On Tue, 2 May 2000, Karen Cravens wrote:

KC>Hotmail apparently had one of the servers fill up its disk; if you've got
KC>a lot of hotmail subscribers, you'll notice a lot of auto-unsubs tomorrow
KC>morning.  They'll have gotten a "Hey, you got kicked off" notice by then,
KC>which will hopefully go through, and they can resubscribe... otherwise,
KC>they'll get another one in three days.

Further clarification to this (warning, it gets a little technical... skip
to the bottom if you don't care about technostuff).

You'll notice that your status report gives you a reason when it can.  As
mentioned earlier, AOL gives a wacky 250 error sometimes.  Now, 200-series
responses are supposed to be *good*, so AOL's delivery report was
returning the wrong response... it wasn't really an SMTP failure.  (SMTP
is the type of actual transaction between our machine and the mail
destination.  The SMTP transaction was successful in the AOL example...
AOL *later* returned a message saying the delivery had failed, but that
was after we'd finished giving them the message via SMTP.)

Anyway, if something can't make an SMTP connection, you get either a
400-series or 500-series error.  (Same like you get a 404 in an HTTP
connection with your web browser.)  A 400-series means "Things are screwed
up right now, try again later," and the Phoenyx will keep trying every so
often until the mail either goes through or returns a 500-series.  
500-series means "Things are unlikely to ever resolve themselves to where
I can accept that mail.  Don't ever talk to me about it again."  The most
common 500-series is, of course, "550 Requested action not taken: mailbox
unavailable" which is used for a mailbox not found.  (The actual text is
variable, so some systems will actually tell you that the user doesn't
exist.)

There is a specific 400-series error ("452 Requested action not taken:
insufficient system storage") in the RFC that is supposed to be used for
temporary/non-serious disk full errors.  Hotmail isn't using it... they're
responding with "552 Requested mail action aborted: exceeded storage
allocation," which is the "permanent" version... this is supposed to
indicate that the storage situation is a serious, multi-day situation.

End of technobabble.

The upshot is, normally the Phoenyx will not unsubscribe people for
transient errors like this... it'll queue the mail for three days or
whatever we're currently set to (Carl?) which is usually more than long
enough for errors like this to get fixed.  (If you're reading this from a
hotmail account and your headers include hotmail's "law5" server, they've
already gotten it fixed.)  But Hotmail is, in effect, telling the Phoenyx
that it *isn't* likely to get fixed in a reasonable time period and their
server would like to not be pestered about it, so the Phoenyx is
obediently unsubbing those users.

-- --------------------------------------------------------------
Listowner tools are found at http://www.phoenyx.net/listowners