And from what I can tell there is still a Bug with 100% CPU usage.
I did have a quick look with strace on the crazy thread:
this repeats very fast. maybe its a indication for whats going on.
(10:21:20 AM) dwd: I know exactly what that bug is, but it'll be tricky to trace it in Java.
(10:21:52 AM) dwd: It's a session that's closed (see the read call returning zero), but we're not closing the socket, and so it's causing epoll() to return immediately.
I think we declared victory over this.