Not sure on the cause but if you do:

mp.set_start_method('fork', force=True)

There is no semaphore leakage

yeah sure

now it has log, but only the initial one

So the subprocesses are not logged ?

Let me check, see what can be learned ...

Hmm let me check something

i can only see the init log

Let me know if it solved it, if it did I'll make sure we push the RC

Why are you initializing 3 diff Tasks ?

and one experiment takes 40 hours to run, so i let them run in parallel

YEY

seems the not logging problem is back

No sure, here is the code

Wouldn't it make sense to use a single one ?

No sure what O'm seeing here

Nice job