"Expected coroutine function, got method" error when portforwarding

[dhorkel]

Which project are you reporting a bug for?

kr8s

What happened?

When using a jumpbox to forward ports (ultimately for connecting to a database) and error is raised.

from kr8s.objects import Service

service = Service.get("my-service", namespace="my-namespace)
portforward = service.portforward(45667, 45667)

Results in the error:

Exception in thread Thread-1 (run_forever):
Traceback (most recent call last):
  File "C:\Python312\Lib\threading.py", line 1073, in _bootstrap_inner
>>>     self.run()
  File "C:\Python312\Lib\threading.py", line 1010, in run

    self._target(*self._args, **self._kwargs)
  Fi    return run_sync(self.run_forever)()  # type: ignore\.venv\Lib\site-packages\kr8s\portforward.py", line 31, in run_forever
           ^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\dhorkel\code\source\repos\encord_interface\.venv\Lib\site-packages\kr8s\_async_utils.py", line 123, in run_sync
    raise TypeError(f"Expected coroutine function, got {coro.__class__.__name__}")

Anything else?

I have confirmed this error does not occur in 0.19.1 and does occur in both 0.20.2 and 0.20.1. I am using Python 3.12.2 and using a Windows machine.

[jacobtomlinson]

This is almost certainly related to the changes in #551

[jacobtomlinson]

I'm guessing you're calling also portforward.run_forever() which you didn't include in your example. I found some bad sync wrapping in that method that was introduced in #551, should be fixed in #567.

Run forever is a little tricky to test because it does just that, it blocks forever.

[jacobtomlinson]

Just pushed a v0.20.3 tag with this fix.

[dhorkel]

I'm guessing you're calling also portforward.run_forever() which you didn't include in your example. I found some bad sync wrapping in that method that was introduced in #551, should be fixed in #567.

Run forever is a little tricky to test because it does just that, it blocks forever.

I'm actually just running portforward.start() after but saw the error raised before it got there.

Thank you for the quick fix! I'll test it as soon as it goes through.

[jacobtomlinson]

Ah yeah, start() just calls run_forever() in a background thread, so the fix should still be right.