[systemd-devel] [PATCH] core: make parsing of chkconfig headers conditional

Lennart Poettering lennart at poettering.net
Mon Mar 24 10:58:07 PDT 2014


On Mon, 24.03.14 17:46, Michael Biebl (mbiebl at gmail.com) wrote:

> > Turning off chkconfig support does a), not b). And I'd really prefer if
> > sysv scripts would behave the same way on all systemd installations... I
> > mean that's what we try to do after all, provide a unified interface
> > for developers...
> 
> Since Debian doesn't have chkconfig, sysv init script already behave
> differently when it comes to a), so the point about having sysv init
> scripts behave the same on all systemd installations is moot.

Well, but that's a difference in install time behaviour, not runtime
behaviour...

> It really doesn't make sense to parse the chkconfig headers on distros
> which don't support that interface. The result is worse then not
> parsing it at all.
> I don't mint terribly if you want to keep the # description: parsing
> in, but the # chkconfig: and # pidfile: parsing should not be done on
> distros where does headers never have been used/tested.

So, what about this: what actually breaks for you? the runlevel info or
the priority info? Or something else from the chkconfig header?

We could probably just drop the priority info parsing entirely, since it
is pretty useless these days: as no native units carry a priority number
we cannot make any useful use of the chkconfig priority anyway
anymore... It is only useful to order sysv scripts with such a number
against other sysv scripts with such a number, but not against any other
service...

Hence, given it is broken/useless already, if we drop that specific part
of chkconfig support entirely, would that be enough for your issue as
well?

It's simply that the PID file info in the chkconfig header is just
increadibly useful (since it allows us to identify the main process of a
service) and I'd really like to make sure we make use of it wherever
possible. So that chkconfig header bit is what I am interested in, not
the priority number...

Lennart

-- 
Lennart Poettering, Red Hat


More information about the systemd-devel mailing list