[jira] Updated: (HIVE-276) input3_limit.q fails under 0.17

["Zheng Shao (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HIVE-276?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Zheng Shao updated HIVE-276:
----------------------------

    Attachment: HIVE-276.1.patch

Modified the query to do another SORT at the end.

I also thought about propagating the sort order to the reduce sink operator of limit, but that does not seem very easy to do.


> input3_limit.q fails under 0.17
> -------------------------------
>
>                 Key: HIVE-276
>                 URL: https://issues.apache.org/jira/browse/HIVE-276
>             Project: Hadoop Hive
>          Issue Type: Bug
>            Reporter: Zheng Shao
>         Attachments: HIVE-276.1.patch
>
>
> The plan ql/src/test/results/clientpositive/input3_limit.q.out shows that there are 2 map-reduce jobs:
> The first one is distributed and sorted as is specified by the query. The reducer side has LIMIT 20.
> The second one (single reducer job imposed by LIMIT 20) does not have the same sort order, so the final result is non-deterministic.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.