Hmmmm. First guess would be that Hanlon's Razor applies here.
You've been very active on this site for a long time; perhaps you have just hit an unexpected limit (total time on, total nodes (written or read), something like that?) which the underlying code treats as a less-than-graceful failure. I've seen systems which run perfectly for years and then suddenly all hell breaks loose because a DB field was originally set up as INTEGER instead of BIGINT years ago and finally some user has clocked it.
Have a nice cup of tea, count to 10 and then report it as a bug. We'll see what happens after that.