[
https://issues.apache.org/jira/browse/HIVE-23550?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Adesh Kumar Rao resolved HIVE-23550.
------------------------------------
Fix Version/s: (was: 4.0.0)
Resolution: Invalid
HIVE-21641 changed the usage of CalciteAnalyzer from just
CalciteAnalyzer.genLogicalPlan to fully analyzing (CalciteAnalyzer.analyze) the
query. That includes the retry logic for calcite semantic exceptions too.
Closing this.
> GetSplits does not retries queries for CacliteSemanticException
> ---------------------------------------------------------------
>
> Key: HIVE-23550
> URL: https://issues.apache.org/jira/browse/HIVE-23550
> Project: Hive
> Issue Type: Bug
> Components: UDF
> Affects Versions: 3.1.0
> Reporter: Adesh Kumar Rao
> Assignee: Adesh Kumar Rao
> Priority: Major
>
> Reproducible case:
> {noformat}
> create table t1 (c1 int, c2 int, c3 int);
> select get_splits("select c2, count(distinct c3) from t1 group by c2 having
> count(distinct c3) > 1",0);{noformat}
>
> Error:
> {noformat}
> Error: java.io.IOException:
> org.apache.hadoop.hive.ql.optimizer.calcite.CalciteSemanticException:
> Distinct without an aggregation. (state=,code=0)
> {noformat}
> This happens because calcite does not understand the query "select c2,
> count(distinct c3) from t1 group by c2 having count(distinct c3) > 1" and
> throws calciteSemanticException
>
> If this query is run directly via beeline, hiveserver2 catches this exception
> and re-analyzes the query by turning off the cbo.
>
> This retrying mechanism is missing in GetSplits UDF.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
