Sorry about the downtime!
1) Table items_selling became corrupted (it happens)
2) Ran out of disk space, repair failed
3) At some point in the last two months apache got installed, which failed to bind to port 80, but which also rewrote one of the nginx configuration files. When I restarted nginx, apache took over. When I removed apache, the bad conf file broke nginx.
So everything should be fixed right now, it’ll process the 70,000 queued uploads over the next few hours. Basically I need to throw more money at the server, or I need to remove the history database download from the developers section, or I need some IRL cash coming in to pay for upgrades :)