Karthik Kambatla commented on YARN-4485:

One potential approach is to add requestTime to ResourceRequest and use it to 
compute the difference at allocation time to get per-container allocation 
latency. This could be optionally rolled up at application and queue level. 

> Capture per-application and per-queue container allocation latency
> ------------------------------------------------------------------
>                 Key: YARN-4485
>                 URL: https://issues.apache.org/jira/browse/YARN-4485
>             Project: Hadoop YARN
>          Issue Type: Improvement
>    Affects Versions: 2.7.1
>            Reporter: Karthik Kambatla
>            Assignee: Karthik Kambatla
>              Labels: supportability, tuning
> Per-application and per-queue container allocation latencies would go a long 
> way towards help with tuning scheduler queue configs. 
> This umbrella JIRA tracks adding these metrics. 

This message was sent by Atlassian JIRA

Reply via email to