[
https://issues.apache.org/jira/browse/HIVE-19102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Matt McCline updated HIVE-19102:
--------------------------------
Description:
There are known bugs recently found and reported that occur when vectorization
is turn on in Q files. Until those bugs are fixed, add SET statements to the
top of the Q files that suppress vectorization.
Change a few EXPLAIN VECTORIZATION to EXPLAIN where it isn't needed and creates
an unnecessary Q file difference during enable by default experiments.
* +TestCliDriver+
** +Execution Failures+
*** *input_lazyserde.q*
**** HIVE-19088
*** *input_lazyserde2.q*
**** HIVE-19088, too.
*** *nested_column_pruning.q*
**** HIVE-19016
*** *parquet_map_of_arrays_of_ints.q*
**** HIVE-19015
*** *parquet_map_of_maps.q*
**** HIVE-19015, too.
*** *parquet_nested_complex.q*
**** HIVE-19016
** +Wrong Results+
*** *delete_orig_table.q*
**** HIVE-19109
*** *offset_limit_global_optimizer.q*
**** Added ORDER BY clauses to fix different vectorization intermediate
results due to LIMIT clause.
*** *parquet_ppd_decimal.q*
**** HIVE-19108
*** *udf_context_aware.q*
**** Test detects vectorization and output changes.
Added "set hive.test.vectorized.execution.enabled.override=none;"
*** *vector_udf3.q*
**** Test detects vectorization and output changes.
Added "set hive.test.vectorized.execution.enabled.override=none;"
* +TestContribCliDriver+
** +Wrong Results+
*** *udf_example_arraymapstruct.q*
**** HIVE-19110
was:
There are known bugs recently found and reported that occur when vectorization
is turn on in Q files. Until those bugs are fixed, add SET statements to the
top of the Q files that suppress vectorization.
Change a few EXPLAIN VECTORIZATION to EXPLAIN where it isn't needed and creates
an unnecessary Q file difference during enable by default experiments.
* +Execution Failures+
** *input_lazyserde*
*** HIVE-19088
** *input_lazyserde2*
*** HIVE-19088, too.
** *nested_column_pruning*
*** HIVE-19016
** *parquet_map_of_arrays_of_ints*
*** HIVE-19015
** *parquet_map_of_maps*
*** HIVE-19015, too.
** *parquet_nested_complex*
*** HIVE-19016
* +Wrong Results+
** *udf_context_aware*
*** Test detects vectorization and output changes.
Added "set hive.test.vectorized.execution.enabled.override=none;"
> Vectorization: Suppress known Q file bugs
> -----------------------------------------
>
> Key: HIVE-19102
> URL: https://issues.apache.org/jira/browse/HIVE-19102
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 3.0.0
> Reporter: Matt McCline
> Assignee: Matt McCline
> Priority: Critical
> Attachments: HIVE-19102.01.patch
>
>
> There are known bugs recently found and reported that occur when
> vectorization is turn on in Q files. Until those bugs are fixed, add SET
> statements to the top of the Q files that suppress vectorization.
> Change a few EXPLAIN VECTORIZATION to EXPLAIN where it isn't needed and
> creates an unnecessary Q file difference during enable by default experiments.
> * +TestCliDriver+
> ** +Execution Failures+
> *** *input_lazyserde.q*
> **** HIVE-19088
> *** *input_lazyserde2.q*
> **** HIVE-19088, too.
> *** *nested_column_pruning.q*
> **** HIVE-19016
> *** *parquet_map_of_arrays_of_ints.q*
> **** HIVE-19015
> *** *parquet_map_of_maps.q*
> **** HIVE-19015, too.
> *** *parquet_nested_complex.q*
> **** HIVE-19016
> ** +Wrong Results+
> *** *delete_orig_table.q*
> **** HIVE-19109
> *** *offset_limit_global_optimizer.q*
> **** Added ORDER BY clauses to fix different vectorization intermediate
> results due to LIMIT clause.
> *** *parquet_ppd_decimal.q*
> **** HIVE-19108
> *** *udf_context_aware.q*
> **** Test detects vectorization and output changes.
> Added "set hive.test.vectorized.execution.enabled.override=none;"
> *** *vector_udf3.q*
> **** Test detects vectorization and output changes.
> Added "set hive.test.vectorized.execution.enabled.override=none;"
> * +TestContribCliDriver+
> ** +Wrong Results+
> *** *udf_example_arraymapstruct.q*
> **** HIVE-19110
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)