Bug: Logger error with exec_info=True

[flatherskevin]

Expected Behaviour

I would expected the output to be

AWS logger
NoneType: None

Current Behaviour

--- Logging error ---
Traceback (most recent call last):
  File "/Users/kevin.flathers/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/logging/__init__.py", line 1160, in emit
    msg = self.format(record)
          ^^^^^^^^^^^^^^^^^^^
  File "/Users/kevin.flathers/.local/share/uv/python/cpython-3.12.7-macos-aarch64-none/lib/python3.12/logging/__init__.py", line 999, in format
    return fmt.format(record)
           ^^^^^^^^^^^^^^^^^^
  File "/Users/kevin.flathers/codebase/cz-common-python/.venv/lib/python3.12/site-packages/aws_lambda_powertools/logging/formatter.py", line 197, in format
    extracted_exception, extracted_exception_name = self._extract_log_exception(log_record=record)
                                                    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/Users/kevin.flathers/codebase/cz-common-python/.venv/lib/python3.12/site-packages/aws_lambda_powertools/logging/formatter.py", line 391, in _extract_log_exception
    return self.formatException(log_record.exc_info), log_record.exc_info[0].__name__  # type: ignore
                                                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
AttributeError: 'NoneType' object has no attribute '__name__'. Did you mean: '__ne__'?
Call stack:
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/Users/kevin.flathers/codebase/cz-common-python/z.py", line 10, in <module>
    aws_logger.warning('AWS logger', exc_info=True)
  File "/Users/kevin.flathers/codebase/cz-common-python/.venv/lib/python3.12/site-packages/aws_lambda_powertools/logging/logger.py", line 544, in warning
    return self._logger.warning(
Message: 'AWS logger'
Arguments: ()

Code snippet

import logging
from aws_lambda_powertools import Logger


python_logger = logging.getLogger(__name__)
aws_logger = Logger()

python_logger.warning('Python logger', exc_info=True)
print('\n\n')
aws_logger.warning('AWS logger', exc_info=True)

Possible Solution

It appears like AWS Power Tools Logger is throwing an error somewhere when exc_info is set to True, but an exception was not raised. This is counter to how the Python logger works, where it just prints NoneType: None

Steps to Reproduce

Run the provided script

Powertools for AWS Lambda (Python) version

latest

AWS Lambda function runtime

3.12

Packaging format used

PyPi

Debugging logs

[boring-cyborg]

Thanks for opening your first issue here! We'll come back to you as soon as we can.
In the meantime, check out the #python channel on our Powertools for AWS Lambda Discord: Invite link

[leandrodamascena]

Thank you for raising this issue to improve the developer experience @flatherskevin!

You're correct. This behavior occurs in Powertools because setting exc_info triggers the addition of exception information to the logging message.

To clarify:

1. When exc_info is set, Powertools attempts to include exception details in the log.
2. We also try to include a comprehensive stack tracer.

I agree that we should test if an exception occurs before adding this key, but I'm curious to hear from you in which scenarios you want to log warning with exc_info=True without a real exception.

Thanks.

[flatherskevin]

@leandrodamascena the use case was mostly just we migrated from the Python logger to the AWS PowerTools Logger, and this situation showed up and started failing. We have fixed the line in our code because using exc_info outside of an exception is not necessarily useful, but I wanted to raise it here as it is a breaking bug from the core Python logging behavior.

[leandrodamascena]

@leandrodamascena the use case was mostly just we migrated from the Python logger to the AWS PowerTools Logger, and this situation showed up and started failing. We have fixed the line in our code because using exc_info outside of an exception is not necessarily useful, but I wanted to raise it here as it is a breaking bug from the core Python logging behavior.

Hi @flatherskevin! Thanks a lot for clarifying the use case and I agree that sense for us to fix this.

If you have some bandwidth, I'd be more than happy to have your input here. But if not, don't worry, I can fix this in the next release, which will be in 15 days.

[waffleshop]

I encountered the same issue when code defers exceptions by storing them in a list, logging them later, then raising them together as an ExceptionGroup.

[leandrodamascena]

I encountered the same issue when code defers exceptions by storing them in a list, logging them later, then raising them together as an ExceptionGroup.

Hey @waffleshop! Can you share a snippet that I can reproduce this error?

[waffleshop]

@leandrodamascena Here is a vague snippet that reproduces my code's situation. I was able to work around the issue by using asyncio.as_completed() instead of asyncio.gather().

import asyncio
from random import randint

from aws_lambda_powertools import Logger

logger = Logger()


async def random_sleep():
    """Simulates IO wait and errors"""
    if randint(1, 5) == 5:
        raise Exception("Simulated exception")

    await asyncio.sleep(randint(1, 5))


async def main():
    """Main function to run the async tasks"""
    results = await asyncio.gather(
        *[random_sleep() for _ in range(20)], return_exceptions=True
    )
    exceptions = []

    for result in results:
        if isinstance(result, Exception):
            logger.exception(result)
            exceptions.append(result)

    if exceptions:
        raise ExceptionGroup("Some tasks failed", exceptions)


if __name__ == "__main__":
    asyncio.run(main())