -0500 (EST), woods@??? (Greg A.
Woods) wrote:
>[ On Tuesday, February 6, 2001 at 11:07:52 (+0100), Marc Haber wrote: ]
>> People demand it. So we do it.
>
>Are you sure? Do they really know what they're asking for? Have you
>ever challenged them and asked them to explain exactly, and technically,
>why they need such a service? Do they really understand what the
>purpose is and what problems it causes in the modern Internet?
Most of our customers run Exchange, Notes or more exotic stuff. They
usually don't have clueful personnel, so it frequently happens that a
customer mail server goes down for a longer period of time (a week is
not unusual). Being secondary MX comfortably enables us to bump up
retry time for these hosts. Additionally, we notice a customer's mail
server going down when we see mail for that customer piling up on our
MX.
>I suspect most people would only have answers such as "Well I read about
>it in a book", or "Some dim-witted consultant I hired said I needed
>one!" Tell them such advice lost its relevance for the real big-I
>Internet back in 1985 or even before, and was almost certainly totally
>wrong by 1995.
I can't concur with that argumentation. Getting incoming e-mail on my
own machine is generally a good thing because what is at our side, can
be controlled by us.
>If I understand correctly the problem is that you are seeing "bounces"
>that are sent to your system since it is an MX for the target domain in
>questoin, but which your mailer cannot deliver (the customer's mailer
>gives an SMTP error code instead of blindly accepting everything your
>mailer tosses at it).
No. the problem is "errors_copy" for certain recipient addresses that
I know I can't do anything about because these simply are non-existent
addresses that found their way into the spammer's databases.
Generally, I want to see that kind of bounces because it makes me
aware of problems my customers might have.
>Just wait until you get more spammers sending with an empty sender
>address (i.e. "<>"), and choosing your MX host instead of first trying
>the primary MX. This is happening to many folks now.
Here, not yet.
I will probably try to give my exim a "negative list" of addresses
that I know don't exist, making my secondary MX reject these addresses
during SMTP time.
>> Mail comes in for the internet for foobar@??? from
>> sender@???. mx.we.example.com accepts mail,
>> spools it, delivers it to mx.customer.example.com.
>> mx.customer.example.com says "500 unknown local part foobar in
>> <foobar@???>. mx.we.example.com generates bounce to
>> sender@???, copies to
>> postmaster@???. This is the scenario I am talking about.
>
>Does the bounce get delivered successfully to the sender?
Yes.
>If so then you want to simply turn off errors_copy (it is supposedly off
>by default)! [In smail it is called error_copy_postmaster, and I get no
>end of complaints from people who turn it on without thinking and then
>say they get too many bounces! :-)] Unless you're debugging something
>you do not ever want to see copies on all bounces!
I am always debugging something ;)
Greetings
Marc
--
-------------------------------------- !! No courtesy copies, please !! -----
Marc Haber | " Questions are the | Mailadresse im Header
Karlsruhe, Germany | Beginning of Wisdom " | Fon: *49 721 966 32 15
Nordisch by Nature | Lt. Worf, TNG "Rightful Heir" | Fax: *49 721 966 31 29
