[
https://issues.apache.org/jira/browse/HIVE-17939?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16227173#comment-16227173
]
Deepak Jaiswal commented on HIVE-17939:
---------------------------------------
It is the same predictability problem with spark. updating the test to use
order by.
> Bucket map join not being selected when bucketed tables is missing bucket
> files
> -------------------------------------------------------------------------------
>
> Key: HIVE-17939
> URL: https://issues.apache.org/jira/browse/HIVE-17939
> Project: Hive
> Issue Type: Bug
> Reporter: Deepak Jaiswal
> Assignee: Deepak Jaiswal
> Attachments: HIVE-17939.1.patch
>
>
> Looks like the following logic kicks in during
> OpTraitsRulesProcFactory.TableScanRule.checkBucketedTable(), which prevents
> the table from being considered a proper bucketed table:
> // The number of files for the table should be same as number of
> // buckets.
> if (fileNames.size() != 0 && fileNames.size() != numBuckets) {
> return false;
> }
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
