Varun Vasudev commented on YARN-3534:

Thanks for the path [~elgoiri]! Is it possible for you to split up the patch 
into two - one to record the memory and cpu utilization in 
NodeResourceMonitorImpl and one for the RPC changes. If you could file a 
separate JIRA for the NodeResourceMonitorImpl changes that you have made(they 
are definitely required and should go in), that would be ideal.

I'm not sure if we should use RPC  to expose the stats. Using RPC restricts the 
stats to a limited number of services whereas using REST lets a larger number 
of services access the stats - which is why YARN-3332 wants to use REST to 
expose the NM stats. Using REST should also make it easier to add more stats 
with respect to utilization.

> Report node resource utilization
> --------------------------------
>                 Key: YARN-3534
>                 URL: https://issues.apache.org/jira/browse/YARN-3534
>             Project: Hadoop YARN
>          Issue Type: New Feature
>          Components: nodemanager, resourcemanager
>    Affects Versions: 2.7.0
>            Reporter: Inigo Goiri
>            Assignee: Inigo Goiri
>         Attachments: YARN-3534-1.patch, YARN-3534-2.patch, YARN-3534-3.patch, 
> YARN-3534-3.patch
>   Original Estimate: 336h
>  Remaining Estimate: 336h
> YARN should be aware of the resource utilization of the nodes when scheduling 
> containers. For this, this task will implement the NodeResourceMonitor and 
> send this information to the Resource Manager in the heartbeat.

This message was sent by Atlassian JIRA

Reply via email to