[
https://issues.apache.org/jira/browse/HIVE-3938?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13634523#comment-13634523
]
Mithun Radhakrishnan commented on HIVE-3938:
--------------------------------------------
Ok, Dilip's comment about support for multi-table inserts poses a dilemma.
Thus far, HiveMetaStore::add_partitions_core() seems to have assumed (at least
in part) that all the parts belong to the same table (evidenced by the
logInfo() line). If one might call add_partitions() with parts in different
tables, should they all be considered to be part of a single atomic set? Or
ought I to categorize them (per table) and send separate Events out?
I'm inclined to do the former (because you retain the ability to break the
partition-sets up downstream), but I'm open to suggestions.
(+ [~ashutoshc], who wrote the initial version of this.)
> Hive MetaStore should send a single AddPartitionEvent for atomically added
> partition-set.
> -----------------------------------------------------------------------------------------
>
> Key: HIVE-3938
> URL: https://issues.apache.org/jira/browse/HIVE-3938
> Project: Hive
> Issue Type: Bug
> Components: Metastore
> Affects Versions: 0.10.0
> Reporter: Mithun Radhakrishnan
> Assignee: Mithun Radhakrishnan
> Attachments: HIVE-3938.trunk.patch
>
>
> HiveMetaStore::add_partitions() currently adds all partitions specified in
> one call using a single meta-store transaction. This acts correctly. However,
> there's one AddPartitionEvent created per partition specified.
> Ideally, the set of partitions added atomically can be communicated using a
> single AddPartitionEvent, such that they are consumed together.
> I'll post a patch that does this.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
