Restarting ocserv doesn't clean up all workers

[Nikos Mavrogiannopoulos]

On Wed, 2014-09-24 at 17:35 +0200, Niels Peen wrote:
> > On 24 Sep 2014, at 16:44, Nikos Mavrogiannopoulos <n.mavrogiannopoulos at gmail.com> wrote:
> > 
> > Could you provide the log output of such cases? Unless killed with
> > SIGKILL ocserv shouldn't have left children hanging.
> 
> I use the TERM signal to kill the main ocserv process (using the pid file).
> 
> Prior to restart:
> 
> root     15305  0.0  0.3  53432  6484 ?        Ss   05:24   0:02 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf
> root     15306  0.0  0.1  48704  2144 ?        S    05:24   0:00 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf
> nobody   27889  0.0  0.0  48748  1592 ?        S    21:40   0:03 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf
> 
> After restart:
> 
> [ ok ] Restarting ocserv: ocserv.
> nobody   27889  0.0  0.0  48748  1592 ?        S    21:40   0:03 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf
> root     29495  0.0  0.0  48696  1372 ?        Ss   23:25   0:00 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf
> root     29497  0.0  0.0  48704  1404 ?        S    23:25   0:00 /usr/local/sbin/ocserv -c /etc/ocserv/ocserv.conf

There is a ps flag that prints the process name rather than argv so it
would be more clear what type of ocserv process is that.

> 27889 should’ve been killed but wasn’t. To kill it manually at this point I need to use SIGKILL - TERM won’t work.
> The log doesn’t show the shutdown, only the (re)start:

It is printed on debug level. I've modified it now to print it on info
level, but you should see something like:
"main: termination request received; waiting for children to die"

Seeing it again there may be an issue in the way waitpid() is handled.
That should fix it:
http://git.infradead.org/ocserv.git/commitdiff/accdb24050a1de06c0408c9d783aa0575e35e831

regards,
Nikos