Application does not emit logs on stderr when auto-instrumented

Describe your environment

OS: (e.g, Ubuntu)
Python version: (e.g., Python 3.8.10)
Instrumentation image: ghcr.io/open-telemetry/opentelemetry-operator/autoinstrumentation-python:0.46b0

What happened?

I have a container running a simple application running on a k8s pod (code attached below).
The application emits the logs messages through stderr when running.

When instrumentation is attached to it, the logs are streamed to the backend, as expected.
Yet, the application stops emitting logs through stderr.

Application code

import logging
import random
import time

random.seed(time.time_ns())

app_name = f'app_{random.randint(1, 100)}'

logger = logging.getLogger(f"{app_name}_logger")

for i in range(3000):
    sleep_ms = 10 * random.randint(1, 100)
    logger.warning(f"Python app '{app_name}' sleeping for {sleep_ms} ms")
    time.sleep(sleep_ms / 1000)

Dockerfile

FROM python:3.9.6

WORKDIR /usr/app
RUN python -m venv /usr/app/venv
ENV PATH="/usr/app/venv/bin:$PATH"

COPY ./requirements.txt .
RUN pip install -r requirements.txt

COPY ./app .

CMD ["python", "-u", "./app.py"]

Steps to Reproduce

Build app, deploy on k8s.

Expected Result

Logs should be streamed through stderr even when the application is instrumented.

Actual Result

Logs are not produced to stderr when the app is instrumented using auto-instrumentation.

Additional context

No response

Would you like to implement a fix?

None

Start with the attached app.py and Dockerfile, deploying the application with the stated Python auto-instrumentation image on a Kubernetes pod. Compare stderr output with and without instrumentation, then trace the auto-instrumentation logging setup. Done means application warning logs continue to appear on stderr while still reaching the backend.