TEST: samba4.blackbox.dbcheck.release-4-0-0 fails

Andrew Bartlett abartlet at samba.org
Wed Dec 3 13:19:08 MST 2014

On Wed, 2014-12-03 at 16:32 +0100, Jelmer Vernooij wrote:
> On Wed, Dec 03, 2014 at 04:22:12PM +0100, Andreas Schneider wrote:
> > On Thursday 20 November 2014 09:50:07 Andreas Schneider wrote:
> > > Hi,
> > > 
> > > could someone shed some light on what this test is looking for? This test is
> > > failing for me on master.
> See the comments in source4/selftest/tests.sh. Basically, these are meant to make
> sure that nothing mangles the database in horrible ways.

That is correct for the general per-env dbcheck runs, but this one is
slightly different.  It is trying to see that running dbcheck on an old
database still works, and then the ldapcmp is trying to ensure that the
database matches (within reason) what we would get from a new install
right now. 

> I don't like how they are done, because it means we do not find out that something
> dirties the environment until the end of the test run, so we don't get to find
> out which specific test suite is problematic.

This is true for the main dbcheck tests, yes.  We don't find out that we
have 'corrupted' (in the eyes on dbcheck) the DB until the very end.  It
also means the start state for dbcheck isn't well defined - we should
have more corrupt database snapshots to test on.

> How is it failing for you?

The failure          
> msDS-Behavior-Version =>
> ['4']
> ['2']

Is very interesting, as indeed we recently changed the default domain
functional level to 2008R2 in b9701a0a79dd15dd6f53075638fba9a2a3d92e19,
and patched upgradeprovision-oldrelease.sh, but not this file.  Why
doesn't this fail for everyone?


Andrew Bartlett

Andrew Bartlett
Authentication Developer, Samba Team  http://samba.org
Samba Developer, Catalyst IT          http://catalyst.net.nz/services/samba

More information about the samba-technical mailing list