[
https://issues.apache.org/jira/browse/HIVE-16669?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16174083#comment-16174083
]
Eugene Koifman edited comment on HIVE-16669 at 9/21/17 1:09 AM:
----------------------------------------------------------------
OK, this is more than just fine tuning. Suppose we have
base_8
delta_9
delete_delta_10 - this affects rows in base_8
Minor compaction (as currently implemented (inherited from Acid 1)) will
produce delta_9_10 which means all deletes by txn 10 affecting rows in base_8
are lost.
so HIVE-17089 is effectively incomplete w/o this
was (Author: ekoifman):
OK, this is more than just fine tuning. Suppose we have
base_8
delta_9
delete_delta_10 - this affects rows in base_8
Minor compaction (as currently implemented (inherited from Acid 1)) will
produce delta_9_10 which means all deletes by txn 10 affecting rows in base_8
are lost.
> Fine tune Compaction to take advantage of Acid 2.0
> --------------------------------------------------
>
> Key: HIVE-16669
> URL: https://issues.apache.org/jira/browse/HIVE-16669
> Project: Hive
> Issue Type: Bug
> Components: Transactions
> Reporter: Eugene Koifman
> Assignee: Eugene Koifman
> Priority: Blocker
>
> * There is little point using 2.0 vectorized reader since there is no
> operator pipeline in compaction
> * If minor compaction just concats delete_delta files together, then the 2
> stage compaction should always ensure that we have a limited number of Orc
> readers to do the merging and current OrcRawRecordMerger should be fine
> * ...
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
