I have to wonder if it was a routing issue. If I would have been at my computer during that time a traceroute would have told the tale. If it was a server problem, it seems unlikely that it would have fixed itself.