Question

Recently in a project, I had a multiprocessing Process that crashed. A child process was supposed to do a calculation, then send it to the parent process in a Pipe. If the child crashed, the parent would freeze while reading from the pipe. Is there a ‘correct’ way to send data that would avoid blocking the parent forever if a child dies?

This is an example that reproduces the problem I’m having:

import multiprocessing as mp

def f(pipe):
    a = 1/0
    pipe.send('hola')

parent, child = mp.Pipe()

proc = mp.Process(target=f, args=(child,))

proc.start()

print "Grabbing result"
print "Result: {0}".format(parent.recv())

proc.join()

Answer 1

The parent process could use the connection’s poll(...) method to determine if any result was forthcoming with a reasonable time limit:

import multiprocessing as mp

timelimit = 3

def f(pipe):
    a = 1/0
    pipe.send('hola')

parent, child = mp.Pipe()

proc = mp.Process(target=f, args=(child,))

proc.start()

print "Grabbing result"
if parent.poll(timelimit):
    print "Result: {0}".format(parent.recv())
else:
    print "No data available after {0} seconds...".format(timelimit)

proc.join()

When I run this code I get the following results:

Grabbing result
Process Process-1:
Traceback (most recent call last):
  File "/usr/lib/python2.7/multiprocessing/process.py", line 258, in _bootstrap
    self.run()
  File "/usr/lib/python2.7/multiprocessing/process.py", line 114, in run
    self._target(*self._args, **self._kwargs)
  File "s.py", line 6, in f
    a = 1/0
ZeroDivisionError: integer division or modulo by zero
No data available after 3 seconds...