jackylk commented on a change in pull request #3452: [CARBONDATA-3579] Support merging index files when adding new partition
URL:
https://github.com/apache/carbondata/pull/3452#discussion_r346649371
##########
File path: integration/spark2/src/main/scala/org/apache/spark/sql/execution/command/partition/CarbonAlterTableAddHivePartitionCommand.scala
##########
@@ -153,6 +153,23 @@ case class CarbonAlterTableAddHivePartitionCommand(
CarbonLoaderUtil.addDataIndexSizeIntoMetaEntry(newMetaEntry, loadModel.getSegmentId, table)
// Make the load as success in table status
CarbonLoaderUtil.recordNewLoadMetadata(newMetaEntry, loadModel, false, false)
+
+ // Normally, application will use Carbon SDK to write files into a partition folder, then
+ // add the folder to partitioned carbon table.
+ // If there are many threads writes to the same partition folder, there will be many
+ // carbon index files, and it is not good for query performance since all index files
+ // need to be read to spark driver.
+ // So, here trigger to merge the index files by sending an event
+ val alterTableModel = AlterTableModel(
Review comment:
For normal minor compaction, is it triggering index merge for all segments?
One solution is to add one parameter in AlterTableMergeIndexEvent to specify the segmentId to do merge index. What do you think?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[hidden email]
With regards,
Apache Git Services
Locked
