[Buildd-tools-devel] Bug#485871: Bug#485871: Session Clone is false for lvm-snapshot

[Kees Cook]

On Thu, Jun 12, 2008 at 11:49:22PM +0100, Roger Leigh wrote:
> Kees Cook <kees at outflux.net> writes:
> > Version: 2.0.0-1
> 
> I'm pretty sure this is wrong; 1.3.0 is the latest in git.  What
> version does --version report?

Agh, sorry, I did this bug report manually.  That should be 1.2.0-1.

> > $ schroot --info -c hardy-6caf62a7-5594-410f-841f-8607ffaaa98f
> > ...
> >   Type                   lvm-snapshot
> > ...
> >   Session Cloned         false
> > ...
> >   LVM Snapshot Device    /dev/systemvg/hardy-6caf62a7-5594-410f-841f-8607ffaaa98f
> >
> >
> > This used to be "true", and is causing sbuild to spend time removing
> > build-deps instead of just releasing the schroot normally.
> 
> It turned out that "Session Cloned" was ambiguous, so "Session Purged"
> was added to correct that, to indicate that it would be purged, and so
> cleanup could be skipped.  The version of sbuild in unstable should
> work with the schroot in unstable (and testing), but if you are using
> another distribution, this might not hold (the dependencies could be
> made tighter if this is the case).

Hmm, while I am using Ubuntu primarily, the sbuild version is caught up
with Debian unstable.  (0.57.3-1ubuntu1)  Checking in unstable, I still
see the same old code:

  cleanup_packages:
    if (defined ($session->{'Session Cloned'}) &&
        $session->{'Session Cloned'} == 1) {
        print PLOG "Not removing build depends: cloned chroot in use\n";
    } else {
        uninstall_deps($session);
    }

This was fixed, as you mentioned, here:
http://git.debian.org/?p=buildd-tools/sbuild.git;a=commitdiff;h=9bf1147149e23e9c21ede73a33a4e531cbd0553c

But I think something must have gotten confused, given the published
unstable change:
http://packages.debian.org/changelogs/pool/main/s/sbuild/current/changelog

(If I change those checks to 'Session Purged', things work the way I was
expecting, so this is clearly just a mix-up in sbuild publication, not a
bug in schroot.)

-- 
Kees Cook                                            @outflux.net