Re: Enabling/disabling/floating services in runlevels
Le ven 09/04/2004 à 21:35, J.D. Hood a écrit :
> --- Patrice Fortier <Patrice.Fortier@u-bordeaux3.fr> wrote:
> > Which is wrong, as it _is_ configured.
> No it isn't, because a total absence of symlinks is not a valid
Of course it is a valid config.
Just tell me where there is written, on generic sysV doc (ie: not on
debian doc, or the update-rc.d doc), about init/rc?.d that this is not a
valid (forbidden) configuration?
I agree that it is not a valid config for update-rc.d (and only for
_this one_), and I'm saying that it is a bug as it is a valid config
for _everybody else_.
> A total absence of symlinks isn't left alone on upgrade because
> update-rc.d interprets this as the unconfigured state. This is
> by design.
I understand that, but where does this interpretation comes from?
> We really aren't arguing in a circle. We are talking past each other.
> I am talking about how sysv-rc should be configured _today_. The
> people who want to delete rc symlinks are all talking about how they
> _wished_ sysv-rc would work.
No I'm talking about how it is working everywhere I know except on
debian _today_ (and also yesterday).
There admins like me who use this. There are utils like chkconfig,
webminn or even your rcconf that use it.
Don't say that the authors don't know what they're doing, these utils
work perfectly well.
And I'm talking about how I wish _update-rc.d_ should work on upgrade.
> Given that there is one correct way of disabling a service, I
> don't understand why you persist in demanding a second way
> which isn't currently supported.
because there are several correct ways, and you only care about one.
I just want to keep my configuration in /etc after an upgrade.
There is _no_ good reason systematically destroy a config during
AFAIK, the update-rc.d author is not the admin of my servers. So,
please, keep out of my config files!