Mailing-List: contact core-dev-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: core-dev@hadoop.apache.org
Message-ID: <934040132.1217535391832.JavaMail.jira@brutus>
Date: Thu, 31 Jul 2008 13:16:31 -0700 (PDT)
From: "Doug Cutting (JIRA)" <jira@apache.org>
To: core-dev@hadoop.apache.org
Subject: [jira] Updated: (HADOOP-153) skip records that fail Task
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


     [ https://issues.apache.org/jira/browse/HADOOP-153?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Doug Cutting updated HADOOP-153:
--------------------------------

    Status: Open  (was: Patch Available)

Sigh.  Sorry it's taken so long for me to get back to this.  The patch has now gone stale!  My bad.

It looks much better.

The configuration parameters controlling skipping should all start with "mapred.skip".  Private constants should probably be defined for these, since they're referred to more than once.  That way the compiler will check that they're all spelled correctly.  The term "isSkipModeKickedOff" might instead be just "inSkipMode" or simply "isSkipping", and "skip.mode.kicked.off" might be "mapred.skip.on" or somesuch.

It  would also be good to add javadoc to the new public Task methods and to SortedRange's public methods.  We don't (yet) publish the javadoc for Task, but it is a pretty central class and deserves to be well documented.  In general, public methods in public classes should have javadoc.  Sometimes classes (like Task) which didn't used to be public are made public, and lots of their methods are then missing javadoc, but that's not an excuse to continue the practice.  And it never hurts to have javadoc, even for non-public methods in non-public classes...

> skip records that fail Task
> ---------------------------
>
>                 Key: HADOOP-153
>                 URL: https://issues.apache.org/jira/browse/HADOOP-153
>             Project: Hadoop Core
>          Issue Type: New Feature
>          Components: mapred
>    Affects Versions: 0.2.0
>            Reporter: Doug Cutting
>            Assignee: Sharad Agarwal
>             Fix For: 0.19.0
>
>         Attachments: 153_1.patch, 153_2.patch, 153_3.patch, 153_4.patch, 153_5.patch, skipRecords_wip1.patch
>
>
> MapReduce should skip records that throw exceptions.
> If the exception is thrown under RecordReader.next() then RecordReader implementations should automatically skip to the start of a subsequent record.
> Exceptions in map and reduce implementations can simply be logged, unless they happen under RecordWriter.write().  Cancelling partial output could be hard.  So such output errors will still result in task failure.
> This behaviour should be optional, but enabled by default.  A count of errors per task and job should be maintained and displayed in the web ui.  Perhaps if some percentage of records (>50%?) result in exceptions then the task should fail.  This would stop jobs early that are misconfigured or have buggy code.
> Thoughts?

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