[
https://issues.apache.org/jira/browse/HADOOP-5282?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Amar Kamat updated HADOOP-5282:
-------------------------------
Attachment: HADOOP-5282-v1.0.patch
Attaching a fix that logs task-failed line only if the attempt was successful
on the lost tracker. So for an attempt that was successful on the lost tracker
the history lines will look like
{code}
Task Reduce Start
Attempt Reduce Start
Attempt Reduce Killed
Task reduce Killed
{code}
and for the attempt that was running on the lost tracker, the history lines
will look like
{code}
Task Reduce Start
Attempt Reduce Start
Attempt Reduce Failed
{code}
> Running tasks on lost trackers are incorrectly logging the attempt and task
> failures
> ------------------------------------------------------------------------------------
>
> Key: HADOOP-5282
> URL: https://issues.apache.org/jira/browse/HADOOP-5282
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Affects Versions: 0.20.0
> Reporter: Amar Kamat
> Assignee: Amar Kamat
> Priority: Blocker
> Fix For: 0.20.0
>
> Attachments: HADOOP-5282-v1.0.patch
>
>
> If a running attempt is lost on a tracker, a attempt-level cleanup with the
> same attempt-id is launched for it. In such cases the events are not logged
> correctly and instead of logging
> {code}
> Task Reduce Start
> Attempt Reduce Start
> Attempt Reduce Killed
> Task reduce Killed
> {code}
> it logs
> {code}
> Task Reduce Start
> Task Reduce Failed
> Attempt Reduce Start
> Attempt Reduce Killed
> {code}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
