I have a problem which I have been beating my head against for weeks now--namely, connecting from one red5 web application to another generates a slew of FIFO pipes that persist after the actual socket has been closed, eventually hitting the per-process open files limit, blocking new connections and hanging around until the process is explicitly killed.
After putting in debugging statements, the number of pipes jumps up right after the creation of an NioSocketConnector. The function that creates it is here: http://pastebin.com/m46bec4bd The NioSocketAcceptor is created here: http://pastebin.com/m6e3227ef Does it look like anything is missing? Should the SocketConnector be created differently? Your help would be greatly appreciated. Thank you, MLH
