[
https://issues.apache.org/jira/browse/YARN-5100?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15286552#comment-15286552
]
Junping Du commented on YARN-5100:
----------------------------------
Thanks [~xgong] for reporting the issue and deliver the patch to fix it.
Comparing with reverse the sequence of events, I would prefer the fix on event
handling logic: when application state is already in FINISHED state, then we
can skip event with STATE_UPDATED_EVENT_TYPE. That sounds more robust. What do
you think?
BTW, the patch is not apply to trunk. Would you rebase the patch against trunk?
> The YarnApplicationState is always running in ATS no matter the application
> is running or finishes.
> ---------------------------------------------------------------------------------------------------
>
> Key: YARN-5100
> URL: https://issues.apache.org/jira/browse/YARN-5100
> Project: Hadoop YARN
> Issue Type: Bug
> Affects Versions: 2.8.0
> Reporter: Xuan Gong
> Assignee: Xuan Gong
> Priority: Blocker
> Attachments: YARN-5100.1.patch
>
>
> After YARN-5029, we add one more event : APP_STATE_UPDATE event which is used
> by RM to sync up the App state between Timeline Server. But when we get
> appReport from TimelineServer, we parse all events with timestamp descending
> order (Finish event-->App State update event --> create event) which causes
> this issue.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
