You are viewing a plain text version of this content. The canonical link for it is here.

Posted to issues@hive.apache.org by "Sergey Shelukhin (JIRA)" <ji...@apache.org> on 2017/10/09 23:45:00 UTC

[jira] [Commented] (HIVE-17695) HiveInputFormat.processForWriteIds(Path dir, JobConf conf, ValidTxnList validTxnList, List finalPaths)

    [ https://issues.apache.org/jira/browse/HIVE-17695?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16197887#comment-16197887 ] 

Sergey Shelukhin commented on HIVE-17695:
-----------------------------------------

Cannot be done with list bucketing in the picture (it's delta_.../lb-stuff/union-suffix iirc). 

> HiveInputFormat.processForWriteIds(Path dir, JobConf conf,       ValidTxnList validTxnList, List<Path> finalPaths)
> ------------------------------------------------------------------------------------------------------------------
>
>                 Key: HIVE-17695
>                 URL: https://issues.apache.org/jira/browse/HIVE-17695
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Transactions
>            Reporter: Eugene Koifman
>
> this has special handling for writes resulting from Union All query
> In full Acid case at least, these subdirs get collapsed in favor of statementId based dir names (delta_x_y_stmtId).  It would be cleaner/simpler to make MM follow the same logic.  (full acid does it Hive.moveFiles() I think)



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)