[BEAM-9119] apache_beam.runners.portability.fn_api_runner_test.FnApiRunnerTest[...].test_large_elements is flaky (sickbayed)

• Type: Improvement
• Status: Resolved
• Priority: P3
• Resolution: Fixed
• Fix Version/s: Missing
• Component/s: sdk-py-core
• Labels: beam-fixit sickbay
• Assignee: Unassigned
• Reporter: Valentyn Tymofieiev
• Created: 14/Jan/20 23:40
• Updated: 13/Jan/22 21:52
• Resolved: 29/Sep/21 00:47

Description

Saw 3 errors today, all manifest with:

IndexError: index out of range in apache_beam/coders/slow_stream.py", line 169, in read_byte_py3.

https://builds.apache.org/job/beam_PreCommit_Python_Phrase/1369
https://builds.apache.org/job/beam_PreCommit_Python_Phrase/1365
https://builds.apache.org/job/beam_PreCommit_Python_Phrase/1370

Sample logs:

12:10:27  =================================== FAILURES ===================================
12:10:27  ____________ FnApiRunnerTestWithDisabledCaching.test_large_elements ____________
12:10:27  [gw0] linux -- Python 3.6.8 /home/jenkins/jenkins-slave/workspace/beam_PreCommit_Python_Phrase/src/sdks/python/test-suites/tox/py36/build/srcs/sdks/python/target/.tox-py36-gcp-pytest/py36-gcp-pytest/bin/python
12:10:27  
12:10:27  self = <apache_beam.runners.portability.fn_api_runner_test.FnApiRunnerTestWithDisabledCaching testMethod=test_large_elements>
12:10:27  
12:10:27      def test_large_elements(self):
12:10:27        with self.create_pipeline() as p:
12:10:27          big = (p
12:10:27                 | beam.Create(['a', 'a', 'b'])
12:10:27                 | beam.Map(lambda x: (
12:10:27                     x, x * data_plane._DEFAULT_SIZE_FLUSH_THRESHOLD)))
12:10:27      
12:10:27          side_input_res = (
12:10:27              big
12:10:27              | beam.Map(lambda x, side: (x[0], side.count(x[0])),
12:10:27                         beam.pvalue.AsList(big | beam.Map(lambda x: x[0]))))
12:10:27          assert_that(side_input_res,
12:10:27                      equal_to([('a', 2), ('a', 2), ('b', 1)]), label='side')
12:10:27      
12:10:27          gbk_res = (
12:10:27              big
12:10:27              | beam.GroupByKey()
12:10:27              | beam.Map(lambda x: x[0]))
12:10:27  >       assert_that(gbk_res, equal_to(['a', 'b']), label='gbk')
12:10:27  
12:10:27  apache_beam/runners/portability/fn_api_runner_test.py:617: 
12:10:27  _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
12:10:27  apache_beam/pipeline.py:479: in __exit__
12:10:27      self.run().wait_until_finish()
12:10:27  apache_beam/pipeline.py:459: in run
12:10:27      self._options).run(False)
12:10:27  apache_beam/pipeline.py:472: in run
12:10:27      return self.runner.run_pipeline(self, self._options)
12:10:27  apache_beam/runners/portability/fn_api_runner.py:472: in run_pipeline
12:10:27      default_environment=self._default_environment))
12:10:27  apache_beam/runners/portability/fn_api_runner.py:480: in run_via_runner_api
12:10:27      return self.run_stages(stage_context, stages)
12:10:27  apache_beam/runners/portability/fn_api_runner.py:569: in run_stages
12:10:27      stage_context.safe_coders)
12:10:27  apache_beam/runners/portability/fn_api_runner.py:889: in _run_stage
12:10:27      result, splits = bundle_manager.process_bundle(data_input, data_output)
12:10:27  apache_beam/runners/portability/fn_api_runner.py:2076: in process_bundle
12:10:27      part, expected_outputs), part_inputs):
12:10:27  /usr/lib/python3.6/concurrent/futures/_base.py:586: in result_iterator
12:10:27      yield fs.pop().result()
12:10:27  /usr/lib/python3.6/concurrent/futures/_base.py:432: in result
12:10:27      return self.__get_result()
12:10:27  /usr/lib/python3.6/concurrent/futures/_base.py:384: in __get_result
12:10:27      raise self._exception
12:10:27  apache_beam/utils/thread_pool_executor.py:44: in run
12:10:27      self._future.set_result(self._fn(*self._fn_args, **self._fn_kwargs))
12:10:27  apache_beam/runners/portability/fn_api_runner.py:2076: in <lambda>
12:10:27      part, expected_outputs), part_inputs):
12:10:27  apache_beam/runners/portability/fn_api_runner.py:2020: in process_bundle
12:10:27      expected_outputs[output.transform_id]).append(output.data)
12:10:27  apache_beam/runners/portability/fn_api_runner.py:285: in append
12:10:27      windowed_key_value = coder_impl.decode_from_stream(input_stream, True)
12:10:27  apache_beam/coders/coder_impl.py:1153: in decode_from_stream
12:10:27      value = self._value_coder.decode_from_stream(in_stream, nested)
12:10:27  apache_beam/coders/coder_impl.py:749: in decode_from_stream
12:10:27      for i, c in enumerate(self._coder_impls)])
12:10:27  apache_beam/coders/coder_impl.py:749: in <listcomp>
12:10:27      for i, c in enumerate(self._coder_impls)])
12:10:27  apache_beam/coders/coder_impl.py:493: in decode_from_stream
12:10:27      return in_stream.read_all(nested)
12:10:27  apache_beam/coders/slow_stream.py:158: in read_all
12:10:27      return self.read(self.read_var_int64() if nested else self.size())
12:10:27  apache_beam/coders/slow_stream.py:175: in read_var_int64
12:10:27      byte = self.read_byte()
12:10:27  _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
12:10:27  
12:10:27  self = <apache_beam.coders.slow_stream.InputStream object at 0x7f2390d80c88>
12:10:27  
12:10:27      def read_byte_py3(self):
12:10:27        # type: () -> int
12:10:27        self.pos += 1
12:10:27  >     return self.data[self.pos - 1]
12:10:27  E     IndexError: index out of range
12:10:27  
12:10:27  apache_beam/coders/slow_stream.py:169: IndexError

Issue Links

• links to: GitHub Pull Request #11425