[
https://issues.apache.org/jira/browse/HIVE-15899?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16013166#comment-16013166
]
Eugene Koifman edited comment on HIVE-15899 at 8/16/17 7:40 PM:
----------------------------------------------------------------
When bucketing requirement is relaxed, CTAS can be successful. How will acid
write look like?
It cannot retain 1/ and 2/ subdirectories. Should each leg rely on writerId to
create a set of unique tranche files across all subdirs so that acid "move" can
put all of them into a delta?
Then how do we later remove the "move" for S3?
perhaps it should rely on statementId and create delta_x_x_0, delta_x_x_1, etc
was (Author: ekoifman):
When bucketing requirement is relaxed, CTAS can be successful. How will acid
write look like?
It cannot retain 1/ and 2/ subdirectories. Should each leg rely on writerId to
create a set of unique tranche files across all subdirs so that acid "move" can
put all of them into a delta?
Then how do we later remove the "move" for S3?
> check CTAS over acid table
> ---------------------------
>
> Key: HIVE-15899
> URL: https://issues.apache.org/jira/browse/HIVE-15899
> Project: Hive
> Issue Type: Task
> Reporter: Eugene Koifman
> Assignee: Eugene Koifman
>
> need to add a test to check if create table as works correctly with acid
> tables
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
