[
https://issues.apache.org/jira/browse/YARN-6246?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Karthik Kambatla updated YARN-6246:
-----------------------------------
Summary: Identifying starved apps does not need the scheduler writelock
(was: Move app starvation identification out of the update thread)
> Identifying starved apps does not need the scheduler writelock
> --------------------------------------------------------------
>
> Key: YARN-6246
> URL: https://issues.apache.org/jira/browse/YARN-6246
> Project: Hadoop YARN
> Issue Type: Sub-task
> Components: fairscheduler
> Affects Versions: 2.9.0
> Reporter: Karthik Kambatla
> Assignee: Karthik Kambatla
>
> Given the update thread holds the scheduler write-lock, we are probably
> better of computing starvation and identification of starved apps in a
> different thread.
> I am averse to adding a thread that runs on a *configurable timeout*, but
> maybe we could trigger this thread after every update run, or do this in the
> update thread but outside of the write-lock.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
