dbus-daemon-1 --system seems to crash

[Bryan Clark]

On Wed, 2004-08-04 at 21:05 +0200, Kay Sievers wrote:
> On Wed, 2004-08-04 at 09:49 -0400, Bryan Clark wrote:
> > On Wed, 2004-08-04 at 09:23 +0000, Kristof Vansant wrote:
> > > Are there known problems with dbus 0.21?
> > > Because it seems to crash with --system
> > > if I start dbus-daemon-1 --system
> > > ps shows no dbus-daemon-1 process running
> > > and if I try to run dbus-daemon-1 again, it complaints that pid file
> > > already exists
> > > can I enable debug or something?
> > 
> > yes, debugging dbus is a little tricky, but it will probably tell you
> > exactly why it isn't starting up.  This is how I debug dbus, maybe
> > someone else has a slightly cleaner system.  This is all done as root,
> > just to be clear.
> > 
> > First backup your system.conf file [*]
> > 	cp /etc/dbus-1/system.conf /etc/dbus-1/system.conf.bak
> > 
> > Edit the system.conf file to remove this section
> > 	<!-- Fork into daemon mode -->
> > 	<fork/>
> 
> Yes, I'm doing this all the time, but it's really bad if you have
> forgotten it to revert, and the init script hangs forever on bootup.
> And every make install overwrites the edited version :(

I had copied my edited file to system.conf.nofork so I can always
overwrite the file that make install applies. Kind of hacky I know ;-)
Which is why I later edited my init script.

> Wouldn't it be nice to have a command line option instead, so the
> initscript can issue 'dbus-daemon-1 --system --fork' and it will always
> work? 'fork=yes|no' may also be a option to overwrite the one in the
> config file.

It actually does have the option --fork, so you can put that in your
init script and not have to worry about this.  There is no option for
overriding when the config file specifies a fork.  You'd then have to
turn init scripts off in your configure to make sure it's not
overwritten in a make install.

Maybe patches would be accepted for this, but I don't know if it's
really necessary since this is a development thing and you can patch
your init script to get around it; I guess it's up to the maintainers.

~ Bryan