Naganarasimha G R commented on YARN-3045:

Hi [~djp],
Sorry for the delayed response! and some points to discuss for your queries as 
follows :
bq why we hook the track of container start event in ContainerManagerImpl, but 
for container finished event, we do it inside of ContainerImpl? We should try 
to keep NMTimelinePublisher get referenced in one place if no necessary for 
other places. 
This was intentionally done to avoid resending of timelineevents during 
recovery. In RM's case also it was happening(which is being handled in 
YARN-3127) hence to avoid duplicate events have kept it there. If any better 
ways to avoid, i am open for it .

Other comments will take care, some of it are due to missing to revert the code 
while testing ...

> [Event producers] Implement NM writing container lifecycle events to ATS
> ------------------------------------------------------------------------
>                 Key: YARN-3045
>                 URL: https://issues.apache.org/jira/browse/YARN-3045
>             Project: Hadoop YARN
>          Issue Type: Sub-task
>          Components: timelineserver
>            Reporter: Sangjin Lee
>            Assignee: Naganarasimha G R
>         Attachments: YARN-3045-YARN-2928.002.patch, 
> YARN-3045-YARN-2928.003.patch, YARN-3045-YARN-2928.004.patch, 
> YARN-3045-YARN-2928.005.patch, YARN-3045.20150420-1.patch
> Per design in YARN-2928, implement NM writing container lifecycle events and 
> container system metrics to ATS.

This message was sent by Atlassian JIRA

Reply via email to