The definition of "up" is when the website is online and accessable. If not, even if the machine's little bolp continues to give out a nice green strong light, it's considered down.
Responding to requests is up. As nginx is actually returning a HTTP code it's up.
Also quit your whining, it's a free service, you don't like the uptime, go make your own.
Thanks to the admins for what I assume was a chunk of their weekend spent in front of a screen waiting for something to happen.
I felt compelled to come to this guy's defense...just on moral grounds. I use TVDB via SickBeard, so the maintenance affects me too, but whatev's. I've been working in datacenters and random linux admin jobs for about 10 years now -- I understand maintenance and I'm not complaining (just putting that out there.) These guys do a great job and provide an awesome service for free.
That said, maybe you've heard of "POSIWID" - The purpose of a system is what it does. (Basically, all the steps and technology in between do not matter. The final result "is what it does") A future server being powered on, apache/lighty/nginx installed and responding, the page rendering but a database not being available; all of these little things don't matter. In this case, the purpose is to give us a fun free TV show database -- right now it isn't doing that. So with POSIWID in mind, the system isn't doing what it's supposed to do. It's down.
Maybe a plainer exampleis this: I work for a major tech company. You'd have to be under a rock for the last decade not to know who we are. If I have a cluster of servers that are handing out response codes but not serving pages (and therefore not generating ad revenue) we do not consider such a situation as "up". If certain features are available but others are not, we'd refer to that as degraded but still not up.
Hopefully that clarifies things a bit. The difference between machines or certain services being "up" versus the site functioning in it's regularly expected capacity.