21

I have a module A that does a basic map/reduce by taking data and sending it to modules B, C, D etc for analysis and then joining their results together.

But it appears that modules B, C, D etc cannot themselves create a multiprocessing pool, or else I get

AssertionError: daemonic processes are not allowed to have children

Is it possible to parallelize these jobs some other way?

For clarity, here's an (admittedly bad) baby example. (I would normally try/catch but you get the gist).

A.py:

  import B
  from multiprocessing import Pool

  def main():
    p = Pool()
    results = p.map(B.foo,range(10))
    p.close()
    p.join()
    return results


B.py:
  
  from multiprocessing import Pool

  def foo(x):
    p = Pool()
    results = p.map(str,x)
    p.close()
    p.join()
    return results
Emi OB
  • 2,814
  • 3
  • 13
  • 29
rhombidodecahedron
  • 7,693
  • 11
  • 58
  • 91

1 Answers1

31

is it possible to have a pool inside of a pool?

Yes, it is possible though it might not be a good idea unless you want to raise an army of zombies. From Python Process Pool non-daemonic?:

import multiprocessing.pool
from contextlib import closing
from functools import partial

class NoDaemonProcess(multiprocessing.Process):
    # make 'daemon' attribute always return False
    def _get_daemon(self):
        return False
    def _set_daemon(self, value):
        pass
    daemon = property(_get_daemon, _set_daemon)

# We sub-class multiprocessing.pool.Pool instead of multiprocessing.Pool
# because the latter is only a wrapper function, not a proper class.
class Pool(multiprocessing.pool.Pool):
    Process = NoDaemonProcess

def foo(x, depth=0):
    if depth == 0:
        return x
    else:
        with closing(Pool()) as p:
            return p.map(partial(foo, depth=depth-1), range(x + 1))

if __name__ == "__main__":
    from pprint import pprint
    pprint(foo(10, depth=2))

Output

[[0],
 [0, 1],
 [0, 1, 2],
 [0, 1, 2, 3],
 [0, 1, 2, 3, 4],
 [0, 1, 2, 3, 4, 5],
 [0, 1, 2, 3, 4, 5, 6],
 [0, 1, 2, 3, 4, 5, 6, 7],
 [0, 1, 2, 3, 4, 5, 6, 7, 8],
 [0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
 [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10]]

concurrent.futures supports it by default:

# $ pip install futures # on Python 2
from concurrent.futures import ProcessPoolExecutor as Pool
from functools import partial

def foo(x, depth=0):
    if depth == 0:
        return x
    else:
        with Pool() as p:
            return list(p.map(partial(foo, depth=depth-1), range(x + 1)))

if __name__ == "__main__":
    from pprint import pprint
    pprint(foo(10, depth=2))

It produces the same output.

Is it possible to parallelize these jobs some other way?

Yes. For example, look at how celery allows to create a complex workflow.

jfs
  • 399,953
  • 195
  • 994
  • 1,670