Respect vllm logging configs in vllm_spyre

[kazunoriogata]

Description

vLLM has mechanism to change logging configuration by using environment variables. However, vllm_spyre does not refers to these environment variables, but it always configure python logging based on the hard coded default of vLLM. As a result, changing logging configuration results in mixing log messages of different configurations and generates inconsistent logs.

This PR updates initialization of python logger in vllm_spyer to respect the logging configs and use the same logging configuration as vllm.

[github-actions]

👋 Hi! Thank you for contributing to vLLM support on Spyre.
Just a reminder: Make sure that your code passes all the linting checks, otherwise your PR won't be able to be merged. To do so, first install the linting requirements, then run format.sh and commit the changes. This can be done with uv directly:

uv sync --frozen --group lint --active --inexact

Or this can be done with pip:

uv pip compile --group lint > requirements-lint.txt
pip install -r requirements-lint.txt
bash format.sh

Now you are good to go 🚀

[joerunde]

Thanks @kazunoriogata!

Do you think it's possible to read the config from the vllm logger instead of reloading the file without validation?

[kazunoriogata]

@joerunde Thank you for the comment, but I don't think we can get the configs from vllm logger.

_configure_vllm_root_logger does not save the config dict. The config dict is thrown away after calling dictConfig(). I couldn't find a method to extract logging config or a list of formats/handlers/loggers from an existing logger (e.g., the root logger). So, I think we need to reloading the file.

Since errors like file-not-found or invalid-format should have raised an exception and should not reach the vllm-spyre logger initialization, I think we don't need the same file validation again.

[joerunde]

Nice, thanks for looking into it! I'll get this merged