On Tue 24 Jan 2012 at 08:43PM, Enda O'Connor wrote:
> On 24/01/2012 18:07, Alan Coopersmith wrote:
> >On 01/24/12 10:02 AM, Enda o'Connor - Oracle Ireland - Software Engineer
> >wrote:
> >>Hi
> >>I installed s11 large-server install
> >>then added
> >>pkg install x11/session/xauth x11/server/xvnc solaris-desktop
> >>library/motif
> >>
> >>to get GUI support for oracle db installer
> >>
> >>then added OSC 4.0 via
> >>pkg install ha-cluster-full
> >>
> >>
> >>then configured my cluster.
> >>
> >>Afterwards I notice that /etc/init.d has group of root which causes
> >>pkg verify
> >>to not like things so much, is there an easy way to determine what has
> >>done
> >>this, aside from looping though each pkg and doign pkg contents -m and
> >>looking
> >>for etc/init.d.
> >
> >"pkg search -l /etc/init.d" should narrow down the possibilities quickly.
> >
> >For full automation:
> >
> >pkg search -l -H -o pkg.fmri /etc/init.d | \
> >xargs pkg contents -o group,pkg.fmri -a path=etc/init.d
> >
> >I'd suspect though that's it's something outside of an IPS package that
> >changed
> >it, since IPS shouldn't allow conflicting owners for packages.
> >
>
> at a guess I suspect oracle 11.2.0.3 DB is the culprit, it installs
> -rwxr-xr-x 1 root root 8802 Jan 9 16:42 init.ohasd
> -rwxr-xr-x 1 root root 6722 Jan 9 16:42 ohasd
>
> into /etc/init.d.
> I'd need to redo the install to verify though
FWIW, I've found a dtrace one-liner like the following to be a big help
in tracking things like this down.
# dtrace -wn 'syscall::fchownat:entry /arg3 == 0 && basename(copyinstr(arg1))
== "init.d"/ { stop(); system("ptree %d", pid); system("prun %d", pid)}'
--
Mike Gerdts
Solaris Core OS / Zones http://blogs.oracle.com/zoneszone/
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
