Mailing-List: contact dev-help@pig.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@pig.apache.org
Date: Thu, 12 Jan 2017 00:10:16 +0000 (UTC)
From: "Rohini Palaniswamy (JIRA)" <jira@apache.org>
To: pig-dev@hadoop.apache.org
Message-ID: <JIRA.13033992.1484173706000.2458.1484179816739@Atlassian.JIRA>
In-Reply-To: <JIRA.13033992.1484173706000@Atlassian.JIRA>
References: <JIRA.13033992.1484173706000@Atlassian.JIRA> <JIRA.13033992.1484173706042@arcas>
Subject: [jira] [Updated] (PIG-5088) HashValuePartitioner has skew when
 there is only map fields
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit
archived-at: Thu, 12 Jan 2017 00:10:18 -0000


     [ https://issues.apache.org/jira/browse/PIG-5088?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Rohini Palaniswamy updated PIG-5088:
------------------------------------
      Resolution: Fixed
    Hadoop Flags: Reviewed
          Status: Resolved  (was: Patch Available)

Committed to branch-0.16 and trunk. Thanks for the review Koji.

> HashValuePartitioner has skew when there is only map fields
> -----------------------------------------------------------
>
>                 Key: PIG-5088
>                 URL: https://issues.apache.org/jira/browse/PIG-5088
>             Project: Pig
>          Issue Type: Bug
>            Reporter: Rohini Palaniswamy
>            Assignee: Rohini Palaniswamy
>             Fix For: 0.17.0, 0.16.1
>
>         Attachments: PIG-5088-1.patch
>
>
>   One user who just did union of map fields, got a skew with everything going to one input file as hashcode for map was calculated on the size. [~knoguchi] pointed out that my assumption of map hashcode being non-deterministic and can change with jdk versions is wrong as it just does a sum of hashcode of its entries.


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