Github user ravipesala commented on a diff in the pull request:
https://github.com/apache/carbondata/pull/1681#discussion_r158197465
--- Diff:
integration/spark2/src/main/scala/org/apache/spark/sql/execution/command/management/CarbonLoadDataCommand.scala
---
@@ -488,7 +490,24 @@ case class CarbonLoadDataCommand(
}
InternalRow.fromSeq(data)
}
- LogicalRDD(attributes, rdd)(sparkSession)
+ if (updateModel.isDefined) {
+ sparkSession.sparkContext.setLocalProperty(EXECUTION_ID_KEY, null)
+ // In case of update, we don't need the segmrntid column in case
of partitioning
+ val dropAttributes = attributes.dropRight(1)
+ val finalOutput = relation.output.map { attr =>
+ dropAttributes.find { d =>
+ val index = d.name.lastIndexOf("-updatedColumn")
--- End diff --
It requires change the order in UpdateCommand, so it impacts the actual
flow of IUD. So I guess better handle here
---
