hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Doug Cutting (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-1431) Map tasks can't timeout for failing to call progress
Date Fri, 25 May 2007 16:24:16 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-1431?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12499140
] 

Doug Cutting commented on HADOOP-1431:
--------------------------------------

Comparators can be performance-sensitive, so I'd like to see some benchmarks of a ReportingComparator
before we accept that solution.

The long-term best patch for this would be to modify SequenceFile to actually report progress
as it sorts.  This is a long-standing issue.

The current bug is that the progress thread is running when there's no sorting going on. 
It should only run during calls to MapTask#sortAndSpillToDisk(), not during the entire map.
 This should be easy to patch for the short term.

> Map tasks can't timeout for failing to call progress
> ----------------------------------------------------
>
>                 Key: HADOOP-1431
>                 URL: https://issues.apache.org/jira/browse/HADOOP-1431
>             Project: Hadoop
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.13.0
>            Reporter: Owen O'Malley
>         Assigned To: Arun C Murthy
>             Fix For: 0.13.0
>
>         Attachments: HADOOP-1431_1_20070525.patch
>
>
> Currently the map task runner creates a thread that calls progress every second to keep
the system from killing the map if the sort takes too long. This is the wrong approach, because
it will cause stuck tasks to not be killed. The right solution is to have the sort call progress
as it actually makes progress. This is part of what is going on in HADOOP-1374. A map gets
stuck at 100% progress, but not done.

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


Mime
View raw message