APM + Flask + Multiprocessing = No Spans

Elastic Observability APM
1

I am trying to capture some spans in my Python Flask app using the decorator:
@elasticapm.capture_span()

This works perfectly, until I execute a couple of functions in parallel using python's multiprocessing module. These spans are then never captured (the spans set in the main process are still captured). Within these functions, I am also not able to send an error or message (apm.capture_message()).

I use multiprocessing.Process and multiprocessing.Pipe to parallelize the execution of a list of functions.

I would love for each of these functions to be captured as individual spans within the request transaction.

Does anyone have experience with capturing parallel spans using Flask and Elastic APM?

2

Hi @nagellack

unfortunately, this is not something we currently support. The information about ongoing transactions/spans is stored in a thread local (on Python 3.7 it's in a contextvar variable, but that amounts to to more or less the same thing in this case).

I will look into ways to achieve this, but it's a non-trivial problem and I can't say when we might support this use case.

3

Hey - Thanks for the response.

I look forward to it - possibly - being supported!