[
https://issues.apache.org/jira/browse/HIVE-22298?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16952802#comment-16952802
]
Peter Vary commented on HIVE-22298:
-----------------------------------
There are 4 group of changes here:
# OrcSplit.java - We start using
{{VectorizedOrcAcidRowBatchReader.canUseLlapIoForAcid}} for deciding when we
are able to use LlapIO
# LlapRecordReader.java - When reading original files we should not expect ACID
file format
# VectorizedOrcAcidRowBatchReader.java - Again, we have to be prepared to
handle original files too
# VectorizedOrcAcidRowBatchReader#DeleteRecordKey - This was a race condition
where we used a static variable in {{compareTo}}
Added a small change into {{acid_vectorization_original.q}} with which I was
able to repro a test case very similar to HIVE-17921.
[~gopalv]: Could you please review?
Thanks,
Peter
> Allow Llap IO cache for reading tables without delete delta
> -----------------------------------------------------------
>
> Key: HIVE-22298
> URL: https://issues.apache.org/jira/browse/HIVE-22298
> Project: Hive
> Issue Type: Improvement
> Reporter: Peter Vary
> Assignee: Peter Vary
> Priority: Major
> Attachments: HIVE-22298.2.patch, HIVE-22298.3.patch,
> HIVE-22298.4.patch, HIVE-22298.5.patch, HIVE-22298.patch
>
>
> Since HIVE-19985 we can use Llap IO cache for queries which are not
> explicitly reading ROW_ID even for original files.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
