BIND 10 #2468: b10-stats-httpd_test.py not stopping

[BIND 10 Development]

#2468: b10-stats-httpd_test.py not stopping
-------------------------------------+-------------------------------------
                   Reporter:  jreed  |                 Owner:
                       Type:         |                Status:  new
  defect                             |             Milestone:  New Tasks
                   Priority:         |            Resolution:
  medium                             |             Sensitive:  0
                  Component:         |           Sub-Project:  Core
  statistics                         |  Estimated Difficulty:  0
                   Keywords:         |           Total Hours:  0
            Defect Severity:  N/A    |
Feature Depending on Ticket:         |
        Add Hours to Ticket:  0      |
                  Internal?:  0      |
-------------------------------------+-------------------------------------

Comment (by naokikambe):

 Thank you for the comment.

 Replying to [comment:6 jinmei]:
 > If the server instance is only needed for that specific test, the
 > change itself may make sense, but I don't understand why the original
 > code requires too many open files, and I'm a bit reluctant to give it
 > a go without knowing why it solves the problem.
 >
 > I'd be also concerned about the use of a separate thread for unit
 > tests in the first place.  Sometimes such a tricky setup might be the
 > real last resort, but we should at least fully consider other options
 > with some proper abstraction of the service.  It's often possible for
 > such a dynamic language as Python.

 I know many problems in ``b10-stats-httpd_test.py``. We should consider
 other options like #1668. We can also work for revising ``b10-stats-
 httpd_test.py`` on another ticket.

 But now ``b10-stats-httpd_test.py`` hangs at the buildbot. I think we
 should fix it instantly. If we don't need to do so, we might need to
 revert the change splitting the unittest(revert
 3a32db0518e8ea18bad9f04df67ad04dcc7afe45). At least before the change,
 such a failure(too many open files) didn't happen at the buildbot, I
 think. That change was just for observation of hanging but it caused
 another problem. So can I revert it?

 > Regarding the code itself, the use of hasattr seems to be a bit
 > cryptic.  While we need more lines of code, it seems to me more
 > natural if we set it to None in setUp() and do shutdown if it is not
 > None in tearDown.

 If ``test_mccs()`` is unexpectedly stopped, the test needs to shutdown the
 ``stats_server`` object after that. It just does in ``tearDown()``. That
 doesn't make sense much.

 Regards,

-- 
Ticket URL: <http://bind10.isc.org/ticket/2468#comment:7>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development