Admins Accuse Microsoft of Hotmail Cap

kurmudgeon writes "The Register is fielding reader tips that Hotmail has placed Draconian limits on the number of Hotmail recipients who can receive an email. The first 10 Hotmail addresses included in a mass email go through just fine, according to these reports. But any additional addresses are returned to sender with a message that reads: "552 Too many recipients." (Microsoft denies it has placed any such restriction on the number of senders.) This would appear to be a violation of RFC 2821, which states: "Rejection of messages (for excessive recipients) with fewer than 100 RCPT commands is a violation of this specification."

Comments

And the problem is...?

[Kelson]

Let's look at that phrasing: "Rejection of messages (for excessive recipients) with fewer than 100 RCPT commands is a violation of this specification." (emphasis added).

Are they rejecting messages, or are they rejecting recipients?

According to this, they're rejecting recipients with an obvious "try this again" code. Really that should be 452, not 552, but that same RFC 2821 says that senders should treat a 552 as temporary:

RFC 821 [30] incorrectly listed the error where an SMTP server exhausts its implementation limit on the number of RCPT commands ("too many recipients") as having reply code 552. The correct reply code for this condition is 452. Clients SHOULD treat a 552 code in this case as a temporary, rather than permanent

So whatever sending server runs into these limits should retransmit the message to the remaining recipients on the next queue run. Okay, it'll only reach 10 recipients at a time, which is annoying. It shouldn't be kicking back the error to the client.

Really, assuming Microsoft has actually put this limit in place, the only thing I can see that's wrong, from a practical standpoint, is using the outdated 552 code instead of the more specific 452 -- but that same RFC people are waving around says that their servers should treat it as temporary anyway.

Am I missing something?

Re:And the problem is...?

[Obfuscant]

No, according the standards, every recipient rejected for "too many" stays in the queue and delivery is attempted at the next queue run. While Hotmail's violation of the standard seems bad, the worst effect it should have is to slow the delivery, not prevent it.

If a client actually stops trying to deliver based on a 552 error, then it, too, is violating the standard, in a way that actually prevents delivery. I consider that a more serious violation.

Re:And the problem is...?

[dgatwood]

Only if the sending SMTP server is broken. SMTP has two types of return codes: permanent failures and temporary failures. A permanent failure causes the message to bounce to the sender. A temporary failure causes the message to be queued and resent. Upon resending, only recipients for whom an error was generated are retried. Thus, if this error occurs after ten recipients, the remainder won't get the message in the first pass, but the next ten will get it when the sending server retries (usually after an hour, IIRC). This should continue until the recipient list is exhausted. Even this assumes that the sending SMTP server is extremely dumb and doesn't really understand anything about this error code at all beyond that it is a temporary error.... If it actually understands the code, it should try resending to additional recipients immediately, and divide the message into smaller batches, in which case it would delay delivery by a few minutes at most.

In theory, in some extreme cases, the recipient might never get the message. If it retries once an hour for a week (fairly typical), that would effectively cap the number of HoTMaiL recipients of a single message at 10 * 24 * 7 = 1680 recipients. Of course, a proper sending SMTP server should already be able to split messages into batches of a hundred or less because a limit of 100+ is considered acceptable behavior by the receiving server. Thus, in effect, because 1680 is larger than 100, short of a very long term net outage after the initial connection attempt, all the recipients should receive the message in every case. If this does not occur, the sending SMTP server is broken.

This is, of course, just my opinion.

Re:And the problem is...?

[walt-sjc]

Dealing with mail servers that handle newsletters with subscribers numbering in the 100K+ each range, I have several special configurations on my servers to handle various broken behavior. In fact, I have a "butthead-TMR" list that contains hotmail for this EXACT reason. I also have a "butthead-TMC" list for a few broken servers that start doing the same 500 level errors for "too many simultaneous connections".

It's one thing to have anti-spam and anti-abuse mechanisms in place, it's another to deliberately break basic functionality in direct violation of the standards that make email work. There are MUCH better ways of handling situations where you want to rate limit inbound mail that are fully compliant with the RFCs, that allow all valid mail to get through.

It simply amazes me how many IDIOTS are running servers at large ISP's / sites. It is well known by most competent email admins that hotmail is totally broken and unreliable. Anyone still using hotmail for everyday use should have their head examined.

