Github user kunal642 commented on a diff in the pull request: https://github.com/apache/carbondata/pull/2109#discussion_r178237981 --- Diff: integration/spark2/src/main/scala/org/apache/spark/sql/execution/command/management/CarbonLoadDataCommand.scala --- @@ -149,7 +143,16 @@ case class CarbonLoadDataCommand( val LOGGER: LogService = LogServiceFactory.getLogService(this.getClass.getCanonicalName) val carbonProperty: CarbonProperties = CarbonProperties.getInstance() carbonProperty.addProperty("zookeeper.enable.lock", "false") - + currPartitions = if (table.isHivePartitionTable) { --- End diff -- Actually i had to move getPartitions to processData because when this method is called for child tables then we are expecting validSegments to be set. In processMeta they are not set. I have passed carbonTable to avoid metastore call. Should not cause problems.c
---