I found this particular section in the Python multiprocessing docs helpful. The following program:
from multiprocessing import Process, Queue, current_process
import time
def f(q):
name = current_process().name
config = q.get()
print "%s got config: %s" % (name, config)
print "%s beginning processing at %s" % (name, time.asctime())
time.sleep(5)
print "%s completing processing at %s" % (name, time.asctime())
if __name__ == '__main__':
q = Queue()
processes = []
cfg = { 'my' : 'config', 'data' : 'here' }
for i in range(3):
p = Process(target=f, args=(q,))
processes.append(p)
p.start()
q.put(cfg)
for p in processes:
p.join()
demonstrates the main script (which is the "multiprocessing manager" in your question) creating 3 processes and sending them each a configuration (shown here as a dictionary).
Each process reads the configuration, does its processing (here, just sleep for 5 secs) then terminates. The output from running this script is:
Process-1 got config: {'my': 'config', 'data': 'here'}
Process-1 beginning processing at Tue Jun 23 23:34:23 2009
Process-2 got config: {'my': 'config', 'data': 'here'}
Process-2 beginning processing at Tue Jun 23 23:34:23 2009
Process-3 got config: {'my': 'config', 'data': 'here'}
Process-3 beginning processing at Tue Jun 23 23:34:23 2009
Process-1 completing processing at Tue Jun 23 23:34:28 2009
Process-2 completing processing at Tue Jun 23 23:34:28 2009
Process-3 completing processing at Tue Jun 23 23:34:28 2009