[jira] [Created] (IMPALA-6334) test_compute_stats_tablesample failing on Isilon builds

["David Knupp (JIRA)" <ji...@apache.org>]

David Knupp created IMPALA-6334:
-----------------------------------

             Summary: test_compute_stats_tablesample failing on Isilon builds
                 Key: IMPALA-6334
                 URL: https://issues.apache.org/jira/browse/IMPALA-6334
             Project: IMPALA
          Issue Type: Bug
    Affects Versions: Impala 2.12.0
            Reporter: David Knupp
            Priority: Critical


MainThread: Comparing QueryTestResults (expected vs actual):
Expected:
{noformat}
3660,3660,12,regex:.*B,'NOT CACHED','NOT CACHED','TEXT','false','hdfs://10.17.95.12:8020/test-warehouse/test_compute_stats_tablesample_16dd5daf.db/alltypesnopart'
{noformat}
Actual:
{noformat}
3661,3661,12,'238.68KB','NOT CACHED','NOT CACHED','TEXT','false','hdfs://10.17.95.12:8020/test-warehouse/test_compute_stats_tablesample_16dd5daf.db/alltypesnopart'
{noformat}

Stacktrace
{noformat}
self = <test_stats_extrapolation.TestStatsExtrapolation object at 0x3aa1890>
vector = <tests.common.test_vector.ImpalaTestVector object at 0x31fd810>
unique_database = 'test_compute_stats_tablesample_16dd5daf'

    @CustomClusterTestSuite.with_args(impalad_args=('--enable_stats_extrapolation=true'))
    def test_compute_stats_tablesample(self, vector, unique_database):
      # Create a partitioned and unpartitioned text table. Use the existing files from
      # functional.alltypes as data because those have a known, stable file size. This
      # test is sensitive to changes in file sizes across test runs because the sampling
      # is file based. Creating test tables with INSERT does not guarantee that the same
      # file sample is selected across test runs, even with REPEATABLE.
    
      # Create partitioned test table. External to avoid dropping files from alltypes.
      part_test_tbl = unique_database + ".alltypes"
      self.client.execute(
        "create external table %s like functional.alltypes" % part_test_tbl)
      alltypes_loc = self._get_table_location("functional.alltypes", vector)
      for m in xrange(1, 13):
        part_loc = path.join(alltypes_loc, "year=2009/month=%s" % m)
        self.client.execute(
          "alter table %s add partition (year=2009,month=%s) location '%s'"
          % (part_test_tbl, m, part_loc))
    
      # Create unpartitioned test table.
      nopart_test_tbl = unique_database + ".alltypesnopart"
      self.client.execute("drop table if exists %s" % nopart_test_tbl)
      self.client.execute(
        "create table %s like functional.alltypesnopart" % nopart_test_tbl)
      nopart_test_tbl_loc = self._get_table_location(nopart_test_tbl, vector)
      # Remove NameNode prefix and first '/' because PyWebHdfs expects that
      if nopart_test_tbl_loc.startswith(NAMENODE):
        nopart_test_tbl_loc = nopart_test_tbl_loc[len(NAMENODE)+1:]
      for m in xrange(1, 13):
        src_part_loc = alltypes_loc + "/year=2009/month=%s" % m
        # Remove NameNode prefix and first '/' because PyWebHdfs expects that
        if src_part_loc.startswith(NAMENODE): src_part_loc = src_part_loc[len(NAMENODE)+1:]
        file_names = self.filesystem_client.ls(src_part_loc)
        for f in file_names:
          self.filesystem_client.copy(path.join(src_part_loc, f),
                                      path.join(nopart_test_tbl_loc, f))
      self.client.execute("refresh %s" % nopart_test_tbl)
    
>     self.run_test_case('QueryTest/compute-stats-tablesample', vector, unique_database)

custom_cluster/test_stats_extrapolation.py:84: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
common/impala_test_suite.py:424: in run_test_case
    self.__verify_results_and_errors(vector, test_section, result, use_db)
common/impala_test_suite.py:297: in __verify_results_and_errors
    replace_filenames_with_placeholder)
common/test_result_verifier.py:404: in verify_raw_results
    VERIFIER_MAP[verifier](expected, actual)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

expected_results = <tests.common.test_result_verifier.QueryTestResult object at 0x3a84850>
actual_results = <tests.common.test_result_verifier.QueryTestResult object at 0x3aa0c10>

    def verify_query_result_is_equal(expected_results, actual_results):
      assert_args_not_none(expected_results, actual_results)
>     assert expected_results == actual_results
E     assert Comparing QueryTestResults (expected vs actual):
E       3660,3660,12,regex:.*B,'NOT CACHED','NOT CACHED','TEXT','false','hdfs://10.17.95.12:8020/test-warehouse/test_compute_stats_tablesample_16dd5daf.db/alltypesnopart' != 3661,3661,12,'238.68KB','NOT CACHED','NOT CACHED','TEXT','false','hdfs://10.17.95.12:8020/test-warehouse/test_compute_stats_tablesample_16dd5daf.db/alltypesnopart'

common/test_result_verifier.py:231: AssertionError
{noformat}




--
This message was sent by Atlassian JIRA
(v6.4.14#64029)