[
https://issues.apache.org/jira/browse/IMPALA-14142?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17980457#comment-17980457
]
Zoltán Borók-Nagy commented on IMPALA-14142:
--------------------------------------------
Yes, and only in release builds. In ARM debug builds the test passes.
That said, the test is flawed as we cannot expect COMPUTE STATS with
TABLESAMPLE to be completely deterministic (not even with REPEATABLE clause),
see:
[https://github.com/apache/impala/blob/276f48ee85944f87ebb43a5b404020dcb38a5e66/tests/metadata/test_stats_extrapolation.py#L51-L54]
It's interesting that the test is quite deterministic on these small Iceberg
tables, and also in ARM release builds we get the same results on repeated
runs, they just differ from the results we get elsewhere.
Anyway, I'm planning to rewrite the test completely to work similarly to
TestStatsExtrapolation.test_compute_stats_tablesample.
> TestIcebergV2Table.test_compute_stats_table_sampling fails in exhaustive
> builds
> -------------------------------------------------------------------------------
>
> Key: IMPALA-14142
> URL: https://issues.apache.org/jira/browse/IMPALA-14142
> Project: IMPALA
> Issue Type: Bug
> Reporter: Zoltán Borók-Nagy
> Assignee: Zoltán Borók-Nagy
> Priority: Major
> Labels: broken-build, impala-iceberg
>
> h3. Error Message
> {noformat}
> assert Comparing QueryTestResults (expected vs actual):
> 'action','STRING',2,0,5,4.333333492279053,-1,-1 !=
> 'action','STRING',12,0,5,4.333333492279053,-1,-1
> 'event_time','TIMESTAMP',2,0,16,16,-1,-1 !=
> 'event_time','TIMESTAMP',12,0,16,16,-1,-1 'id','INT',3,0,4,4,-1,-1 !=
> 'id','INT',18,0,4,4,-1,-1 'user','STRING',2,0,4,4,-1,-1 !=
> 'user','STRING',12,0,4,4,-1,-1{noformat}
> h3. Stacktrace
> {noformat}
> query_test/test_iceberg.py:1554: in test_compute_stats_table_sampling
> self.run_test_case('QueryTest/iceberg-v2-compute-stats-table-sampling',
> vector)
> /data/jenkins/workspace/impala-asf-master-exhaustive-release-arm/repos/Impala/tests/common/impala_test_suite.py:915:
> in run_test_case
> self.__verify_results_and_errors(vector, test_section, result, use_db)
> /data/jenkins/workspace/impala-asf-master-exhaustive-release-arm/repos/Impala/tests/common/impala_test_suite.py:747:
> in __verify_results_and_errors
> replace_filenames_with_placeholder)
> /data/jenkins/workspace/impala-asf-master-exhaustive-release-arm/repos/Impala/tests/common/test_result_verifier.py:523:
> in verify_raw_results
> VERIFIER_MAP[verifier](expected, actual)
> /data/jenkins/workspace/impala-asf-master-exhaustive-release-arm/repos/Impala/tests/common/test_result_verifier.py:305:
> in verify_query_result_is_equal
> assert expected_results == actual_results
> E assert Comparing QueryTestResults (expected vs actual):
> E 'action','STRING',2,0,5,4.333333492279053,-1,-1 !=
> 'action','STRING',12,0,5,4.333333492279053,-1,-1
> E 'event_time','TIMESTAMP',2,0,16,16,-1,-1 !=
> 'event_time','TIMESTAMP',12,0,16,16,-1,-1
> E 'id','INT',3,0,4,4,-1,-1 != 'id','INT',18,0,4,4,-1,-1
> E 'user','STRING',2,0,4,4,-1,-1 !=
> 'user','STRING',12,0,4,4,-1,-1{noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
