On Fri, 2002-03-01 at 01:48, Frank Cusack wrote:
> On Thu, Feb 28, 2002 at 10:17:47AM -0800, Frank Cusack wrote:
> > On Thu, Feb 28, 2002 at 02:37:41PM -0300, Eduardo Roldan wrote:
> > > Why the included init script(/redhat/rc.radiusd-redhat) takes the
> > > arguments restart and reload as the same?
> > >
> > > I think that the restart command should do a STOP, START. And the reload
> > > should send a HUP signal to the radiusd daemon to reload the
> > > configuration files. I have changed my script to reflect this. There are
> > > any drawbacks to make this the default behavior in the distributed
> > > script?
> >
> > Yeah, that should be the default behaviour. I'll commit a fix.
>
> I lied. :-) radiusd does not write a pidfile (even though it does parse
> this arg from the config file). I am not sure how it will respond to
> HUP on systems like Linux where each thread is seen as a process by most
> utilities (like ps). So... I'll leave it for someone else to look at.
When I send a HUP signal to the radiusd threads nothing happens.
So, i think that sending a HUP signal to these threads isn't bad.
If we add the following statements in the "case" clause to the redhat
init script:
reload)
echo -n 'Sending Signal to RADIUSD server: '
killproc $RADIUSD -1
echo
;;
this calls the pidofproc() function that finnally use the pidof program
to get the pid of a program. This funciton returns a list of pids, like:
5873 5872 5871 5870 5869 5868 5867
If you think that sending a HUP singal to the radiusd threads isn't
safe, you always can extract the minimum pid of this list. That list is
the main proccess. I'm correct?
The default beahvoir of the script is send a HUP signal to all proccess
returned by pidofproc.
So Frank, be a good boy and commit that fix. Can you? :)
-
List info/subscribe/unsubscribe? See http://www.freeradius.org/list/users.html
