[ 
https://issues.apache.org/jira/browse/HADOOP-153?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12575614#action_12575614
 ] 

Owen O'Malley commented on HADOOP-153:
--------------------------------------

This sounds like roughly the right direction.

Maybe we should extend diagnostic info to contain a record number (or offset?) 
and use that to record the error that was thrown? For errors in 
RecordReader.next, you are right that it probably makes sense to skip ahead to 
the next sync marker/newline. Since the meta data is equally at risk as the 
record data. So having a "gotoSync" in RecordReader would help.

Mappers and Reducers should be able to tolerate killer records, just like the 
reader.

We should think about what is required for the RecordWriter to recover from a 
failed write... Of course I suspect there aren't any good answers until HDFS 
has truncate and append.

> skip records that throw exceptions
> ----------------------------------
>
>                 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: Devaraj Das
>             Fix For: 0.17.0
>
>
> 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.

Reply via email to