hadoop-mapreduce-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Patrick Angeles (JIRA)" <j...@apache.org>
Subject [jira] Created: (MAPREDUCE-2391) Improve fair-scheduler documentation regarding 'weight' parameter
Date Wed, 16 Mar 2011 20:36:29 GMT
Improve fair-scheduler documentation regarding 'weight' parameter

                 Key: MAPREDUCE-2391
                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2391
             Project: Hadoop Map/Reduce
          Issue Type: Improvement
          Components: contrib/fair-share
            Reporter: Patrick Angeles

The weight that is displayed on the FairSchedulerServlet is the 'job' weight, which is not
to be confused with 'pool' weight which is specified in the configuration (fair-scheduler.xml
or whathaveyou).

The pool weight influences the fair-share allocation for a pool (and by extension the jobs
within a pool).

The job weight affects a job's share or priority within a pool. This is affected by priority
and also if you specify sizebasedweight=true or if you specify a class for mapred.fairscheduler.weightadjuster
(for example, org.apache.hadoop.mapred.NewJobWeightBooster). This is the weight that shows
up on the FairSchedulerServlet.

This can be quite confusing, and is worth documenting.

This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message