[Release][Windows] Dashboard agent broken in 1.2.0 release candidate

What is the problem?

In the 1.2.0 release candidate, the dashboard agent seems to be completely broken and spams the console from the background. This infinite loop of errors makes Ray pretty much unusable.

(p37) C:\Users\Administrator>python
Python 3.7.9 (default, Aug 31 2020, 17:10:11) [MSC v.1916 64 bit (AMD64)] :: Anaconda, Inc. on win32
Type "help", "copyright", "credits" or "license" for more information.
>>> import ray
>>> ray.init()
2021-02-10 03:05:34,529 INFO services.py:1174 -- View the Ray dashboard at http://127.0.0.1:8265
{'node_ip_address': '172.31.11.8', 'raylet_ip_address': '172.31.11.8', 'redis_address': '172.31.11.8:6379', 'object_store_address': 'tcp://127.0.0.1:58119', 'raylet_socket_name': 'tcp://127.0.0.1:56755', 'webui_url': '127.0.0.1:8265', 'session_dir': 'C:\\Users\\ADMINI~1\\AppData\\Local\\Temp\\2\\ray\\session_2021-02-10_03-05-33_888831_4556', 'metrics_export_port': 54734, 'node_id': 'dd5091720488d7393b326ddfac13adab6bfb97e048484f5ed87d08d8'}
>>> 2021-02-10 03:05:45,544     WARNING worker.py:1107 -- The agent on node EC2AMAZ-6T8IHN5 failed with the following error:
Traceback (most recent call last):
  File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 312, in <module>
    loop.run_until_complete(agent.run())
  File "C:\ProgramData\Anaconda3\envs\p37\lib\asyncio\base_events.py", line 587, in run_until_complete
    return future.result()
  File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 188, in run
    await asyncio.gather(check_parent_task,
UnboundLocalError: local variable 'check_parent_task' referenced before assignment

(pid=None) Traceback (most recent call last):
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 323, in <module>
(pid=None)     raise e
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 312, in <module>
(pid=None)     loop.run_until_complete(agent.run())
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\asyncio\base_events.py", line 587, in run_until_complete
(pid=None)     return future.result()
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 188, in run
(pid=None)     await asyncio.gather(check_parent_task,
(pid=None) UnboundLocalError: local variable 'check_parent_task' referenced before assignment
(pid=None) --- Logging error ---
(pid=None) Traceback (most recent call last):
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\logging\handlers.py", line 69, in emit
(pid=None)     if self.shouldRollover(record):
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\logging\handlers.py", line 183, in shouldRollover
(pid=None)     self.stream = self._open()
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\logging\__init__.py", line 1116, in _open
(pid=None)     return open(self.baseFilename, self.mode, encoding=self.encoding)
(pid=None) NameError: name 'open' is not defined
(pid=None) Call stack:
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\aiohttp\client.py", line 337, in __del__
(pid=None)     self._loop.call_exception_handler(context)
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\asyncio\base_events.py", line 1645, in call_exception_handler
(pid=None)     self.default_exception_handler(context)
(pid=None)   File "C:\ProgramData\Anaconda3\envs\p37\lib\asyncio\base_events.py", line 1619, in default_exception_handler
(pid=None)     logger.error('\n'.join(log_lines), exc_info=exc_info)
(pid=None) Message: 'Unclosed client session\nclient_session: <aiohttp.client.ClientSession object at 0x0000020B6436E488>'
(pid=None) Arguments: ()
2021-02-10 03:05:47,482 WARNING worker.py:1107 -- The agent on node EC2AMAZ-6T8IHN5 failed with the following error:
Traceback (most recent call last):
  File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 312, in <module>
    loop.run_until_complete(agent.run())
  File "C:\ProgramData\Anaconda3\envs\p37\lib\asyncio\base_events.py", line 587, in run_until_complete
    return future.result()
  File "C:\ProgramData\Anaconda3\envs\p37\lib\site-packages\ray\new_dashboard/agent.py", line 188, in run
    await asyncio.gather(check_parent_task,
UnboundLocalError: local variable 'check_parent_task' referenced before assignment

Ray version and other system information (Python version, TensorFlow version, OS):

Reproduction (REQUIRED)

Please provide a short code snippet (less than 50 lines if possible) that can be copy-pasted to reproduce the issue. The snippet should have no external library dependencies (i.e., use fake or mock data / environments):

Use python 3.7 (for example using conda on the aws windows 2016 deep learning ami). Install the 1.2.0 release candidate with

pip install --index-url https://test.pypi.org/simple/ ray

run

import ray

ray.init()

If the code snippet cannot be run by itself, the issue will be closed with “needs-repro-script”.

• I have verified my script runs in a clean environment and reproduces the issue.
• I have verified the issue also occurs with the latest wheels.

[pcmoritz]

fixed by https://github.com/ray-project/ray/pull/19575

[wuisawesome]

Unfortunately it just doesn’t work on windows. We explicitely disabled it because it to contain the damage, since those errors made the rest of ray unusuable because it was spamming its errors.

Contributions are certainly welcome from people with windows expertise 😉