Run process with timeout

[Micah Elliott]

On Oct 17, Alex Martelli wrote:
> Natan <natanvivo at gmail.com> wrote:
> > I have a python script under linux where I poll many hundreds of
> > interfaces with mrtg every 5 minutes. Today I create some threads and
> > use os.system(command) to run the process, but some of them just hang.
> > I would like to terminate the process after 15 seconds if it doesn't
> > finish, but os.system() doesn't have any timeout parameter.
> 
> Use the subprocess module.  With a subprocess.Popen object, you can
> ... kill it (use its .pid attribute).

The problem I've run into with this approach is the inability to kill
the pid's children.  Most often I'm not so fortunate to be able to
depend on the process to not be doing its own forking.  So here's a
simplified use case:

   $ cat sleep10.sh
   #! /bin/bash
   sleep 10  # does not get killed
   $
   $ cat to3.py
   #! /usr/bin/env python

   from subprocess import Popen
   from time import sleep
   from os import kill
   from signal import SIGTERM

   p = Popen(['./sleep10.sh'])
   sleep(3)
   kill(p.pid, SIGTERM)     # Oops, won't kill p.pid's children.
   ##kill(-p.pid, SIGTERM)  # Won't work since not group leader,
                            #  and I'd rather avoid fork/dup/exec.
   $
   $ ./to3.py
   $ # to3.py finished but sleep 10 still running 

If you try this you will see that sleep10.sh gets killed, but its 
"sleep 10" subprocess does not, and runs for an additional 7 seconds.

I presently rely on an ugly script to do this properly.  It uses low
level calls such as pipe, close, dup2, fork, exec, setpgrp, etc.  I
won't post that here for brevity's sake (unless requested).  For this
case it would fork/exec sleep10.sh, make it a group leader, and the
parent would kill its group.

Is there any way to enable Python's subprocess module to do (implicit?)
group setup to ease killing of all children?  If not, is it a reasonable
RFE?

-- 
Micah Elliott
<mde at micah.elliott.name>