[
https://issues.apache.org/jira/browse/MAPREDUCE-5348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13691331#comment-13691331
]
Karthik Kambatla commented on MAPREDUCE-5348:
---------------------------------------------
MR2 (the version of MR in branch-2) runs over YARN and doesn't have
JobTracker/TaskTracker. Some of the methods there are aimed towards maintaining
compatibility with 1.x releases.
> There are innconsistencies in the design of the Cluster and other classes.
> --------------------------------------------------------------------------
>
> Key: MAPREDUCE-5348
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5348
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Affects Versions: 2.1.0-beta, 2.0.5-alpha
> Reporter: Pranay Varma
>
> Cluster seems to keep TaskTracker information using class TaskTrackerInfo[].
> getTaskTrackers returns TaskTrackerInfo[]. Ideally, I would like it to simply
> return TaskTracker[] or else have the method be called getTaskTrackerInfos.
> A user would expect JobTracker information as well. For consistency, I would
> expect a getJobTracker method to return JobTrackerInfo or JobTracker. This
> method is not available, instead I see getJobTrackerStatus which returns
> JobTrackerStatus.
> Although Cluster has TaskTracker and JobTracker, it uses QueueInfo instead of
> JobQueueInfo, which leaves one wondering what other kinds of queues are
> there. There is no documentation.
> Not clear why Scheduling info is a totally unstructured string. Perhaps some
> information should be structured and the class should contain some String
> member to keep the unstructured information for extensibility.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira