Jason Lowe commented on YARN-2202:

I don't currently have any plans for this.  Off the top of my head, we could 
persist metric changes to the state store so we can recover their previous 
values upon restart.  One issue with that approach is that recovery causes 
resources and containers to progress through their state transitions almost as 
if they were starting again from scratch, so we would need to watch out for 
double-counting during the recovery process.

> Metrics recovery for nodemanager restart
> ----------------------------------------
>                 Key: YARN-2202
>                 URL: https://issues.apache.org/jira/browse/YARN-2202
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: nodemanager
>            Reporter: Jason Lowe
> Per a side discussion in the review of YARN-1341, we should investigate what 
> metrics need to be persisted and recovered as part of NM restart.

This message was sent by Atlassian JIRA

Reply via email to