[
https://issues.apache.org/jira/browse/HIVE-29014?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17972756#comment-17972756
]
Denys Kuzmenko edited comment on HIVE-29014 at 6/13/25 10:31 AM:
-----------------------------------------------------------------
The problem is that minor compaction doesn't explicitly remove the deletes,
referencing the rewritten data files; however, iceberg took care of orphan
deletes during commit before
https://github.com/apache/iceberg/pull/11131#discussion_r1815710103
was (Author: dkuzmenko):
The problem is that minor compaction doesn't explicitly remove the deletes,
referencing the rewritten data files; however, iceberg took care of orphan
deletes during commit prior to
https://github.com/apache/iceberg/pull/11131#discussion_r1815710103
> Wrong row count in statistics after minor compacting an Iceberg table
> ---------------------------------------------------------------------
>
> Key: HIVE-29014
> URL: https://issues.apache.org/jira/browse/HIVE-29014
> Project: Hive
> Issue Type: Bug
> Components: Iceberg integration
> Reporter: Krisztian Kasa
> Priority: Major
>
> In {{iceberg_minor_compaction_unpartitioned.q.out}} after the second minor
> compaction of the table {{ice_orc}} the row count should remain 7 but 6 is
> shown in the describe formatted output.
> Run {{iceberg_minor_compaction_unpartitioned.q}} to reproduce the issue.
> The issue is caused by an optimization in Iceberg:
> https://github.com/apache/iceberg/pull/11131#issuecomment-2913183379
> https://github.com/apache/iceberg/pull/11131#discussion_r1815710103
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
