[
https://issues.apache.org/jira/browse/YARN-4597?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15648992#comment-15648992
]
Arun Suresh commented on YARN-4597:
-----------------------------------
[~jianhe], tried another round at refactoring out the 'killedBeforeStart' flag
from the ContainerLaunch. I now feel, it should not be clubbed with the
'shouldLaunchContainer' flag. They signify very different states: the former is
used to ensure the container is never started while the latter is used to
capture the event that the container has already been launch, and thereby
'should not be launched again'. I feel a better name for
'shouldLaunchContainer' should have been 'containerAlreadyLaunched'.
I have raised YARN-5860 and YARN-5861 to track the remaining tasks need to get
this truly feature complete. If [~jianhe] is fine with the current state of
this patch, I would like to check this in. It would be nice though if
[~kkaranasos] also takes a look at this to verify if I havn't missed anything
in the {{ContainerScheduler}} functionality vis-a-vis the
{{QueuingContainerManagerImpl}}
> Add SCHEDULE to NM container lifecycle
> --------------------------------------
>
> Key: YARN-4597
> URL: https://issues.apache.org/jira/browse/YARN-4597
> Project: Hadoop YARN
> Issue Type: New Feature
> Components: nodemanager
> Reporter: Chris Douglas
> Assignee: Arun Suresh
> Labels: oct16-hard
> Attachments: YARN-4597.001.patch, YARN-4597.002.patch,
> YARN-4597.003.patch, YARN-4597.004.patch, YARN-4597.005.patch,
> YARN-4597.006.patch, YARN-4597.007.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]
