exim at inode
Feb 17, 2005, 3:38 AM
Post #14 of 47
(679 views)
Permalink
|
Christian Schmidt wrote:
>> RFC2821 states "while mail that cannot be transmitted immediately
>> MUST be queued and periodically retried by the sender."
Yes, but the thought, that we have to queue every single ******* message
in the first place just because someone's greylisting, makes me kind of
sick.
>> That "MUST" means that regardless of my use of greylisting or not,
>> the sender *has* to provide resources on the assumption that I
>> won't be able to accept the messages.
I assume you have never run a system that processes several million
messages a day, have you? These "assumption" you're talking about is of
course made, but it is based on average and real life values, and you
don't calculate with "If every message gets rejected temporarily, I'll
need space and power to store and process a queue of 20 Million messages.".
> And that's why I hold the opinion that the implementation of
> greylisting is somewhat equal to "wasting other people's resources".
I completely agree with that.
> => IMO greylisting may be a temporary solution, but goes to the cost
> of others and will successful as long as spammers don't use any
> queueing mechanisms.
Yes, but it isn't usefull in any way even now, when spammers abuse other
peoples systems. (Via open relay, open proxies, etc.)
And by the way, spammers wouldn't have to really "queue" mails, just try
to send every temporary failed message a second time after they finished
processing their recipient list.
But as we're talking about greylisting, and many people here seem to use
it, what about the hints database problem I mentioned here some time
ago? As you're using greylisting, I'm sure you all have a solution to
this situation:
A mail comes in, addressed to a recipient at a remote host. This remote
host uses greylisting and says "mailbox temporarily unavailable". The
mail is put into the queue. Another mail arrives, addressed to the same
remote recipient, but from a different sender. The mail in the queue
hasn't been processed a second time, (or maybe it has, but to earliy for
greylisting to accept) because the're so many mails in it. (Due to many
people using greylisting out there.) The retry time for the recipient
has passed, exim starts a delivery, and the remote host says "mailbox
temporarily unavailable". The mail is put into the queue, and so on.
This causes mails to stay in the queue surprisingly long, like some days
or so. You may think this is not very likely, but we really have lots of
them here.
I would be really gratefull, if someone could point at a simple
solution, as I can't hear people complaining "I'm using greylisting, and
my mails are getting delayed!" anymore.
lg,
daniel
PS: Don't get me wrong, this is nothing personal, I'm just very annoyed
by all these wannabe spam solutions out there, that try to make fun
with SMTP communication in one way or the other.
|
smime.p7s
1
