[
https://issues.apache.org/jira/browse/HIVE-26337?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17564178#comment-17564178
]
Ayush Saxena commented on HIVE-26337:
-------------------------------------
hmm, that isn't something good, if it is true, this would bother
hive-replication performance as well.
Do you plan to raise a PR with the fix, let me know if you face any issues
there, I can try help...
> Duplicate and single event log under NOTIFICATION_LOG for drop-partition
> unlike add-partition event which overloads the metadata table
> --------------------------------------------------------------------------------------------------------------------------------------
>
> Key: HIVE-26337
> URL: https://issues.apache.org/jira/browse/HIVE-26337
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 3.1.0
> Environment: HDInsight 4.1.8
> Hive 3.1.2
> Reporter: Sindhu Subhas
> Priority: Major
>
> Multiple events are generated for each drop partition under NOTIFICATION_LOG
> whereas a single event is generated for add_partition during msck sync
> partitions as part of partition management discovery.
> Say, Hive is to add 5 new partitions and remove 100 partitions into SQL
> server, one event is generated for 1 add_partition event entry whereas 100
> drop_partitions events are generated under NOTIFICATION_LOG. This results in
> overloading of the table and also indexes associated with the table.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
