Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.python > #10674 > unrolled thread

Notifications when process is killed

Back to article view | Back to comp.lang.python

Contents

  Notifications when process is killed Andrea Di Mario <anddimario@gmail.com> - 2011-08-01 17:39 +0200
    Re: Notifications when process is killed AndDM <anddimario@gmail.com> - 2011-08-02 00:30 -0700
      Re: Notifications when process is killed Thomas Rachel <nutznetz-0c1b6768-bfa9-48d5-a470-7603bd3aa915@spamschutz.glglgl.de> - 2011-08-02 10:26 +0200
        Re: Notifications when process is killed Kushal Kumaran <kushal.kumaran+python@gmail.com> - 2011-08-02 16:45 +0530
        Re: Notifications when process is killed Thomas Rachel <nutznetz-0c1b6768-bfa9-48d5-a470-7603bd3aa915@spamschutz.glglgl.de> - 2011-08-02 13:40 +0200
      Re: Notifications when process is killed Chris Angelico <rosuav@gmail.com> - 2011-08-02 10:01 +0100
    Re: Notifications when process is killed chrisallick <chrisallick@gmail.com> - 2011-08-03 20:41 -0700

#10674 — Notifications when process is killed

Thanks Thomas, it is what i'm looking for.

Regards

-- 
Andrea Di Mario

[toc] | [next] | [standalone]

#10698

On Aug 1, 5:39 pm, Andrea Di Mario <anddima...@gmail.com> wrote:
> Thanks Thomas, it is what i'm looking for.
>
> Regards
>
> --
> Andrea Di Mario

Hi, i've a little problem, here the code that i use:

        def receive_signal(signum, stack):
                logging.info('Received: %s' % signum)
                reactor.stop()
        signal.signal(signal.SIGTERM, receive_signal)
        signal.signal(signal.SIGHUP, receive_signal)
        signal.signal(signal.SIGINT, receive_signal)

The function works for SIGHUP and SIGINT, but it doesn't work for
SIGTERM. I've tried with simple killall and with -15 option.
Have you some ideas?

Thanks, regards.

[toc] | [prev] | [next] | [standalone]

#10701

Am 02.08.2011 09:30 schrieb AndDM:

> The function works for SIGHUP and SIGINT, but it doesn't work for
> SIGTERM. I've tried with simple killall and with -15 option.
> Have you some ideas?

SIGTERM cannot be caught - it kills the process the hard way.

HTH,


Thomas

[toc] | [prev] | [next] | [standalone]

#10718

On Tue, Aug 2, 2011 at 1:56 PM, Thomas Rachel
<nutznetz-0c1b6768-bfa9-48d5-a470-7603bd3aa915@spamschutz.glglgl.de>
wrote:
> Am 02.08.2011 09:30 schrieb AndDM:
>
>> The function works for SIGHUP and SIGINT, but it doesn't work for
>> SIGTERM. I've tried with simple killall and with -15 option.
>> Have you some ideas?
>
> SIGTERM cannot be caught - it kills the process the hard way.
>

You must mean SIGKILL.  There's nothing special about SIGTERM, except
that it's the default for the kill command.

-- 
regards,
kushal

[toc] | [prev] | [next] | [standalone]

#10719

Am 02.08.2011 10:26 schrieb Thomas Rachel:
> Am 02.08.2011 09:30 schrieb AndDM:
>
>> The function works for SIGHUP and SIGINT, but it doesn't work for
>> SIGTERM. I've tried with simple killall and with -15 option.
>> Have you some ideas?
>
> SIGTERM cannot be caught - it kills the process the hard way.

Thank you for pointing out (via email) that this is wrong:

SIGTERM is a normal signal which can be caught in the normal way.
SIGKILL is (besides SIGSTOP) the one which cannot be caught, blocked or 
ignored.


Thomas

[toc] | [prev] | [next] | [standalone]

#10702

On Tue, Aug 2, 2011 at 8:30 AM, AndDM <anddimario@gmail.com> wrote:
>        def receive_signal(signum, stack):
>                logging.info('Received: %s' % signum)
>                reactor.stop()
>        signal.signal(signal.SIGTERM, receive_signal)
>        signal.signal(signal.SIGHUP, receive_signal)
>        signal.signal(signal.SIGINT, receive_signal)
>
> The function works for SIGHUP and SIGINT, but it doesn't work for
> SIGTERM. I've tried with simple killall and with -15 option.
> Have you some ideas?

You won't be able to catch SIGTERM, as Thomas said, but if you need to
know what caused a process to end, the best way is to have code in the
parent process to catch SIGCHLD. When the child ends, for any reason,
its parent is sent SIGCHLD with some parameters, including the signal
number that caused the termination; you can then log anything you
want.

Chris Angelico

[toc] | [prev] | [next] | [standalone]

#10836

On Aug 1, 11:39 am, Andrea Di Mario <anddima...@gmail.com> wrote:
> Thanks Thomas, it is what i'm looking for.
>
> Regards
>
> --
> Andrea Di Mario

Catch a Kill:

def signal_handler(signal, frame):
	print "Received exit command."
	#server.running = False
	sys.exit()

signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGTERM, signal_handler)

Find and Kill A Process:

import os, signal

process = "websocket.py"

found = False
for line in os.popen("ps ax | grep python"):
	fields = line.split()
	pid = fields[0]
	for field in fields:
		if field.find(process) >= 0:
			print pid
			print field
			os.kill(int(pid), signal.SIGTERM)
			found = True
			break
	if found == True:
		break

if found == True:
	print "found and killed web server process."
else:
	print "could not find web server process.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.python

csiph-web