On Thu, Oct 09, 2003 at 12:42:11PM -0700, [EMAIL PROTECTED] wrote:
> On Thu, 9 Oct 2003, Jeff Godin wrote:
> 
> > I don't see this. The domain apparently continued to resolve until
> > ~10 days past the expiration date, at which time it started
> > resolving to sitefinder.
> 
> Are you saying that it continued resolving to mp3.com's website for up to
> 10 days after expiration?

[snip bit about inadequate monitoring technique, fault]

> > At the time that *someone* noticed and fixed the problem, sitefinder
> > was still operational. This would seem to support the theory that
> > the wildcard's removal was not the catalyst that caused anyone to
> > notice the problem, and therefore the wildcard's presence had no
> > significant role in preventing anyone from noticing the problem.
>
> The problem was "noticed and fixed" on Oct 6, over 24 hours after
> sitefinder was disabled on Oct 4.
>

I made an inference based on the information in the "Day the music
died at MP3.com" article. 

The domain's former expiration would seem to have been 2003-09-26, based
on whois showing a current expiration date of 2004-09-26.

The Register's story was posted at 2003-10-06 at 23:01 GMT according
to their site.

The article begins "Yesterday was almost the day the music died for
MP3.com after the firm forgot to renew its domain name". 

Steve Cox is quoted as saying:
``"When I went to my artist site URL this morning, I saw one of the
infamous VeriSign redirects instead of my music, complete with the
dreaded message, 'We didn't find: artists.mp3s.com'. There is no
Website at this address."''

John Moppett is quoted as saying: 
``"Went to MP3.Com, but can't pick up any links, they all end up
with Verisign!!"''

This seems to indicate that sitefinder was active when the problems
were noticed and corrected, and this seems to indicate that the
breakage was major, and likely didn't go unnoticed for days. 

All of this seems to indicate that the domain continued to resolve
for a period of time after expiring.

But since we're just working from whois and the article above,
there's plenty of room for error and such -- especially with the
lack of specific time/date information.

Taking this down to a simpler situation...

For purposes of example:

I have registered example.com and example.NET in the usual fashion.

I have a major web site at http://www.example.com/ that also
references example.NET urls for portions of its content.

If I neglect to pay the renewal fee before the domain example.NET 
expires, at some point the domain will be removed from the NET.
zonefile.

At that time, things will break. With sitefinder, things break with
a sitefinder "hey, there is no website at this address" message.
Without sitefinder, things break with a "no such host" or similar
NXDOMAIN equivalent.

Where is the ``bigger link to Verisign's Sitefinder than what the
article reports'' that is mentioned in the "FALLOUT FROM SITEFINDER"
weblog entry?

Where is the specific unintended consequence or non-standard
behavior that resulted in ``The expiration most likely went
unnoticed due to Verisign's wildcard record in the gTLD''?

Verisign's action with the insertion of the wildcard broke many
things, but in this case I fail to see where the wildcard or
sitefinder resulted in anything other than a different color of egg
on mp3.com's face.

-jeff

-- 
Jeff Godin <[EMAIL PROTECTED]>
Jedin Technologies

Reply via email to