Re: downtimed and start-stop-daemon.
tisdag den 3 april 2012 klockan 10:06 skrev Samuel Thibault detta:
> Mats Erik Andersson, le Tue 03 Apr 2012 04:00:41 +0200, a écrit :
> > tisdag den 3 april 2012 klockan 02:43 skrev Samuel Thibault detta:
> > > > I now find that "invoke-rc.d" is not handling the daemonisation
> > > > properly: a restart will leave the old daemon running, but still start
> > > > a new instance. Is there some knowledge I should acquire on this
> > > > matter?
> > >
> > > IIRC it's simply a yet-not-investigated bug somewhere in what
> > > invoke-rc.d uses.
> > One reason is that "invoke-rc.d" cannot handle the outcome
> > $ /sbin/runlevel
> > unknown
> > with GNU/Hurd. That disability is known to relate to #497664,
> > which the sysvinit maintainers are notoriously ignoring.
> ignoring? The bug is currently marked as closed. If you expect
> something to be done about it, reopen the bug, otherwise it'll be out of
> anybody's radar.
Sorry, number off by one:
#497665: invoke-rc.d: no-op test for failure to determine runlevel
The report even contains a patch suggestion.
A further investigation displays that the stop call using start-stop-daemon,
as implemented in the init-script, is also malfunctioning with GNU/Hurd.
It is the standard form
s-s-d --stop --quiet --retry=TERM/30/KILL/5 --pidfile $PID --name $NAME
but it returns status 1 in spite of the executable running and the pid-file
being in existence with correct value. Does the problem seem familiar?
There is no functionality flaw with GNU/Linux or with GNU/kFreeBSD.
Mats E A