You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kylin.apache.org by "Shaofeng SHI (JIRA)" <ji...@apache.org> on 2016/10/17 05:54:58 UTC
[jira] [Commented] (KYLIN-2097) Get 'Column does not exist in row
key desc" on cube has TopN measure
[ https://issues.apache.org/jira/browse/KYLIN-2097?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15581316#comment-15581316 ]
Shaofeng SHI commented on KYLIN-2097:
-------------------------------------
Made change on master with commit id: c3bfd770af93db7e7c295846898851c4bdc458c5
> Get 'Column does not exist in row key desc" on cube has TopN measure
> --------------------------------------------------------------------
>
> Key: KYLIN-2097
> URL: https://issues.apache.org/jira/browse/KYLIN-2097
> Project: Kylin
> Issue Type: Bug
> Components: Query Engine
> Reporter: Shaofeng SHI
> Assignee: Shaofeng SHI
> Fix For: v1.6.0
>
>
> Define a cube which has a TopN measure to collect the top tags on Twitter; when query with the tag as a "where" condition,
> {code}
> select TWITTER_TAG3.HOUR_START , hashtag, count(*) as num from TWITTER_TAG3 where hashtag = 'ALDUBAnew' group by HOUR_START, hashtag order by HOUR_START
> {code}
> As TopN doesn't put the UHC column into row key, it couldn't be used as the filtering condition, then Kylin reports error: "Column DEFAULT.TWITTER_TAG3.HASHTAG does not exist in row key desc"
> In this ccase TopN should say no for this query as it doesn't have the capacity, then Kylin can rounte to other cubes.
> For user who only has one cube, he should change to use "having" clause instead of "where", although the performance is not good, it can return result in the end.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)