From: Jeremy Hinegardner <jeremy@...>
Date: 2007-09-08T16:48:26+09:00
Subject: Re: Why don't threads get killed in this case?

Hi Chad,

On Sat, Sep 08, 2007 at 11:30:04AM +0900, grocery_stocker wrote:
> Here is the code that is running on OpenBSD 4.2
> 
> #!/usr/local/bin/ruby
> 
> threads=[]
> 
> 4.times do |i|
>         threads[i]=Thread.new do
>         %x{/usr/libexec/getty std.19200 tty00}
>         end
> end
> 
> threads.each{|thr| thr.join}
> 
> I run it as a unix background job, I get the following
> 
> -bash-3.2$ ./whore.rb &
> [1] 11801
> -bash-3.2$ ps waux | grep djdoboy
> djdoboy  15158  0.0  0.3   844  1540 pi  Ss     9:41PM    0:00.05 -
> bash
> (bash)
> djdoboy  11801  0.0  0.4   996  1872 pi  S     10:03PM    0:00.01
> /usr/local/bin/ruby ./whore.rb
> djdoboy   4493  0.0  0.1   372   556 pi  S     10:03PM    0:00.02
> /usr/libexec/getty std.19200 tty00
> djdoboy  19020  0.0  0.1   332   556 pi  S     10:03PM    0:00.01
> /usr/libexec/getty std.19200 tty00
> djdoboy    456  0.0  0.1   364   564 pi  S     10:03PM    0:00.02
> /usr/libexec/getty std.19200 tty00
> djdoboy  11853  0.0  0.1   260   584 pi  S     10:03PM    0:00.01
> /usr/libexec/getty std.19200 tty00
> djdoboy  23871  0.0  0.1   532   408 pi  R+    10:03PM    0:00.00 ps
> -waux
> djdoboy  27426  0.0  0.1   220   524 pi  S+    10:03PM    0:00.00 grep
> djdoboy
> 
> Now I get and relogin
> -bash-3.2$ exit
> logout
> Connection closed by foreign host.
> [cdalten@localhost ~]$ telnet
> 
> -bash-3.2$ ps waux | grep djdoboy
> djdoboy   6724  0.0  0.3  1012  1548 p6  Ss    10:04PM    0:00.02 -
> bash
> (bash)
> djdoboy  25303  0.0  0.1   576   448 p6  R+    10:04PM    0:00.00 ps
> -waux
> djdoboy   8516  0.0  0.1  1012   408 p6  R+    10:04PM    0:00.00 grep
> djdoboy (bash)
> djdoboy  11801  0.0  0.4   996  1872 pi- S     10:03PM    0:00.01
> /usr/local/bin/ruby ./whore.rb
> djdoboy  11853  0.0  0.1   260   584 pi- I     10:03PM    0:00.01
> /usr/libexec/getty std.19200 tty00
> djdoboy    456  0.0  0.1   364   564 pi- I     10:03PM    0:00.02
> /usr/libexec/getty std.19200 tty00
> djdoboy   4493  0.0  0.1   372   556 pi- I     10:03PM    0:00.02
> /usr/libexec/getty std.19200 tty00
> djdoboy  19020  0.0  0.1   332   556 pi- I     10:03PM    0:00.01
> /usr/libexec/getty std.19200 tty00
> 
> And the threads are still there. How come ruby threads don't get
> killed
> on exit? And more to the point, how do I kill them on exit?
> 

This is not a ruby isssue, this is related to your shell options.

% uname -rs
 OpenBSD 4.0
% find ./ -type f > /dev/null &
[1] 6962
% exit

# log back in

% ps -p 6962
 PID TT   STAT      TIME COMMAND
6962 p4  D       0:00.95 find ./ -type f
% kill 6962
% shopt huponexit
 huponexit       off

man bash: 

       If the huponexit shell option has  been  set  with  shopt,
       bash  sends a SIGHUP to all jobs when an interactive login
       shell exits.

So lets try it.

% shopt -s huponexit
% find ./ -type f > /dev/null &
[1] 29709
% ps -p 29709
  PID TT   STAT      TIME COMMAND
29709 p4  D       0:01.65 find ./ -type f
% exit

 # log back in

% ps -p 29709
 PID TT   STAT      TIME COMMAND
%

enjoy,

-jeremy

-- 
========================================================================
 Jeremy Hinegardner                              jeremy@hinegardner.org