Re: [Monotone-devel] buildbots

[Justin Patrin]

On 10/25/07, Markus Schiltknecht <address@hidden> wrote:
> Hi,
>
> I promised to have a look at the buildbot results. I didn't have much
> spare time recently, but here's what I can tell up to now.
>
> On Lapo's machines (i386-freebsd6, amd64-freebsd6) the unit test
> refiner_various_counts still fails due to lack of memory. I remember
> measuring somewhere around 50mb peak during that test on my machine
> (linux), Lapo, can you give those machines more memory? If not
> permanently, then at least to check how much memory that test would
> require on your freebsd systems?
>
> Additionally, amd64-freebsd6 fails on these tests, all having to do with
> annotate:
>
>   37 alias_command                                 FAIL (line 61)
>   38 annotate_accidental_clean_merge               FAIL (line 59)
>   41 annotate_file_whose_name_changed              FAIL (line 38)
>   42 annotate_where_line_splits                    FAIL (line 54)
>   43 annotate_where_lineage_depends_on_traversal   FAIL (line 53)
>   44 annotate_where_one_parent_is_full_copy        FAIL (line 49)
>   45 annotate_with_human_output                    FAIL (line 17)
>
>
> These seem to fail for the same reason: for annotate, the argument
> '--brief' has been renamed to '--revs-only'. For some strange reason,
> however, the tests on that machine still use '--brief'. At least
> according to the logs I downloaded, see for example these lines:
>
> > runcmd: /usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/mtn, 
> > local_redir = false, requested = nil
> > alias_command:59: 
> > /usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/mtn --norc 
> > --root=/usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/tester_dir/tests/alias_command
> >  
> > --confdir=/usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/tester_dir/tests/alias_command
> >  --rcfile 
> > /usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/tester_dir/tests/alias_command/test_hooks.lua
> >  
> > --db=/usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/tester_dir/tests/alias_command/test.db
> >  --keydir 
> > /usr/home/bot_mtn/mtn/slave-dir/full-amd64-freebsd6/build/tester_dir/tests/alias_command/keys
> >  address@hidden praise foo --brief --rcfile=extra_rc
> > stdout:
> > 0a5087e6.. by tester 2007-10-25: first
> > 28b894c1.. by tester 2007-10-25: second
> > 0a5087e6.. by tester 2007-10-25: third
> > b7831818.. by tester 2007-10-25: fourth
>
> I'm suspecting there are local modifications, however, I don't know how
> to check that. Maybe we should add a 'mtn status' step just after 'mtn
> update' for the build slaves? Or is there another way to tell for sure
> that we are testing an unmodified revision?
>
> Lacking any of OSX or AIX and given those bots are offline, I don't
> think it makes much sense for me to look into those logs. And the
> Windows bot on Visual Studio... well... that probably needs more work
> than I can do now anyway.
>

Oh yeah, my OSX buildbot. I've brought it back up so that the new
build(s) can be tested.

-- 
Justin Patrin