EncodeError: Object of type LimitContext is not JSON serializable

[sentry-io]

Sentry Issue: POSTHOG-N81

TypeError: Object of type LimitContext is not JSON serializable
(3 additional frame(s) were not displayed)
...
  File "__init__.py", line 238, in dumps
    **kw).encode(obj)
  File "json/encoder.py", line 199, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "json/encoder.py", line 257, in iterencode
    return _iterencode(o, 0)
  File "kombu/utils/json.py", line 45, in default
    return super().default(o)
  File "json/encoder.py", line 179, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '

EncodeError: Object of type LimitContext is not JSON serializable
(16 additional frame(s) were not displayed)
...
  File "__init__.py", line 238, in dumps
    **kw).encode(obj)
  File "json/encoder.py", line 199, in encode
    chunks = self.iterencode(o, _one_shot=True)
  File "json/encoder.py", line 257, in iterencode
    return _iterencode(o, 0)
  File "kombu/utils/json.py", line 45, in default
    return super().default(o)
  File "json/encoder.py", line 179, in default
    raise TypeError(f'Object of type {o.__class__.__name__} '

[webjunkie]

@mariusandra this for me breaks async queries. I can take a look later.

[mariusandra]

Oh, this is indeed something I recently added. Tests passed, so didn't expect anything to break. Probably some case to add?

This should fix it according to the internet, but I didn't test it yet: #19102

[webjunkie]

Probably some case to add?

This is in the call to Celery's delay, where Celery puts the arguments into the queue, so hence we bypass it in tests. Would be more on the integration test side of things. I looked it up and not sure how to test, other than manually serializing specifically for this call and checking that it works. But it's not 1:1 what happens inside Celery.

[mariusandra]

Hmm... that does sound involved, and probably better to skip. Thanks for investigating!