You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@drill.apache.org by "Victoria Markman (JIRA)" <ji...@apache.org> on 2015/03/17 23:40:38 UTC
[jira] [Commented] (DRILL-2483) Make buffer that rows are read into
during execution configurable for testing purposes
[ https://issues.apache.org/jira/browse/DRILL-2483?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14366224#comment-14366224 ]
Victoria Markman commented on DRILL-2483:
-----------------------------------------
Hakim (thank you) pointed me to a discussion on dev mailing list that happened two months ago: https://www.mail-archive.com/dev%40drill.apache.org/msg00551.html
> Make buffer that rows are read into during execution configurable for testing purposes
> --------------------------------------------------------------------------------------
>
> Key: DRILL-2483
> URL: https://issues.apache.org/jira/browse/DRILL-2483
> Project: Apache Drill
> Issue Type: Wish
> Reporter: Victoria Markman
>
> We've found a bug recently where if table had multiple duplicate rows and duplicate rows span multiple buffers, merge join returned wrong result. Test case had a table with 10,000 rows.
> The same problem could be reproduced on a much smaller data set if buffer size was configurable.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)