On Mon, Jul 07, 2014 at 05:40:42PM +0200, Lennart Poettering wrote: > On Mon, 07.07.14 11:08, Leonid Isaev ([email protected]) wrote: > > > Hi, > > > > Thanks for the explanation... > > > > On Mon, Jul 07, 2014 at 12:26:03PM +0200, Lennart Poettering wrote: > > > I wasn#t aware of grpck, and quite frankly don't think it makes much > > > sense, what the tool is doing. > > > > Why? Checking syntax can never hurt... > > Well, I am not opposed to that. I am just saying that otherwise the > current logic so nicely considers an account with a missing counterpart > in /etc/shadow disabled with no way to log in, which is exactly what we > want here. However, grpck tool breaks that...
Ah, right. So you mean grpck _and_ pwck.
Is this a new systemd-only thinking, or is it something to be taken to the
shadow upstream (because {grp,pw}ck is provided by shadow)?
>
> > > > Does it mean that on each update, a package manager should touch
> > > > /etc/.updated?
> > >
> > > Hmm? No. A package manager should touch /usr after having done its work.
> >
> > Just to see if I understand: this would mean that ConditionNeedsUpdate =
> > .true.
> > for /etc, no? So, we _do_ want systemd-sysusers.service, ldconfig.service,
> > et
> > al. to run on next boot after an update?
>
> Well, the idea is that they are NOPs if they already ran from postinst...
OK. Fixing post_install scripts of every package is probably not going to
happen, so running the above services on-boot seems indeed better.
Cheers,
--
Leonid Isaev
GPG fingerprints: DA92 034D B4A8 EC51 7EA6 20DF 9291 EE8A 043C B8C4
C0DF 20D0 C075 C3F1 E1BE 775A A7AE F6CB 164B 5A6D
pgpyM02us48oK.pgp
Description: PGP signature
_______________________________________________ systemd-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/systemd-devel
