<html> <head> <style></style></head> <body class='hmmessage'><div dir='ltr'><span style="font-size: 12pt;">> The "kill signal" is not sent 5 seconds later, but immediately after </span><br><div>> progressing past the stop/pre-stop state. This is the way upstart <br>> signals to the process that it should exit now. There would be no reason <br>> to delay that signal at all.<br><br></div><div>I am experiencing different because I received the TERM signal while still the pre-stop script was running:</div><div><br></div><div>---------------------------------------</div><div><div>## upstart config</div><div>kill timeout 120</div><div>kill signal SIGCONT</div><div>nice -10</div><div><br></div><div>## start WinXP VirtualBox job</div><div>exec /home/tombert/scripts/winxpvm-start.sh</div><div><br></div><div>## stop WinXP VirtualBox job</div><div>pre-stop exec /home/tombert/scripts/winxpvm-stop.sh</div></div><div>---------------------------------------</div><div><br></div><div>My pre-stop script runs for <120 seconds but it received the SIGTERM after some (five) seconds - thats the reason why I had to change to SIGCONT.</div><div>So in my world delaying the termination makes sense since a virtual XP does not shutdown so quickly ...</div><div><br></div><div>If you say different than I have to verify this behavior again?</div><div><br></div><div><span style="font-size: 12pt;">> </span></div><div>> I'd be more inclined to change 'kill signal' to this:<br>> <br>> stop signal SIGxxx<br>> <br>> And deprecate the usage of 'kill signal'. That would in fact be more <br>> clear.<br><br></div><div>Would also be better than the "kill" terminology.</div><div><br></div><div>thx!</div> </div></body> </html>