conflicting information about fork-safety of BatchSpanProcessor

Describe your environment

gunicorn+gevent

What happened?

We are trying to figure out how to add instrumentation while adding the minimum amount of code that we will need to maintain across multiple applications.

Steps to Reproduce

The opentelemetry-python documentation claims that "The BatchSpanProcessor is not fork-safe and doesn't work well with application servers (Gunicorn, uWSGI) which are based on the pre-fork web server model.". The opentelemetry.io website claims that Traces (but not Metrics) export work fine for Gunicorn with multiple workers.

Which is it? This looks inconsistent.

Expected Result

These two pages should be consistent and not look inconsistent.

Actual Result

N/A

Additional context

Could you please update either or both pages to be consistent. If they are already consistent, please make them less confusing so the it does not look inconsistent.

Would you like to implement a fix?

None

Tip

_{React with 👍 to help prioritize this issue. Please use comments to provide useful context, avoiding +1 or me too, to help us triage it. Learn more here.}

Start by reading docs/examples/fork-process-model and the linked opentelemetry.io zero-code Python troubleshooting page, then compare their statements about BatchSpanProcessor, Gunicorn, and worker processes. Confirm the intended fork-safety guidance with the relevant maintainers or existing documentation context. Done means the two pages give consistent, unambiguous guidance.