Hotmail is unreliable anyway

[Anonymous Brave Guy]

Our (100% legitimate, double opt-in) mailing list gets a few Hotmail addresses added to it every now and then. We frequently get people complaining about missing mails and so on. Invariably, it's because of something silly, usually spam filtering that has been set to be so ludicrously aggressive that practically anything not white-listed (i.e., nothing on a new account) gets through.

We have now reached the point where we consider Hotmail an irrelevance. We don't even advise complainants to use another mail client any more, we just ignore them. The list is not run for profit, and the effort of supporting Microsoft's not-playing-ball freebie mail system just isn't worth it for what is basically a hobby set-up run for the benefit of our community.

Re:E-mail is dead for mass communication

[RollingThunder]

I take it you're not on any discussion mailing lists, then?

All MS is doing is cranking up bandwidth costs now. Instead of one copy being sent to all 68 subscribers on the server, my listserv now has to send them 68 copies of the same damned thing. Incredibly inefficient, but the subscribers want the email, so that's what'll happen.

You gotta pay up!

[argux]

I read about something like this a few days ago. A big website was warning people not to register with any Microsoft accounts (MSN, Hotmail, Live) because their mails were bouncing. They also mentioned that if they paid some fee, the cap would be lifted (obviously, they wouldn't in a million years give a cent to these people). Instead of paying, they would only recommend people to use Yahoo or Gmail.

Of course, it's not the exact, same thing, but the similarity between the two situations is spooky, to say the least.

Re:E-mail is dead for mass communication

[SnoopJeDi]

Why are people still using these? Why haven't they been replaced by forums?

At UMBC, almost all student organizations, many classes, club teams, etc. etc. all use a mailing list system powered by Sympa to communicate. It's way more convenient than logging into our blackboard site, browsing to the class, finding the discussions forums, and finding the right thread in the mangled excuse for organization.

With the mailing list, all I have to do is check my email. Email is easier to centralize to the individual than forums, and leaves organization up to the end user. I have to check my email for personal communiques, contact from professors, and automatic notifications ANYWAY, why the hell should I not use the system to stay in the loop in a group, too?

That said, reply-all is the worst thing in the world.

Re:I'm not TERRIBLY pro-MS, but...

[Random832]

Granted, security through obscurity isn't really effective, but why should they bother telling spammers how small to make their batches in order to get things through? Because there are lots of legitimate reasons to send an email to more than 10 recepients on a large service like Hotmail, and batching them up (as opposed to sending the whole email, headers and body, to the server multiple times) saves bandwidth.

I would be pissed off if i were subscribed to something and I were the 11th hotmail user on their list.

Re:RFC 2821 is not (yet) a standard

[Anonymous Coward]

It may not be a recommended IETF standard, but the RFC number (on standards track) makes it an effective (read: recommended) standard. If you're a fan of RFC's you should implement them when they are in the Standards Track Especially when it's a trivial extension of an existing protocol, especially when (most) everyone else does.

Re:Hotmail has many worse problems than this one!

[saarbruck]

OP could have provided more detail but was not trolling: I run an email server on a static IP on my 768/128 DSL line. It's for 2 users. My mother, who has been a Hotmail user for so long that she can't relocate or "no one will find her!" refuses to move to gmail despite my pleas. Hotmail silently drops mail from my server about two thirds of the time. Messages such as "Hey, where are we meeting for Grandma's birthday dinner" disappear into the ether despite me being on her whitelist. Repeated hotmail support requests go something like this:

me: why are you accepting my email with code 250 OK, but never delivering it?

them: we can't talk to you until you submit all the forms at postmaster.hotmail.com

me: submits the forms, which are clearly geared toward businesses (my "site" doesn't have a "privacy policy" or an "opt out form" because I don't SELL ANYTHING).

them: we can't talk to you until you sign up for our email tracking service to analyze your traffic

me: signs up. My server doesn't generate enough traffic for them to even log.

them: you need an SPF record

me: installs an SPF record

them: your SPF record is wrong. RFC blah blah states...

me: IT WAS GENERATED BY YOUR ONLINE TOOL!! And if you want to quote RFCs at me how about the one where if your server accepts email, you're guaranteeing not to drop it for frivolous reasons (RFC 2821, sec. 6.1)?

them: our reasons are not frivolous, but we won't tell you anything.

me: like how your servers drop email sent from thunderbird but let the same messages through when sent from outlook express?

them: we don't filter based on header information

... and so it goes. I understand that I'm a small fish in a big pond and that there's a war on terror, uh, I mean spam, but hotmail just sucks.