You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Asif (Jira)" <ji...@apache.org> on 2023/04/12 22:13:00 UTC
[jira] [Updated] (SPARK-43112) Spark may use a column other than the actual specified partitioning column for partitioning, for Hive format tables
[ https://issues.apache.org/jira/browse/SPARK-43112?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Asif updated SPARK-43112:
-------------------------
Description:
The class org.apache.spark.sql.catalyst.catalog.HiveTableRelation has its output method implemented as
// The partition column should always appear after data columns.
override def output: Seq[AttributeReference] = dataCols ++ partitionCols
But the DataWriting commands of spark like InsertIntoHiveDirCommand, expect that the output from HiveTableRelation is in the order in which the columns are actually defined in the DDL.
As a result, multiple mismatch scenarios can happen like:
1) data type casting exception being thrown , even though the data frame being inserted has schema which is identical to what is used for creating ddl.
OR
2) Wrong column being used for partitioning , if the datatypes are same or cast-able, like date type and long
will be creating a PR with the bug test
was:
The class org.apache.spark.sql.catalyst.catalog.HiveTableRelation has its output method implemented as
// The partition column should always appear after data columns.
override def output: Seq[AttributeReference] = dataCols ++ partitionCols
But the DataWriting commands of spark like InsertIntoHiveDirCommand, expect that the out from HiveTableRelation is in the order in which the columns are actually defined in the DDL.
As a result, multiple mistmatch scenarios can happen like:
1) data type casting exception being thrown , even though the data frame being inserted has schema which is identical to what is used for creating ddl.
OR
2) Wrong column being used for partitioning , if the datatypes are same or castable, like datetype and long
will be creating a PR with the bug test
> Spark may use a column other than the actual specified partitioning column for partitioning, for Hive format tables
> --------------------------------------------------------------------------------------------------------------------
>
> Key: SPARK-43112
> URL: https://issues.apache.org/jira/browse/SPARK-43112
> Project: Spark
> Issue Type: Bug
> Components: SQL
> Affects Versions: 3.3.1
> Reporter: Asif
> Priority: Critical
>
> The class org.apache.spark.sql.catalyst.catalog.HiveTableRelation has its output method implemented as
> // The partition column should always appear after data columns.
> override def output: Seq[AttributeReference] = dataCols ++ partitionCols
> But the DataWriting commands of spark like InsertIntoHiveDirCommand, expect that the output from HiveTableRelation is in the order in which the columns are actually defined in the DDL.
> As a result, multiple mismatch scenarios can happen like:
> 1) data type casting exception being thrown , even though the data frame being inserted has schema which is identical to what is used for creating ddl.
> OR
> 2) Wrong column being used for partitioning , if the datatypes are same or cast-able, like date type and long
> will be creating a PR with the bug test
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org