Zoom outage caused by accidental 'shutting down' of the zoom.us domain

Something is fishy about this. A communication error would not result in a domain being placed on hold. On hold is usually the result of a legal order or in the case of the .us TLD a nexus compliance violation. I've transferred thousands of domains from assorted dodgy registrars into MarkMonitor and can not even imagine a scenario where a miscommunication results in a domain being placed on hold.

Nah, weird stuff that “shouldn’t” happen almost always happens more often than things that “should” happen.

I hear ya but this would more than likely be something like a really sloppy human error such as following the wrong process vs. a miscommunication otherwise I would expect these outages to be much more frequent. I do remember when a fat-finger at UUNET took out most of the internet long ago but that was a human error and is a bit harder to have the same impact today.

To me a communication error implies someone followed erroneous instructions without asking the obvious, " ... but isn't this a big business that is still live and why don't I have a legal order in my hand?" In fairness this did happen recently with he.net because a sub-domain was reported but it was done intentionally even if they failed to do even basic due diligence. After Covid I would expect most people would know zoom.us would be in use by a lot of people whereas only specific groups of people would know what he.net is.

I am curious if the process has changed due to laziness and now registrars can just select any number of domains and click a button to place them on hold without management or executive approval. If so that should be in some audit trail and should require confirmation and approval by a senior leader.

What? Weird stuff happens less by definition.

Correctness doesn't scale. If something has six nines of reliability, you'll probably never see the one-in-million outlier yourself. But if the other side deals with a million requests a month, they are a common occurrence.

Yeah I'm not saying errors don't happen. I've been called into gazillions of them including many that "should not happen". Those make for the best root cause analysis and after action reports.

Rather this does not sound like a communication error unless they are leaving out a lot of critical details and context or the domain management interface has been de-frictioned and dumbed down too much.

Not necessarily. The default could happen 49% of the time, and everything else happens way less than 1%, but is weird.

So 51% of the time it’s weird, but not the same weird.

Every place I've been we measured such weirdness outside of the 95'th and 99'th percentile. Anything out of common occurrence beyond the 99'th could be weird or interesting or fascinating. I still wish I could share the incident of a single NIC on a single server taking down an entire data-center, that was both weird and fascinating.

Isn't the stated reason, a miscommunication with the registrar, far far more likely?

If there is a "default" then "everything else" is not weird. The conclusion is "this thing doesn't work most of the time so it wouldn't be weird if it doesn't".

Well, Zoom also lied about their encryption (or, perhaps more charitably, described it in a misleading way. nah, they just lied) and was directing traffic through chinese servers with no reason for doing it -- it was occurring when all meeting participants and the company paying for the zoom account were outside China -- besides enabling spying.

In normal times, absolutely. These aren’t normal times.

Could it have been something as simple as "hey, zoonn.us is violating Zoom's copyright, please block it" and then someone typos "zoom.us".

What's the escalation theory here mean? The US shut it down to damage a company it doesn't like? And 2-4 hours is meaningful? or China did it? Maybe it was shutdown and used as negotiating leverage and brought back when some agreement was reached?

GoDaddy's involvement really makes me believe that it's a genuine screw up.