Re: Monit doesn't detach from the tty

[Rory Toma]

Which brings up 2 more questions...

1) why use _exit(0) instead of exit(0)?

2) I noticed that we also keep fd 0, 1 and 2 open. Generally, I've used
dup2 to copy 0 1 and 2 to and closed them off.


On Wed, 2002-08-07 at 17:23, Mark Ferlatte wrote:
> Heh, I'm back, fresh from the umask issue (BTW, I'm running a copy of
> monit that does umask save at startup/restore at spawn, and it, at
> least, does what I want... ;), and found another one.
> 
> monit 2.5, when run in daemon mode, doesn't actually detach from the
> tty, which results in (at least) ssh having trouble when you try and log
> out.
> 
> Empirical evidence:
> 
> protagonist:~# fuser /dev/pts/6
> /dev/pts/6:           3453  3662
> protagonist:~# /etc/init.d/monit start
> Starting daemon monitor: monit.
> protagonist:~# fuser /dev/pts/6
> /dev/pts/6:           3453  3667  3675
> protagonist:~# ps ax | grep 3667
>  3667 ?        S      0:00 /usr/sbin/monit -c /etc/monit/monitrc
>  3679 pts/6    S      0:00 grep 3667
> 
> 
> Upon looking in daemonize.c:daemonize(), it looks like either you need
> to fork() again after setsid(), or do
> 
> fd = open("/dev/tty", O_RDONLY);
> if (fd != -1) {
>       if (ioctl(fd, TIOCNOTTY) == -1) {
>               /* print some error about being unable to detach from
>                  tty properly */
>       }
>       (void) close(fd);
> }
> 
> I have a feeling that the double fork method is much more portable, so
> you probably want to do that instead of the ioctl.
> 
> M
-- 
Rory Toma               address@hidden
VP of Run Level 5       http://www.trs80.net
Digeo Digital           http://www.digeo.com

signature.asc
Description: This is a digitally signed message part