More timely and informative rejections of email to @livejournal.com addresses

[pauamma]

Title
More timely and informative rejections of email to @livejournal.com addresses

Short, concise description of the ideaWhen LiveJournal decides to reject email sent to a @livejournal.com (whether a user or an internal address like accounts@, webmaster@, etc...), have it do it as soon as it gets the email, and with a link to an explanation of the

Comments

[pinterface]

Are there still mail servers that actually wait 5 days before considering a message timed out? Oy. Mine bails after 24 hours, and even that seems long.

I'm of the opinion mail servers should be very detailed and specific about why they are rejecting mail. Yes, in some cases that makes getting around those blocks easier for spammers--but that downside is far outweighed by the upside of improving the reliable delivery of e-mail for legitimate senders. Botnets generally can't make FCrDNS work, for instance, because they don't have control over the PTR records; legitimate senders will learn pretty quickly they need it if rejections point them in the right direction.

However, I will note that if the DNS response was SERVFAIL, rather than NXDOMAIN, a 4xy response would be appropriate.

[pauamma]

Are there still mail servers that actually wait 5 days before considering a message timed out? Oy. Mine bails after 24 hours, and even that seems long.
From RFC 5321 section 4.5.4.1, "Sending Strategy": Retries continue until the message is transmitted or the sender gives up; the give-up time generally needs to be at least 4-5 days. It MAY be appropriate to set a shorter maximum number of retries for non-delivery notifications and equivalent error messages than for standard messages. The parameters to the retry algorithm MUST be configurable.However, I will note that if the DNS response was SERVFAIL, rather than NXDOMAIN, a 4xy response would be appropriate.
Agreed, as should all other inconclusive outcomes: timeouts, non-authoritave replies with an empty answer section, etc. (I would be very surprised if Postfix didn't enforce that for smtpd_client_checks.)

[pinterface]

The RFC also defines VRFY, but almost every system which does an address verification callback uses RCPT instead. It allows for the usage of address literals, but you won't find many mailservers that actually accept them (mine included). Or my personal pet peeve: the protocol defines a way to reject a message, but plenty of mailservers instead silently discard mail--or dump it into a spam folder--rather than simply rejecting it outright.

The point being that the RFCs don't always reflect the modern reality of running a mailserver. RFCs have suggested long retry periods since the early days when networks were unreliable and spotty (RFC 1123 § 5.3.1.1--from 1989!), but nowadays users expect e-mail to be pretty quick and if you wait several days to tell them the other guy didn't get their message, they are going to kvetch.

That's not to say there's anything wrong with waiting 5 days if you can pull it off. I can't, and it surprises me that others can. :)

[pauamma]

I think this is getting OT for suggestions fast. :-)