Exception: KeyError: 140058396972800

[davidparks21]

I'm getting the following exception consistently from hanging_threads:

Exception in thread Thread-6:
Traceback (most recent call last):
  File "/usr/lib/python3.6/threading.py", line 916, in _bootstrap_inner
    self.run()
  File "/usr/lib/python3.6/threading.py", line 864, in run
    self._target(*self._args, **self._kwargs)
  File "/usr/local/lib/python3.6/dist-packages/hanging_threads.py", line 97, in monitor
    new_threads = get_current_frames()
  File "/usr/local/lib/python3.6/dist-packages/hanging_threads.py", line 148, in get_current_frames
    for thread_id, frame in sys._current_frames().items()
  File "/usr/local/lib/python3.6/dist-packages/hanging_threads.py", line 148, in <genexpr>
    for thread_id, frame in sys._current_frames().items()
KeyError: 140058396972800

When I look at that code and run the following two commands I get some odd results:

[thread_id for thread_id, frame in sys._current_frames().items()]
[t.ident for t in threading.enumerate()]

The first one is derived from the hanging_threads code using sys._current_frames(), in that output the thread_id for one or two of the entries is changing every time I run that command while in a debugging session. This appears to be the core of the problem, the thread_id is invalid while building the dictionary. In the second case, using threading.enumerate() the thread IDs remain constant.

I've tested this in Python 3.6 and 3.7.4.

I noted the following issues which reference a bug in sys._current_frames() but seems to suggest it was fixed perhaps in 3.4/3.5 (I'm not really familiar with the bug).

• https://stackoverflow.com/questions/8110920/python-threading-sys-current-frames-vs-threading-enumerate
• https://bugs.python.org/issue17094

At this point I'm a little uncertain how to proceed, my options seem to be:

• Experiment with using threading.enumerate().
• Try to understand the underlying issue with sys._current_frames() better.

I'd love some thoughts from someone with more experience here.

[niccokunzmann]

I would think, go for experimentation with threading.enumerate().
It might be that the thread is deleted very fast. threading.enumerate() should be the save option.

[remdragon]

This is not a bug in Python. It is a race condition trying to compare threads between the output sys._current_frames() vs threading.enumerate(). If a thread exits in between the two calls, you get the crash. I submitted issue #89 with a rewrite of get_current_frames() that fixes the crash. Sorry I don't have the time to create a PR right now.