作为一个最小的例子，我想在远程和本地机器上运行“ping website.com”，并将两者的输出并行打印到本地stdout .

我使用多处理来调用两个函数并行运行 . 每个函数都使用subprocess.Popen（“ping ...”） . 如果使用shell = True调用Popen，则仅打印一台机器的输出，最后打印另一台机器的输出 . 如果使用shell = False调用Popen，则输出将并行打印 . 为什么？

最小的例子（shell = True在第一个函数中）：

#!/usr/bin/env python

import multiprocessing
import subprocess
import sys
import rpyc

def run_remote(ip, command, arg):
    '''Connect to a pc using rpyc package, print remote stdout to local stdout.'''
    remote_pc = rpyc.classic.connect(ip) # Connect
    remote_pc.modules.sys.stdout = sys.stdout   # Redirect remote stdout to local stdout
    child = remote_pc.modules.subprocess.Popen([command, arg], shell=True, stdout=remote_pc.modules.subprocess.PIPE)
    while True:
        out = child.stdout.readline()
        if out == '' and child.poll() != None:
            break
        if out != '':
            print "%s" % out
            remote_pc.modules.sys.stdout.flush()

def run_local(command, arg):
    child = subprocess.Popen([command, arg], stdout=subprocess.PIPE)
    while True:
        out = child.stdout.readline()
        if out == '' and child.poll() != None:
            break
        if out != '':
            print "%s" % out
            sys.stdout.flush()

if __name__ == '__main__':
    print "Started."
    remote_ip = "192.168.1.135" # must be running rpyc_classic.py
    # CREATE PROCESSES
    pc1 = multiprocessing.Process(target=run_remote, name="pc1", args=(remote_ip,"ping","yahoo.com"))
    pc3 = multiprocessing.Process(target=run_local, name="pc3", args=("ping","google.com"))
    # START PROCESSES
    pc1.start()
    pc3.start()
    # WAIT FOR PROCESSES
    pc1.join()
    pc3.join()
    print "Done."
python subprocess popen