You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hive.apache.org by "Marton Bod (Jira)" <ji...@apache.org> on 2022/01/06 15:40:00 UTC

[jira] [Updated] (HIVE-25849) Disable insert overwrite for bucket partitioned Iceberg tables

     [ https://issues.apache.org/jira/browse/HIVE-25849?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Marton Bod updated HIVE-25849:
------------------------------
    Description: 
Insert overwrite should be disabled where the target Iceberg table is a bucket partitioned table, since which existing partitions will be overwritten is very hard to predict from a user's POV, as it depends on the bucket hash values calculated for the new dataset's rows. It's better to be on the safe side and disable this operation to avoid unwanted data loss.

Note: this the same approach followed by Impala too.

  was:Insert overwrite should be disabled where the target Iceberg table is a bucket partitioned table, since which existing partitions will be overwritten is very hard to predict from a user's POV, as it depends on the bucket hash values calculated for the new dataset's rows. It's better to be on the safe side and disable this operation to avoid unwanted data loss.


> Disable insert overwrite for bucket partitioned Iceberg tables
> --------------------------------------------------------------
>
>                 Key: HIVE-25849
>                 URL: https://issues.apache.org/jira/browse/HIVE-25849
>             Project: Hive
>          Issue Type: Improvement
>            Reporter: Marton Bod
>            Assignee: Marton Bod
>            Priority: Major
>
> Insert overwrite should be disabled where the target Iceberg table is a bucket partitioned table, since which existing partitions will be overwritten is very hard to predict from a user's POV, as it depends on the bucket hash values calculated for the new dataset's rows. It's better to be on the safe side and disable this operation to avoid unwanted data loss.
> Note: this the same approach followed by Impala too.



--
This message was sent by Atlassian Jira
(v8.20.1#820001)