[
https://issues.apache.org/jira/browse/HUDI-1312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17222361#comment-17222361
]
Vinoth Chandar commented on HUDI-1312:
--------------------------------------
Quickest way to test would be make changes in
HoodieParquetInputFormat#listStatus to avoid calling super.listStatus(),
instead supply those from the metadata table.
then if you fire up a spark-shell using
spark.sql.hive.convertMetastoreParquet=false and query a Hive table, then you
should see the difference [~pwason] . To build higher confidence of the
metadata table itself from a query side perspective, you can control this from
a hive config and then run the same queries back to back and diff the results,
for a week or so.
> Query side use of Metadata Table
> --------------------------------
>
> Key: HUDI-1312
> URL: https://issues.apache.org/jira/browse/HUDI-1312
> Project: Apache Hudi
> Issue Type: Sub-task
> Reporter: Prashant Wason
> Assignee: Udit Mehrotra
> Priority: Major
>
> Add support for opening Metadata Table on the query side and using it for
> eliminating file listings.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
