[jira] [Commented] (KYLIN-1936) Improve enable limit logic (exactAggregation is too strict)

["Zhong Yanghong (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/KYLIN-1936?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16099487#comment-16099487 ] 

Zhong Yanghong commented on KYLIN-1936:
---------------------------------------

Hi [~mahongbin], why we need this precondition for limit push down, could you give me an example? Thanks very much.

{code}
        //if groupsD is clustered at "head" of the rowkey, then limit push down is possible
        int size = groupsD.size();
        if (!groupsD.containsAll(cuboid.getColumns().subList(0, size))) {
            possible = false;
            logger.info("Storage limit push down is impossible because groupD is not clustered at head, groupsD: " + groupsD //
                    + " with cuboid columns: " + cuboid.getColumns());
        }
{code}

> Improve enable limit logic (exactAggregation is too strict)
> -----------------------------------------------------------
>
>                 Key: KYLIN-1936
>                 URL: https://issues.apache.org/jira/browse/KYLIN-1936
>             Project: Kylin
>          Issue Type: Improvement
>    Affects Versions: v1.5.3
>            Reporter: hongbin ma
>            Assignee: hongbin ma
>             Fix For: v1.5.4
>
>
> from zhaotianshuo@meizu.com:
> recently I got the following error while execute query on a cube which is not that big( about 400mb, 20milion record)
> ==================
> Error while executing SQL "select FCRASHTIME,count(1) from UXIP.EDL_FDT_OUC_UPLOAD_FILES group by FCRASH_ANALYSIS_ID,FCRASHTIME limit 1": Scan row count exceeded threshold: 10000000, please add filter condition to narrow down backend scan range, like where clause.
> I guess what  it scan were the intermediate result, but It doesn't any order by,also the result count is limit to just 1.so it could scan to find any record with those two dimension and wala.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)