Casey Allen Shobe wrote:
On Monday 13 June 2005 15:11, Sam Varshavchik wrote:
2) You should avoid bouncing mail after it's already accepted by your mail
server. What you're doing might've been tolerable years ago, but on
today's Internet, where an estimated 75% of all E-mail is spam with forged
return addresses, sending backscatter bounces to forged return addresses --
especially in situations where they are easily preventable -- is considered
abusive.
I'm sorry, but what??? Bounce messages are a normal and reasonable part of
the E-mail infrastructure, and are *not* considered abusive.
what if one says "spam messages are a normal ..."?
We see lots of
mail coming in to local accounts that don't exist, and then the server
automatically sends out bounce messages. These bounces all typically bounce
themselves, and eventually disappear out of the queue. What would you
propose instead?
I certainly expect to receive a message if I accidentally typo an E-mail
address, otherwise I'll have no idea the recipient did not receive the
message.
It's ok to get a bounce from your server, not from a remote server. the
remote server must reject the recipient (with a 5xx), then your server
will generate the bounce for you.
servers should not accept mail that won't be delivered unless
- the sender is trusted
- the failure could not be predicted (quota exceeded but only detected
later, etc).
As Sam says, backscatter is a serious problem, and it has become too
much (remember those silly "you sent a virus" messages?). If you get a
lot of these, you'll probably change your mind:)
uncontrolled bounce can also be used to "spam by reflection". Instead of
sending spam to fo...@example.com, I send it to an invalid address in your
site and hope you'll send an NDR to fo...@example.com. Poor foo...
9 messages in net.sourceforge.lists.courier-maildropRe: [maildropl] Bounce email in .mail...
