You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hive.apache.org by "Abhay (Jira)" <ji...@apache.org> on 2022/12/07 06:29:00 UTC

[jira] [Assigned] (HIVE-26814) Atomic Schema Upgrade for HMS

     [ https://issues.apache.org/jira/browse/HIVE-26814?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Abhay reassigned HIVE-26814:
----------------------------


> Atomic Schema Upgrade for HMS
> -----------------------------
>
>                 Key: HIVE-26814
>                 URL: https://issues.apache.org/jira/browse/HIVE-26814
>             Project: Hive
>          Issue Type: Improvement
>          Components: Metastore
>            Reporter: Abhay
>            Assignee: Abhay
>            Priority: Major
>
> Currently, as schematool applies schema changes during an upgrade, each change is autocommitted. So if it fails after applying a few changes from an upgrade file, those changes remain in the schema. So re-running the schematool fails at the first line because that change already exists.
> Instead, we should create a transactional boundary between each upgrade file, aka commit only after all the changes in an upgrade file are successful.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)