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):

 On my FreeBSD9 environment, I've never seen it. As long as I grepped logs
 at git.bind10.isc.org, it likely never happened on Linux systems.
 {{{
 > find . -name unittests.out | xargs grep -l 'A deadlock might be detec'
 ./BIND10-cppcheck/20121029173012-FreeBSD8-amd64-GCC/logs/unittests.out
 ./BIND10-cppcheck/20121029180746-FreeBSD8-amd64-GCC/logs/unittests.out
 ./BIND10-cppcheck/20121029223001-FreeBSD8-amd64-GCC/logs/unittests.out
 ./BIND10/20120508224610-NetBSD5-amd64/logs/unittests.out
 ./BIND10/20120509055016-NetBSD5-amd64/logs/unittests.out
 ./BIND10/20120509105016-NetBSD5-amd64/logs/unittests.out
 ./BIND10/20120511195000-NetBSD5-amd64/logs/unittests.out
 ./BIND10/20120722185001-Solaris10-sparc-Sunstudio/logs/unittests.out
 ./BIND10/20120808174510-MacOSX10.6-x86_64-Clang-Static/logs/unittests.out
 ./BIND10/20120913153047-Solaris10-sparc-GCC/logs/unittests.out
 ./BIND10/20121030203500-FreeBSD8-i386/logs/unittests.out
 >
 }}}

 BTW according to the failure logs, it seems to happen at line 766 or 775
 in b10-stats-httpd_test.py. They are inside of test_httpd(). test_httpd()
 is a little big. So how about splitting test_httpd() as following? If we
 specify a failed test, we might be able to fix or skip it.

 {{{#!patch
 diff --git a/src/bin/stats/tests/b10-stats-httpd_test.py
 b/src/bin/stats/tests/b10-stats-httpd_test.py
 index 73ed35b..627d50e 100644
 --- a/src/bin/stats/tests/b10-stats-httpd_test.py
 +++ b/src/bin/stats/tests/b10-stats-httpd_test.py
 @@ -761,6 +761,7 @@ class TestStatsHttpd(unittest.TestCase):
              self.assertEqual(ht.address_family, socket.AF_INET)
              self.assertTrue(isinstance(ht.socket, socket.socket))

 +    def test_httpd_anyIPv4(self):
          # any address (IPv4)
          server_addresses = get_availaddr(address='0.0.0.0')
          self.stats_httpd = MyStatsHttpd(server_addresses)
 @@ -769,6 +770,7 @@ class TestStatsHttpd(unittest.TestCase):
              self.assertEqual(ht.address_family,socket.AF_INET)
              self.assertTrue(isinstance(ht.socket, socket.socket))

 +    def test_httpd_anyIPv6(self):
          # any address (IPv6)
          if self.ipv6_enabled:
              server_addresses = get_availaddr(address='::')
 @@ -778,6 +780,7 @@ class TestStatsHttpd(unittest.TestCase):
                  self.assertEqual(ht.address_family,socket.AF_INET6)
                  self.assertTrue(isinstance(ht.socket, socket.socket))

 +    def test_httpd_failed(self):
          # existent hostname
          self.assertRaises(stats_httpd.HttpServerError, MyStatsHttpd,
                            get_availaddr(address='localhost'))}}}

 Regards,

 Replying to [comment:1 jreed]:
 > Here is output from when I killed it last time (I think):
 >
 >
 http://git.bind10.isc.org/~tester/builder/BIND10-cppcheck/20121029223001-FreeBSD8-amd64-GCC/logs/unittests.out
 >
 > Maybe hanging in gethostbyaddr?
 >
 > Also another example, different system:
 >
 >
 http://git.bind10.isc.org/~tester/builder/BIND10/20121030203500-FreeBSD8-i386/logs/unittests.out
 >
 > About 6 failures in the stats/tests in past month and a half.

-- 
Ticket URL: <http://bind10.isc.org/ticket/2468#comment:2>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development