Wangda Tan commented on YARN-2885:

Hi [~asuresh],
Thanks for updating.

Looked at latest patch, I majorly looked at configuration changes and codes 
interact with existing RM components. some comments:
- Do you have real use case that distributed scheduler needs to set different 
properties such as DIST_SCHEDULING_MIN_MEMORY? 
Since MIN_MEMORY is a property that AM needs to know (for purpose of 
calculating how much resources to request), we need to tell AM when MIN_MEMORY 
of local RM is different from central RM. I would suggest to use central RM's 
settings for MIN_MEMORY, etc. if you don't have real use case for now.
- First constructor of ApplicationMasterService, should use {{name}} instead of 
- You can add a isDistributedSchedulingEnabled method to YarnConfiguration to 
avoid duplicated logic like:
314         boolean isDistSchedulingEnabled =
315             conf.getBoolean(YarnConfiguration.DIST_SCHEDULING_ENABLED,
316                 YarnConfiguration.DIST_SCHEDULING_ENABLED_DEFAULT);

> Create AMRMProxy request interceptor for distributed scheduling decisions for 
> queueable containers
> --------------------------------------------------------------------------------------------------
>                 Key: YARN-2885
>                 URL: https://issues.apache.org/jira/browse/YARN-2885
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: nodemanager, resourcemanager
>            Reporter: Konstantinos Karanasos
>            Assignee: Arun Suresh
>         Attachments: YARN-2885-yarn-2877.001.patch, 
> YARN-2885-yarn-2877.002.patch, YARN-2885-yarn-2877.full-2.patch, 
> YARN-2885-yarn-2877.full-3.patch, YARN-2885-yarn-2877.full.patch, 
> YARN-2885-yarn-2877.v4.patch, YARN-2885_api_changes.patch
> We propose to add a Local ResourceManager (LocalRM) to the NM in order to 
> support distributed scheduling decisions. 
> Architecturally we leverage the RMProxy, introduced in YARN-2884. 
> The LocalRM makes distributed decisions for queuable containers requests. 
> Guaranteed-start requests are still handled by the central RM.

This message was sent by Atlassian JIRA

Reply via email to