I have a Python library that does some multiprocessing like:

with multiprocessing.Pool(n) as pool:
    results = pool.map(f, stuff)

It runs fine if I use the library from Python, but when I run it in R via reticulate, it just hangs.

I have not been able to reproduce it minimally, however I found that if I add the following prior to the call, it runs successfully:

mp <- reticulate::import("multiprocessing", delay_load = TRUE)

mp$set_start_method('spawn', force = TRUE)

Is this a known issue, with any solutions aside from setting the start method ?

I am in an Ubuntu (v24.04) container on a RHEL (v9.6) host, if that helps.

I am trying to get a reprex, if I manage I will add it.

Edit:

Reticulate version: 1.43.0
R version: 4.5.1