Raul Miller writes: > > Only works if you're guaranteed that there's only one running instance > > of the daemon.
John Hasler <[EMAIL PROTECTED]> wrote: > ??? > > hasler/~ pidof xterm > 30423 30422 30421 30420 30419 > hasler/~ xterm is not a daemon > > I seem to recally either policy or the packaging manual recommending > > against using pidof for start/stop scripts. > > Can't find any mention of it. Are you thinking of killall? Probably. However, I was assuming that the pidfile is only relevant for determining which process to kill (at which point it should be wiped out -- but that's a detail we might be overlooking in some packages.). [I suppose it could be used as a loose check to see whether the daemon is still running, but that suffers a similar problem.] If there's no need for the pidfile to uniquely represent the daemon process, then there's no need for the pidfile. -- Raul
