[
https://issues.apache.org/jira/browse/IMPALA-15134?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18097011#comment-18097011
]
ASF subversion and git services commented on IMPALA-15134:
----------------------------------------------------------
Commit 3dcd7c4936abc22d10368c6dca6b0bd973010027 in impala's branch
refs/heads/master from Steve Carlin
[ https://gitbox.apache.org/repos/asf?p=impala.git;h=3dcd7c493 ]
IMPALA-15135: Calcite planner: added support for Kudu tables
This commit enables the Calcite planner to support queries on
Kudu tables. Highlights include:
- The Kudu tables can have an auto_incrementing_id column that
does not show up when a "select *" is done on the query. Special
logic has been added to support this. If the user expliclity selects
the auto_incrementing_id column, the query will fall back to the
original planner (IMPALA-15136 has been filed for this).
- A small refactor was done for PrunedPartitionHelper to handle
Kudu tables because it does not support code in the HdfsPartitionPruner.
- A small refactor was done in CalciteDb to handle FeKuduTable
- IMPALA-15134 has been filed on an issue where the cardinality on
the KuduScan table is different from the original planner.
- Added some code in the test framework to handle the section
CALCITE_PLANNER_RUNTIME_PROFILE for when a table format is specified.
Change-Id: Ie22a66f0edad195cade74a8447efc83a8ce526f7
Reviewed-on: http://gerrit.cloudera.org:8080/24528
Tested-by: Impala Public Jenkins <[email protected]>
Reviewed-by: Steve Carlin <[email protected]>
> Calcite planner: Incorrect cardinality in Kudu explain plan
> -----------------------------------------------------------
>
> Key: IMPALA-15134
> URL: https://issues.apache.org/jira/browse/IMPALA-15134
> Project: IMPALA
> Issue Type: Bug
> Reporter: Steve Carlin
> Priority: Major
>
> For the following query in the runtime_filters.test in the functional_kudu db:
> with t1 as (select month x, bigint_col y from alltypes limit 7301),
> t2 as (select distinct int_col x, bigint_col y from alltypestiny limit 3)
> select count(*) from t1, t2 where t1.x = t2.x
> An incorrect cardinality is being calculated for the select node.
> While I haven't researched this fully, it seems that there is a SELECT
> PlanNode added for the Calcite planner that removes some IS NOT NULL data.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
