[
https://issues.apache.org/jira/browse/DRILL-2897?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14533761#comment-14533761
]
Aman Sinha commented on DRILL-2897:
-----------------------------------
This has to do with the combination of singleton distribution trait and
collation. In this particular example, it appears that the child of MergeJoin
is unable to provide a singleton and sorted trait. More investigation is
needed to see why that happens and whether it has to do with the collation
changes done on Calcite.
Since this is independent of LIMIT 0 (you can drop it from your query and still
see the issue), I think you should proceed with the optimization and either
mark the testExampleQueries.testJoin() as ignored or enable exchanges only for
that test. I can file a separate JIRA to track the issue.
> Update Limit 0 to avoid parallelization
> ---------------------------------------
>
> Key: DRILL-2897
> URL: https://issues.apache.org/jira/browse/DRILL-2897
> Project: Apache Drill
> Issue Type: Bug
> Reporter: Jacques Nadeau
> Assignee: Aman Sinha
> Fix For: 1.0.0
>
> Attachments: DRILL-2897.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
