[
https://issues.apache.org/jira/browse/YARN-4597?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Arun Suresh updated YARN-4597:
------------------------------
Attachment: YARN-4597.001.patch
Attaching initial patch.
* Introduced a {{ContainerScheduler}} : The Scheduler can decide when a
Container can be launched. This can be done based on policy. Currently, there
is flag that it checks to see if Queuing is enabled, it performs the role of
the {{QueuingContainerManagerImpl}}
* Removed the {{QueuingContainerManagerImpl}} : As described above, it is not
required anymore. Most of the logic has been moved to the Scheduler.
* There is no need for a {{QueuingContext}} anymore.
* {{TestQueuingContainerManager}} has been moved to
{{TestContainerSchedulerQueuing}}
* The {{LOCALIZED}} container state has been renamed to {{SCHEDULED}}
Since a lot of code paths have been simplified, a lot of code changes are
actually deletions.
Do take a look. (cc: [~kasha], [~chris.douglas], [~vvasudev], [~jianhe],
[~kkaranasos], [~subru])
> Add SCHEDULE to NM container lifecycle
> --------------------------------------
>
> Key: YARN-4597
> URL: https://issues.apache.org/jira/browse/YARN-4597
> Project: Hadoop YARN
> Issue Type: Bug
> Components: nodemanager
> Reporter: Chris Douglas
> Assignee: Arun Suresh
> Attachments: YARN-4597.001.patch
>
>
> Currently, the NM immediately launches containers after resource
> localization. Several features could be more cleanly implemented if the NM
> included a separate stage for reserving resources.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
