How to implement multi-threads Java RMI with shell commands by python?

Question

I used Java to implement a distributed algorithm Birman-Schiper-Stephenson . Every time I run this program, I need to open three terminals and type command java Server 2 , java Client 0 2 , java Client 1 2 separately in three terminals to activate a server and two clients.

Now I want this process to be automatic and I try to use python to write a script to run those java files automatically. I tried the code as below:

import os
import thread

thread.start_new_thread(os.system('java Server 2'))
try:
    thread.start_new_thread(os.system('java Client 0 2'))
    thread.start_new_thread(os.system('java Client 1 2'))
except:
    print "Error: unable to start thread"

However, there's something wrong with the multithreaded process. I have to terminate the Server process then the Client process can be activated. Can anybody tell me how to implement this multiprocess with shell commands by python?

Answer 1

os.system blocks, so each thread you spawn is going to block the parent thread until execution is done (which, in the case of your java Server call, probably means it will never exit).

A simple way to get around this is to use the subprocess module (which is recommended for executing stuff you'd use os.system for).

import subprocess

commands = [
    'java Server 2',
    'java Client 0 2',
    'java Client 1 2'
]

# using subprocess.Popen will run sub-processes in parallel
java_procs = [subprocess.Popen(command, shell=True) for command in commands]

# now wait for the processes to finish?
for p in java_procs: p.wait()

If you need more serious multiprocessing/parallel process execution, the multiprocessing module is usually a better recommendation.