On Tue, Sep 03, 2013 at 02:16:24PM -0400, Simo Sorce wrote:
On Tue, 2013-09-03 at 13:13 +0200, Pavel Březina wrote:
> On 09/02/2013 09:42 AM, Sumit Bose wrote:
> > On Sun, Sep 01, 2013 at 05:34:23PM -0400, Simo Sorce wrote:
> >> The test named test_ss_idmap, takes quite long on my machine (as in
> >> minutes) although the machine is completely idle. Any idea what causes
> >> it ? Any way to avoid it ?
> >
> > sorry, I cannot reproduce this
> >
> > $ /usr/bin/time ./sss_idmap-tests
> > Running suite(s): IDMAP
> > 100%: Checks: 20, Failures: 0, Errors: 0
> > 0.00user 0.00system 0:00.02elapsed 40%CPU (0avgtext+0avgdata 1376maxresident)k
> > 0inputs+160outputs (0major+6696minor)pagefaults 0swaps
> >
> > Does strace gives any indication where the time is spend? Fell free to
> > just send me the strace output.
> >
> > bye,
> > Sumit
> >
> >>
> >> Simo.
>
> Hi,
> there are two idmap tests:
> test_sss_idmap - cmocka based
> sss_idmap-test - check based
>
> Simo was complaining about the first one.
>
> Anyway, both tests takes few miliseconds on my machine, so Sumit's
> request is still valid.
>
> $ time ./test_sss_idmap
> [==========] Running 18 test(s).
> [ RUN ] test_add_domain
> [ OK ] test_add_domain
> [ RUN ] test_map_id
> [ OK ] test_map_id
> [ RUN ] test_map_id_external
> [ OK ] test_map_id_external
> [ RUN ] test_check_sid_id
> [ OK ] test_check_sid_id
> [ RUN ] test_check_sid_id
> [ OK ] test_check_sid_id
> [ RUN ] test_has_algorithmic
> [ OK ] test_has_algorithmic
> [==========] 6 test(s) run.
> [ PASSED ] 6 test(s).
>
> 0 FAILED TEST(S)
>
> real 0m0.026s
> user 0m0.010s
> sys 0m0.013s
>
>
> $ time ./sss_idmap-tests
> Running suite(s): IDMAP
> 100%: Checks: 20, Failures: 0, Errors: 0
>
> real 0m0.030s
> user 0m0.006s
> sys 0m0.021s
./test_sss_idmap takes very little for me too ,, however I realized the
problem is in how test names are printed during make check, as they are
printed *after* the test is performed
so while I thought test_sss-idmap was taking ages, it was probably the
next one which is strtonum-tests ?
Except that if I run it directly it takes very little too ... I ...
don't know what to say, maybe buffers are not flushed to output so it is
some other test down the road ?
btw I've seen that dyndns-tests are flakey, sometimes they fail ... and
as you rerun all is immediately fine ... any ideas ?
Are just all my machines melting ? (apparently I am having kernel issues
on another machine and gnome-sssion issues on a third today ...)
Simo.
Was the dyndns test failing with master? I saw failures in the past, but
Pavel fixed a use-after-free in the timeout test. If you see failures
with master (and I suspect you do because you're developing against
master) then there is a bug.