Re: Full install/removal/upgrade test results available

[Andreas Tille <andreas@an3as.eu>]

On Wed, Nov 17, 2010 at 12:28:54PM -0600, Lucas Nussbaum wrote:
> > I checked for cimg-dev in
> > 
> > http://people.debian.org/~lucas/logs/2010/11/17/cimg-dev.log
> > 
> > and found only one hint of a failure when it says:
> > 
> > -- New version is unstable version: FAILED (UNINSTALLED != 1.3.9-1)
> > 
> > but from reasing the log I can not find any hint for this problem.  Is
> > this possibly one of the cases where some dependencies show a problem?
> > Any help reading the log properly is welcome.
> 
> So, that one is an interesting case, and there are several occurences of
> this. It might be a problem in APT. What happens is that, if you install
> cimg-dev on lenny, and then upgrade to squeeze, cimg-dev gets removed.
> But after the upgrade, you can safely reinstall it. It might be caused
> by the handling of Breaks in APT. I'm not sure.

So I have no idea what to do except waiting for other hints or somebody
confirming that APT has a problem.
 
> > >    dict-wn (U)
> > 
> > http://people.debian.org/~lucas/logs/2010/11/17/dict-wn.log
> > 
> > ...
> >    Installing new version of config file /etc/dictd/dictd.conf ...
> >    Installing new version of config file /etc/init.d/dictd ...
> >    insserv: warning: current stop runlevel(s) (0 1 6) of script `dictd' overwrites defaults (1).
> >    invoke-rc.d: policy-rc.d denied execution of start.
> > -- Finding version of dict-wn: 1:3.0-18
> > -- New version is unstable version: OK
> > E: 10mount: umount: /tmp/lucas/schroot/mount/llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b/dev: device is busy.
> > E: 10mount:         (In some cases useful info about processes that use
> > E: 10mount:          the device is found by lsof(8) or fuser(1))
> > E: llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b: Chroot setup failed: stage=setup-stop
> > Cleanup of chroot llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b failed. Running processes:
> > /dc/debcluster/scripts/tasks/instest.rb:5: command not found: lsof -t +D /tmp/lucas/schroot/mount/llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b
> > E: 10mount: umount: /tmp/lucas/schroot/mount/llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b/dev: device is busy.
> > E: 10mount:         (In some cases useful info about processes that use
> > E: 10mount:          the device is found by lsof(8) or fuser(1))
> > E: llenny64-instest-0fa5a864-f08b-447b-955e-c3b7b7befe5b: Chroot setup failed: stage=setup-stop
> > 
> > This does look like a bug rather in dict than in dict-wn.  However, if
> > my assumption would be correct I would expect all other dict
> > dictionaries to fail and thus I wonder if I missinterpreted this log.
> 
> It seems that after installation of dict-wn, there's a process that is still
> running, so destroying the chroot fails.

I'm quoting the relevant part of dict-wn postinst resp. postrm:

     configure)
        if [ -x /usr/sbin/dictdconfig ]; then dictdconfig -w ;fi

        # if [ -x /etc/init.d/dictd ]; then /etc/init.d/dictd restart; fi
        if which invoke-rc.d >/dev/null 2>&1; then
            invoke-rc.d dictd restart
        else
            /etc/init.d/dictd restart
        fi

        exit 0

    remove|purge)
        if [ -x /usr/sbin/dictdconfig ]; then dictdconfig -w ;fi

        # if [ -x /etc/init.d/dictd ]; then /etc/init.d/dictd restart; fi
        if which invoke-rc.d >/dev/null 2>&1; then
            invoke-rc.d dictd restart
        else
            /etc/init.d/dictd restart
        fi
        exit 0


Can you imagine that this might cause the problem?  The other content of
the package is simply a dictionary.  So either the call of dictd init
script is wrong or dictd itself has a problem.  What do you think?

Kind regards

    Andreas.

-- 
http://fam-tille.de