You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hive.apache.org by "Prasanth J (JIRA)" <ji...@apache.org> on 2014/03/30 01:30:15 UTC

[jira] [Commented] (HIVE-6786) Off by one error in ORC PPD

    [ https://issues.apache.org/jira/browse/HIVE-6786?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13954528#comment-13954528 ] 

Prasanth J commented on HIVE-6786:
----------------------------------

[~gopalv] Thats looks like an off-by-one error. It should be >=. I think this case will only be hit when the stripe statistics is missing, in which case the stripeStatistics.size() will be 0 and all stripes should be included.
The attached patch fixes it.

I am curious how is it missing stripe statistics? Is it an old ORC which did not have stripe statistics or is the ORC file generated via other means (apart from hive)?

> Off by one error in ORC PPD 
> ----------------------------
>
>                 Key: HIVE-6786
>                 URL: https://issues.apache.org/jira/browse/HIVE-6786
>             Project: Hive
>          Issue Type: Bug
>    Affects Versions: 0.13.0
>            Reporter: Gopal V
>            Assignee: Prasanth J
>            Priority: Critical
>             Fix For: 0.13.0
>
>         Attachments: HIVE-6786.1.patch
>
>
> Turning on ORC PPD makes split computation fail for a 10Tb benchmark.
> Narrowed down to the following code fragment
> https://github.com/apache/hive/blob/branch-0.13/ql/src/java/org/apache/hadoop/hive/ql/io/orc/OrcInputFormat.java#L757
> {code}
>                 includeStripe[i] = (i > stripeStats.size()) ||
>                     isStripeSatisfyPredicate(stripeStats.get(i), sarg,
>                                              filterColumns);
> {code}
> I would guess that should be a >=, but [~prasanth_j], can you comment if that is the right fix?



--
This message was sent by Atlassian JIRA
(v6.2#6252)