[
https://issues.apache.org/jira/browse/YARN-396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13591020#comment-13591020
]
Zhijie Shen commented on YARN-396:
----------------------------------
AMResponse seems not to be necessary. Whenever an AllocateResponse is created,
there must be a corresponding AMResponse. Whenever the AllocateResponse is
received, the AMResponse must be retrieved. NumClusterNodes of AMResponse and
the properties of AMResponse doesn't seem to be inconsistent. In fact, one of
the AMResponse's properties ia about the node as well, i.e., UpdatedNodes.
Hence it should be good to collapse the data structure. It's better to keep
AllocateResponse but discard AMResponse, because the API doesn't need to be
changed, and it is more natural to send an AllocateResquest and receive an
AllocateResponse consequently. Therefore, AMResponse can be safely removed, and
its properties are moved to AllocateResponse. The drawback of this strategy is
more touch in the code, compared to keeping AMResponse while discarding
AllocateResponse. However, it should be manageable.
> Rationalize AllocateResponse in RM scheduler API
> ------------------------------------------------
>
> Key: YARN-396
> URL: https://issues.apache.org/jira/browse/YARN-396
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Bikas Saha
> Assignee: Arun C Murthy
>
> AllocateResponse contains an AMResponse and cluster node count. AMResponse
> that more data. Unless there is a good reason for this object structure,
> there should be either AMResponse or AllocateResponse.
--
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
