Re: Bug#652575: rsyslog: /etc/init.d/rsyslog modifications for GNU/Hurd

[Guillem Jover <guillem@debian.org>]

Hi!

On Sun, 2012-02-12 at 14:11:41 +0100, Michael Biebl wrote:
> On 27.01.2012 10:18, Guillem Jover wrote:
> >>  * rsyslog should probably switch to use s-s-d --exec instead (why is
> >>    it using --name anyway? that option has always been more unreliable).
> > 
> > Still pending.
> 
> So, I had another look at it. This seems to be a typical case of
> copy&paste. The rsyslog init script is based on the skeleton file from
> /etc/init.d/ which uses --name in stop.
> 
> Guillem, if --name is unreliable as you suggest, it might make sense to
> update the skeleton file accordingly?

--exec should be in general more reliable yes, but the problem with
it is that it will not work for interpreted scripts, and that's where
--name is useful, but then if the script name is too long it will get
truncated, so it might not match correctly (the current minimum
guaranteed to match is <= 15 chars across any kernel s-s-d supports).

> I guess I just use --exec everywhere, as I don't want to add further
> arch specific checks.

Yes, that'd be ideal.

> Could you please check the attached rsyslog init script, if it works
> properly on hurd?

I skimmed over it and it looked good, didn't have time to test it but
I'd say if it works on GNU/Linux and/or GNU/kFreeBSD then anything else
related to s-s-d --exec not working is a bug in s-s-d.

thanks,
guillem