[
https://issues.apache.org/jira/browse/YARN-1015?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16243117#comment-16243117
]
Haibo Chen commented on YARN-1015:
----------------------------------
bq. Could the allowed resource for overallocation change later while the
function is still running?
assignContainer only happens in the scheduler, which serializes scheduling
requests, so we should
be OK most of the time.
The handling of container decrease/demotion, however, bypasses scheduler and
allocates
RMContainers directly followed by an immediate swap of existing containers. It
will cause the
accounting of opportunistic containers to go up briefly, and if the scheduler
happens to be allocating
opportunistic containers in that window, less opportunistic containers will be
allocated. But the accounting
will drop shortly after AM pulls the updated containers, and we'll be able to
allocate opportunistic
containers later.
> FS should watch node resource utilization and allocate opportunistic
> containers if appropriate
> ----------------------------------------------------------------------------------------------
>
> Key: YARN-1015
> URL: https://issues.apache.org/jira/browse/YARN-1015
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Arun C Murthy
> Assignee: Haibo Chen
> Attachments: YARN-1015-YARN-1011.00.patch,
> YARN-1015-YARN-1011.01.patch, YARN-1015-YARN-1011.02.patch,
> YARN-1015-YARN-1011.prelim.patch
>
>
> FS should looks at resource utilization of nodes (provided by NM in
> heartbeat) and allocate opportunistic containers if the resource utilization
> of the node is below its allocation threshold.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
