Forums

I know that processes are periodically terminated due to runtime limitations or server reboots. How are the processes terminated? The following code should print a message to stdout upon exiting. This works when ^C-ing in a Console, but fails when consoles and tasks are killed (by the user or automatically).

A SIGTERM + grace period + SIGKILL approach would give long running processes a chance to save state gracefully in most cases. I believe that this is what systemd and docker do.

import atexit
import sleep

def complain_about_termination():
    print('please, not the circle of life *#$! again... :-(')

atexit.register(complain_about_termination)

while True:
    time.sleep(5)

We send a SIGKILL straight away. Perhaps you're right about the TERM, wait, KILL, though. We'll have a think about it.