Junping Du commented on YARN-3445:

bq. This doesn't sound right. RM already knows what is running and what isn't. 
Why is NM informing the RM?
I could be mis-leaded by RegisterNodeManagerRequest when NM register to RM (for 
RM restart I think). However, it seems like there is no quick or handy way to 
do NodeId -> runningApps. The existing info there should be RMAppImpl -> 
ranNodes only. May be we need to setup reverse mapping somewhere instead - e.g. 
track runningApplication in RMNodeImpl?  

> NM notify RM on running Apps in NM-RM heartbeat
> -----------------------------------------------
>                 Key: YARN-3445
>                 URL: https://issues.apache.org/jira/browse/YARN-3445
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: nodemanager, resourcemanager
>    Affects Versions: 2.7.0
>            Reporter: Junping Du
>            Assignee: Junping Du
>         Attachments: YARN-3445.patch
> Per discussion in YARN-3334, we need filter out unnecessary collectors info 
> from RM in heartbeat response. Our propose is to add additional field for 
> running apps in NM heartbeat request, so RM only send collectors for local 
> running apps back. This is also needed in YARN-914 (graceful decommission) 
> that if no running apps in NM which is in decommissioning stage, it will get 
> decommissioned immediately. 

This message was sent by Atlassian JIRA

Reply via email to