Re: Bug#601455: marked as done (can't stop daemon using /etc/init.d/foo stop when disabled via /etc/default/foo)
On 09/12/13 15:24, Bas Wijnen wrote:
>> From: Holger Levsen <holger@layer-acht.org> I've decided to close
>> this bug, as this "mis-feature" / bug is actually a main
>> characteristic of system V init scripts: they can be+do anything,
>> including showing the behaviour which led to this bug report.
>
> That doesn't make it an unreasonable expectation. I've seen the
> same issue as well, and it always annoyed me.
I don't think a bug against the "general" pseudo-package is going to
solve this. If you think Policy should explicitly state it, I would
suggest opening a bug against Policy (or possibly reopening and
reassigning #601455); or if you think it's a bug already, please open
bugs in individual packages that have it (the original bug reporter
noted mpd and icecast2).
>> If you don't like this, use a modern init system.
>
> When Debian no longer supports sysv init, a reply like this is
> acceptable. Currently it is not.
If we continue to support sysvinit (even if it's as one of several
alternatives), I think we should promote something like "update-rc.d
foo disable" as the "official" way for a sysadmin to disable init
scripts. Disabling things via /etc/default/foo should be deprecated,
and only kept available for backwards compatibility in packages that
already have it (if at all), because it's so easy to get it wrong like
this.
Recent (i.e. jessie) versions of update-rc.d know how to disable the
corresponding systemd unit, too.
S
Reply to: