On Fr, 2008-11-28 at 11:05 +0100, Sven Sandberg wrote:
> Hi Hakan,
>
> Hakan Kuecuekyilmaz wrote:
> > Sven,
> >
> > can you please explain what this push is about?
>
> In 5.1-rpl and 6.0-rpl, we have a new version of mtr that behaves
> slightly differently from the main one. It is generally better (more
> features, better reporting, and faster execution), but has caused huge
> problems for the replication team because it finds bugs in
> non-replication tests that the main version of mtr does not find. We
> have spent the last 6 months fixing both replication and non-replication
> tests. Other teams have been reluctant to fixing bugs in their tests,
> since it looked like we may have broken the server in 6.0-rpl.
Thanks for the update. Sorry that I did not know about the new mtr, yet.
>
> It is of course very inefficient that we in the replication team fix
> non-replication bugs, because we are only experts in replication. I
> created this new tree in order to disprove that the failing
> non-replication tests had anything to do with replication, thus
> motivating other teams to fix bugs in their tests. The new tree contains
> 6.0 main, but with the test framework and all test cases replaced by the
> test framework and test cases from 6.0-rpl. The tree will never be
> pushed to main or anything, it will only be used to show that it is not
> the replication team's responsibility to fix certain bugs.
>
Ok, got the point.
>
> In the first commit you refer to, I removed the mysql-test/ directory
> and client/mysqltest.c, and then copied mysql-test/ and
> client/mysqltest.cc from 6.0-rpl to the tree (and updated some build
> scripts). In the second push, I made an empty commit since I wanted
> pushbuild to build again (most test failures are sporadic, so it wasn't
> guaranteed that they would show up in the first push).
>
Ok.
> >
> > Also, I see a .moved file, which hints to a merge error. See inline for
> > detail
> >
> > On Do, 2008-11-27 at 12:36 +0100, Sven Sandberg wrote:
> >> 2922 Sven Sandberg 2008-11-27
> >> empty commit just to trigger a push in pushbuild
> >>
> >> 2921 Sven Sandberg 2008-11-26
> >> replaced all tests in 6.0 by those from 6.0-rpl
> >> removed:
> >> client/mysqltest.c
> >> mysql-test/create-test-result
> >
> > [cut]
> >
> >> added:
> >> client/mysqltest.cc
> >>
> >
> > [cut]
> >
> >> mysql-test/include/restart_mysqld.inc.moved
> >
> > Where does this .moved file come from? It looks like a result from a
> > merge error.
>
> It does indeed, thanks for spotting this! Since you worked with this
> restart thing (according to bzr annotate :-), would you mind
> investigating which of restart_mysqld.inc and restart_mysqld.inc.moved
> is the correct version?
>
I just checked our Falcon tree and you got the same version as we do.
Please consider to remove the .moved file to save us from future
confusion.
There is another problem: in the rpl tree falcon_bug_28095.test started
to fail, which uses a fairly new feature of mtr called
--shutdown_server. It is used in restart_mysqld.inc to do a soft server
restart, where soft means that the data dir of MySQL is not deleted
between the restart.
This test falcon_bug_28095.test does not fail for us in our trees.
Best regards,
Hakan
--
Hakan Küçükyılmaz, Senior Software Engineer DBTG/MySQL +49 160
98953296
Sun Microsystems GmbH Sonnenallee 1, DE-85551 Kirchheim-Heimstetten
Geschaeftsfuehrer: Thomas Schroeder, Wolfang Engels, Dr. Roland Boemer
Vorsitz d. Aufs.rat.: Martin Haering HRB MUC 161028 49.011, 8.376
