[jira] [Commented] (DRILL-3069) Wrong result for aggregate query with filter on SF100

["Aman Sinha (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/DRILL-3069?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14542885#comment-14542885 ] 

Aman Sinha commented on DRILL-3069:
-----------------------------------

Note that the plan does not contain Sort before StreamAgg because we are doing a plain aggregate (no group-by).  

While looking at it with [~mehant] we found that adding a SelectionVectorRemover after the Filter in the physical plan produces the correct result; however StreamingAgg is supposed to handle SV2, so there's likely an underlying bug there.  

> Wrong result for aggregate query with filter  on SF100 
> -------------------------------------------------------
>
>                 Key: DRILL-3069
>                 URL: https://issues.apache.org/jira/browse/DRILL-3069
>             Project: Apache Drill
>          Issue Type: Bug
>          Components: Execution - Relational Operators
>    Affects Versions: 1.0.0
>            Reporter: Aman Sinha
>            Assignee: Chris Westin
>
> Wrong result on TPCH sf100: 
> {code}
> 0: jdbc:drill:zk=10.10.103.32:5181> select max(l_suppkey) from lineitem where l_suppkey = 3872;
> +------------+
> |   EXPR$0   |
> +------------+
> | 991683     |
> +------------+
> 1 row selected
> {code}
> Plan looks correct: 
> {code}
> +------------+------------+
> |    text    |    json    |
> +------------+------------+
> | 00-00    Screen
> 00-01      StreamAgg(group=[{}], EXPR$0=[MAX($0)])
> 00-02        UnionExchange
> 01-01          StreamAgg(group=[{}], EXPR$0=[MAX($0)])
> 01-02            Filter(condition=[=($0, 3872)])
> 01-03              Scan(groupscan=[ParquetGroupScan [entries=[ReadEntryWithPath [path=maprfs:/data/parquet/tpch/scale100/lineitem]], selectionRoot=/data/parquet/tpch/scale100/lineitem, numFiles=1, columns=[`l_suppkey`]]])
> {code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)