[
https://issues.apache.org/jira/browse/HIVE-15571?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15989154#comment-15989154
]
Nishant Bangarwa commented on HIVE-15571:
-----------------------------------------
[~jcamachorodriguez] your understanding is correct, this is not safe to be
execute multiple insert into in parallel.
When we create a druid segment we need to allocate a version and a shardSpec to
it based on existing segments, In case of druid it is easier to manage as the
overlord manages segment allocation and uses interval based locks, which can
handle multiple tasks as the allocation and locking is done at a central place.
In the case of hive insert into, i was planning to lock the complete datasource
for the first version and later on extend that to only lock the intervals for
which data is being ingested in a subsequent PR . For interval based locking we
would need to figure out the interval for the data being ingested in
StorageHandler preInsert, which we don't know currently.
> Support Insert into for druid storage handler
> ---------------------------------------------
>
> Key: HIVE-15571
> URL: https://issues.apache.org/jira/browse/HIVE-15571
> Project: Hive
> Issue Type: New Feature
> Components: Druid integration
> Reporter: slim bouguerra
> Assignee: Nishant Bangarwa
> Attachments: HIVE-15571.01.patch
>
>
> Add support of inset into operator for druid storage handler.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
