Bug#834298: nis: problem appears to be in invocation in /etc/init.d/nis
* Christoph Terasa <christoph@kohlio.de> [2018-10-17 11:56 +0200]:
> On Tue, 16 Oct 2018 14:56:57 -0600 Joe Pfeiffer <joseph@pfeifferfamily.net>
> wrote:
> > The problem appears to be in line 172 of /etc/init.d/nis:
> > --exec ${NET}/ypbind -- $broadcast ${YPBINDARGS}
> >
> > If I execute this by hand (for my environment, which does not use
> > -broadcast and has empty YPBINDARGS), ie
> >
> > [...]
> >
> > If I delete the -- from the invocation it works.
> >
>
> Good find, though the "--" is signaling the end of the start-stop-daemon
> options, and is not an option to ypbind. Options to the daemon (ypbind) come
> after the "--". Since your options to ypbind are empty anyway, it works in
> your case, but it might fail otherwise. This still looks like a
> bug/regression in the ypbind binary to me, not like a bug in the init.d
> script.
According to the changelog of 3.17.1-3
Drop -no-dbus from YPBINDARGS= in /etc/default/nis.
We no longer build nis with NetworkManager/D-Bus support, so setting
this option made ypbind fail to start. (Closes: #906436)
Running ypbind with no options and an empty YPBINDARGS= works here. So this bug
isn't relevant any more. Please confirm.
I am thinking about packaging
libnsl 1.2.0
libnss_nis 3.0
ypbind-mt
ypserv 2.5
yp-tools 4.2.2
separately to create clean packages. Whatt do you think?
libtirpc has to be updated to minimum 1.0.2 (not only in experimental!). Any co
maintaining would be great ;-) The packaging of Mark Brown aka broonie is
somewhat, hmmmm, old fashioned? The versions are outdated. Anyway FMPOV not
maintainable.
Elimar
--
Never make anything simple and efficient when a way
can be found to make it complex and wonderful ;-)
Reply to: