Is it the sound of communication breaking down?

[venta]

Nearly a year ago, I had one of those conversations that goes:
"Did you get my email?"
"Er, no."

That email eventually rolled in, over a month later. Ah well, I thought. A randon glitch. Annoying, but I suppose these things happen.

However... they're happening more and more often. No, I don't expect email to be instantaneous. But mails taking days to arrive seems to be getting quite common. I've always assumed that once sent, a mail will arrive reasonably quickly. If not, the sender will receive a bounce message, and will at least know that all has not gone according to plan. Yet an email I sent over a week ago hasn't arrived, and neither has it bounced back to me.

If nothing else, this means I'm going to have to stop sending mails that say things along the lines of "reply to this if X", then assuming not-X if I don't get a reply. I know Outlook provides a request-receipt facility, but I'm assuming this won't work if people read their mail with a different client.

At least when sending SMS messages you can request a receipt, and you know the message has at least arrived on the recipient's phone (though, of course, they might not have read it.) Except... over the last six months I've been having increasing trouble with SMSs too. On New Years Eve, for example, it's accepted that the networks are busy, and messaging isn't reliable. But on an average weekday, I don't expect a message to take upwards of three or four hours to arrive. Similarly, if it has arrived, I'd like the receipt to land back with me reasonably promptly.

So, at what point does it become acceptable to start demanding to know why you haven't had a reply to a message of some kind ? Probably most of the time the reason is that the recipient hasn't got round to it, which is perfectly reasonable. "Did you get my mail?" tends to be a polite fiction, the real question is "Why haven't you answered it?" Yet it looks as if it's a question which needs to be asked - otherwise you never find out that actually, no, the mail did just vanish into the ether.

Comments

[bateleur.livejournal.com]

Get a copy of the headers of one of these messages and take a look at the route timings to see where the real problem lies.

For example: lathany's had quite a bit of delayed incoming mail and it's almost always due to Blueyonder's mailservers.

[venta.livejournal.com]

I should look at headers more often, yes. Though I've been having problems with both my work mail and my home mail, which shouldn't really share any particular routes.

I reckon it's just a further sign of the impending apocalypse :)

Impending?

[wechsler.livejournal.com]

I thought it had happened already...

Re: Impending?

[venta.livejournal.com]

Laugh if you will,
Mock if you may,
The world is going to end...
Yesterday

Dammit, there is not enough Les Barker on line :(

[wimble.livejournal.com]

Work and Home to the same correspondent? This would imply a problem at their end, rather than yours.

[venta.livejournal.com]

No, to a variety of correspondents.

Somethings, like yahoo mailing lists of LJ someone's-replied-to-your-comment alert, I'd expect to be very slow occasionally at their end rather than the actual mailrouting end. But I'm talking about (what appears to me to be) a widespread problem to different people, at different times.

[venta.livejournal.com]

OK, if you'll excuse me posting something very long and boring...

The relevant (I think) part of the header from a mail that took two days to arrive:

Return-Path: <lj_notify@livejournal.com>
Received: from mwinf3105.me.freeserve.com (mwinf3105.me.freeserve.com)
by mwinb3101 (SMTP Server) with LMTP; Wed, 17 Mar 2004 03:24:21 +0100
X-Sieve: Server Sieve 2.2
Received: from mwinf3103.me.freeserve.com (mwinf3103 [172.22.158.25])
by mwinf3105.me.freeserve.com (SMTP Server) with ESMTP id 9401919B891A
for <1+fu6000000000000000002933990@back31-mail01-01.me-wanadoo.net>; Mon,
15 Mar 2004 12:47:51 +0100 (CET)
Received: from livejournal.com (livejournal.com [66.150.15.150])
by mwinf3103.me.freeserve.com (SMTP Server) with ESMTP id E20661800F4A
for <my email address>; Mon, 15 Mar 2004 12:38:41 +0100 (CET)
Received: from localhost.localdomain (livejournal.com [66.150.15.150])
by livejournal.com (Postfix) with ESMTP id 3925016C648
for <my email address>; Mon, 15 Mar 2004 03:33:31 -0800 (PST)

So that looks to me like freeserve got it in a timely manner, then sat on it for a couple of days ? Am I reading it correctly ?

On the other hand, a mail that took over 12 hours to make it to me (it was sent to rpgsocej) seemed to spend the missing time kicking about in the .ox.ac.uk domain.

[wimble.livejournal.com]

Am I reading it correctly ?
Yup. Unless all the machines before that one had their clocks out by two days. Which seems somewhat unlikely.

[reddragdiva]

The SMTP standard says four days. Anything beyond that is officially, technically, a problem to be solved. Checking the headers as advised will help in working out who to ask about it.

[venta.livejournal.com]

Says four days for what, exactly ? I know that, in general, if your message is undeliverable, then the giving-up time is four days. Is it guaranteed that a message will either be delivered or bounced within that time ?

[pm215]

Guaranteed by who? Mail me at work and the retry on the final hop is 30 days... (you should get 'not delivered yet, still trying' notes well before that, though).

[neilh.livejournal.com]

There are no guarantees, SMTP is a best effort system but guarantees nothing.

[reddragdiva]

Hmm. I thought after four days your server would either hand it off to another server or return it to you as undeliverable.

[onebyone.livejournal.com]

Not necessarily. Once it has decided that delivery has failed, it will *try* to send it back to you. But suppose that the path back to you is broken somehow (e.g. the machine it's trying to get back to has fallen out of either DNS or someone's routing tables). It won't keep trying forever, hence it might fail.

[neilh.livejournal.com]

Some (probably most) servers do, but there are no guarantees. There cannot be any guarantees since it is conceivable that there is some kind of circular routing happening - if I send a mail with a return address pointing to a mailbox which is full, to a non-existant mailbox, the mailserver will attempt a delivery passing it to destination MX for routing onward, the destination MX doesn't have access to the final destination user set so accepts the mail to pass on to the destination mailbox. When, say, the destination machine dials in the MX attempts delivery to discover the mail won't be accepted by the destination host. At this point it should send a non-delivery notification to the return address. The non-delivery notification will fail, since the return address mailbox is full, what happens? The mail has to be discarded silently (it *may* get routed to an admin mailbox somewhere where human intervention *may* solve one or other of the problems, but it *may not*). The internet relies on these kinds of protocols not clogging up the whole network, there have been a few incidents where such routing cirlces caused significant problem to parts of the network.

[leathellin.livejournal.com]

So.. did you get my email? :-)

[venta.livejournal.com]

<guilty>
Er, yeah, I just didn't reply :)

[diffrentcolours]

Receipts are not an Outlook-only feature, but they are handled differently by different clients; I suspect that mutt ignores them by default, for example.

[wimble.livejournal.com]

One obvious argument being that they're a transport feature, not a user agent feature. Although "delivery" nowadays is an unclear term: just because it's arrived in a POP3-able mailbox, does that mean the message is delivered?

And certainly you don't want return-receipts being generated before you've processed your spam-filters.

[snow-leopard.livejournal.com]

On an entirely unrelated subject did you find someone to go to the V and A with you or shall I see if someone on my friends list wants to escort you? (Should also really get your snail-mail so I can post you the ticket and my membership card.)

[venta.livejournal.com]

I think I've found someone - I'll get back to you asap when I know for definite. Thanks.