You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "MENG DING (JIRA)" <ji...@apache.org> on 2015/08/31 16:47:45 UTC
[jira] [Commented] (YARN-1651) CapacityScheduler side changes to
support increase/decrease container resource.
[ https://issues.apache.org/jira/browse/YARN-1651?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14723493#comment-14723493 ]
MENG DING commented on YARN-1651:
---------------------------------
[~leftnoteasy], I just realized that there is one issue not discussed in the protocol design regarding the {{ContainerResourceChangeRequestProto}} that may affect the scheduler:
{code}
message ContainerResourceChangeRequestProto {
optional ContainerIdProto container_id = 1;
optional ResourceProto capability = 2;
}
{code}
Shall we add a priority field to {{ContainerResourceChangeRequestProto}}? Without a priority field, how does the scheduler decide the priority between increase/decrease request and new allocation request within the same application? Does it simply assume that the increase/decrease request has the highest priority within the same application? If so, that may not be the correct thing to do ... What do you think?
> CapacityScheduler side changes to support increase/decrease container resource.
> -------------------------------------------------------------------------------
>
> Key: YARN-1651
> URL: https://issues.apache.org/jira/browse/YARN-1651
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: resourcemanager, scheduler
> Reporter: Wangda Tan
> Assignee: Wangda Tan
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)