[ 
https://issues.apache.org/jira/browse/YARN-1012?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14522818#comment-14522818
 ] 

Inigo Goiri commented on YARN-1012:
-----------------------------------

Thanks for adjusting the JIRAs [~vinodkv].

My proposal is to get the data from the ContainerMonitorImpl, aggregate it and 
put it in a new entity called ResourceUtilization which will contain for now 
CPU as a float and memory. Then, this will go into in the NodeHealthStauts.

The other option is to add the utilization per container using the 
ResourceUtilization again but adding it per each container into the 
ContainerStatus.

Both of them would require the new ResourceUtilization which in this case for 
simplicity will just be it without the whole ResourceUtilizations layer.

Which one you guys prefer?

> NM should report resource utilization of running containers to RM in heartbeat
> ------------------------------------------------------------------------------
>
>                 Key: YARN-1012
>                 URL: https://issues.apache.org/jira/browse/YARN-1012
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>            Reporter: Arun C Murthy
>            Assignee: Inigo Goiri
>




--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to