[
https://issues.apache.org/jira/browse/HIVE-17585?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16180289#comment-16180289
]
Tao Li commented on HIVE-17585:
-------------------------------
[~sershe] Thanks for the comments. I thought about that and was a little
concerned with the memory overhead to have a new Hive instance per thread
(especially the embedded metastore scenario), so I chose to go with the
singleton approach. But I agree that having the thread local will give us the
best safety.
Regarding the latency, the major advantage of the loading partitions in
parallel is making the HDFS calls in parallel, so synchronizing on the
metastore client should not be a big concern.
> Improve thread safety when loading dynamic partitions in parallel
> -----------------------------------------------------------------
>
> Key: HIVE-17585
> URL: https://issues.apache.org/jira/browse/HIVE-17585
> Project: Hive
> Issue Type: Bug
> Reporter: Tao Li
> Assignee: Tao Li
> Fix For: 3.0.0
>
> Attachments: HIVE-17585.1.patch, HIVE-17585.2.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
