[ https://issues.apache.org/jira/browse/YARN-3995?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15070660#comment-15070660 ]
Naganarasimha G R commented on YARN-3995: ----------------------------------------- Oops, Sorry my mistake , Thanks [~sjlee0] for correcting me. [~sjlee0] current code is already waiting for a second in a separate thread after AM container is closed (in PerNodeTimelineCollectorsAuxService.stopContainer), but the issue with that approach is: it just closes after 1 second though the events are still coming, but what i am trying to suggest is close/remove the collector only after a period of inactivity in the collector. Will that be good considering it will be usually getting delayed for metrics ? if above approach is not required then already existing approach waits for a second in separate thread, does it req any change ? (least i can think is few threads will be there if more AM's are run from a single NM ) > Some of the NM events are not getting published due race condition when AM > container finishes in NM > ---------------------------------------------------------------------------------------------------- > > Key: YARN-3995 > URL: https://issues.apache.org/jira/browse/YARN-3995 > Project: Hadoop YARN > Issue Type: Sub-task > Components: nodemanager, timelineserver > Affects Versions: YARN-2928 > Reporter: Naganarasimha G R > Assignee: Naganarasimha G R > Labels: yarn-2928-1st-milestone > > As discussed in YARN-3045: While testing in TestDistributedShell found out > that few of the container metrics events were failing as there will be race > condition. When the AM container finishes and removes the collector for the > app, still there is possibility that all the events published for the app by > the current NM and other NM are still in pipeline, -- This message was sent by Atlassian JIRA (v6.3.4#6332)