WebSocketProtocol13.periodic_ping exception

We are running a Tornado TCP server on a Windows server 2012 machine. This machine handles both an HTTP-based dashboard and 21 TCP Websocket connected clients. Occasionally in the logs we will see the following error. It seems to be at a very low level of the Websocket13 protocol. But from a networking standpoint what does it mean? Why would the write buffer not be greater than or equal to 0?

[ERROR] [2017-08-17 11:54:14 AM] [tornado.application] Exception in callback <bound method WebSocketProtocol13.periodic_ping of <tornado.websocket.WebSocketProtocol13 object at 0x01D7AD30>>

Traceback (most recent call last):

File "C:\Python27\lib\site-packages\tornado\ioloop.py", line 1026, in _run

    return self.callback()

File "C:\Python27\lib\site-packages\tornado\websocket.py", line 1030, in periodic_ping

    self.write_ping(b'')

File "C:\Python27\lib\site-packages\tornado\websocket.py", line 790, in write_ping

    self._write_frame(True, 0x9, data)

File "C:\Python27\lib\site-packages\tornado\websocket.py", line 768, in _write_frame

    return self.stream.write(frame)

File "C:\Python27\lib\site-packages\tornado\iostream.py", line 406, in write

    self._handle_write()

File "C:\Python27\lib\site-packages\tornado\iostream.py", line 847, in _handle_write

    assert self._write_buffer_size >= 0

AssertionError

Start with the traceback locations in tornado/websocket.py, especially WebSocketProtocol13.periodic_ping and _write_frame, then follow the write path into tornado/iostream.py and _handle_write. Reproduce or trace the Windows server scenario to determine why _write_buffer_size becomes negative, and consider the issue resolved when the cause and an appropriate fix are verified.