You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-dev@hadoop.apache.org by "Ed Kohlwey (JIRA)" <ji...@apache.org> on 2009/11/19 19:55:39 UTC

[jira] Created: (MAPREDUCE-1223) CompositeInputFormat doesn't consider all tuples when run in a local task tracker

CompositeInputFormat doesn't consider all tuples when run in a local task tracker
---------------------------------------------------------------------------------

                 Key: MAPREDUCE-1223
                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-1223
             Project: Hadoop Map/Reduce
          Issue Type: Bug
    Affects Versions: 0.20.1
         Environment: Yahoo distribution for Hadoop 0.20.1.3041192001 and Cloudera Distribution for Hadoop 0.20.1+133
            Reporter: Ed Kohlwey


The CrossJoin class does not emit all tuples representing the cross product of values for a given key. The issue only occurs when using the local task tracker, and not when running the job on a cluster. 

Example
{noformat}
table 1
k1 -> a

table 2
k1 ->c
k1 ->d
{noformat}

The expected output is
{noformat}
table 1 inner join table 2
k1->ac
k1->ad
{noformat}

Instead one gets
{noformat}
table 1 inner join table 2
k1->ac
{noformat}

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

[jira] Resolved: (MAPREDUCE-1223) CompositeInputFormat doesn't consider all tuples when run in a local task tracker

Posted by "Ed Kohlwey (JIRA)" <ji...@apache.org>.
     [ https://issues.apache.org/jira/browse/MAPREDUCE-1223?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Ed Kohlwey resolved MAPREDUCE-1223.
-----------------------------------

    Resolution: Invalid

After some additional testing I'm marking this as invalid. It appears that the issue was with one of the inputs not being sorted.

> CompositeInputFormat doesn't consider all tuples when run in a local task tracker
> ---------------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-1223
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-1223
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>    Affects Versions: 0.20.1
>         Environment: Yahoo distribution for Hadoop 0.20.1.3041192001 and Cloudera Distribution for Hadoop 0.20.1+133
>            Reporter: Ed Kohlwey
>
> The CrossJoin class does not emit all tuples representing the cross product of values for a given key. The issue only occurs when using the local task tracker, and not when running the job on a cluster. 
> Example
> {noformat}
> table 1
> k1 -> a
> table 2
> k1 ->c
> k1 ->d
> {noformat}
> The expected output is
> {noformat}
> table 1 inner join table 2
> k1->ac
> k1->ad
> {noformat}
> Instead one gets
> {noformat}
> table 1 inner join table 2
> k1->ac
> {noformat}

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