[
https://issues.apache.org/jira/browse/YARN-3491?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14496663#comment-14496663
]
Jason Lowe commented on YARN-3491:
----------------------------------
Could you elaborate a bit on why the submit is time consuming? Unless I'm
mistaken, the FSDownload constructor is very cheap and queueing should be
simply tacking an entry on a queue.
> Improve the public resource localization to do both FSDownload submission to
> the thread pool and completed localization handling in one thread
> (PublicLocalizer).
> -----------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Key: YARN-3491
> URL: https://issues.apache.org/jira/browse/YARN-3491
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: nodemanager
> Affects Versions: 2.7.0
> Reporter: zhihai xu
> Assignee: zhihai xu
> Priority: Critical
>
> Improve the public resource localization to do both FSDownload submission to
> the thread pool and completed localization handling in one thread
> (PublicLocalizer).
> Currently FSDownload submission to the thread pool is done in
> PublicLocalizer#addResource which is running in Dispatcher thread and
> completed localization handling is done in PublicLocalizer#run which is
> running in PublicLocalizer thread.
> Because FSDownload submission to the thread pool at the following code is
> time consuming, the thread pool can't be fully utilized. Instead of doing
> public resource localization in parallel(multithreading), public resource
> localization is serialized most of the time.
> {code}
> synchronized (pending) {
> pending.put(queue.submit(new FSDownload(lfs, null, conf,
> publicDirDestPath, resource,
> request.getContext().getStatCache())),
> request);
> }
> {code}
> Also there are two more benefits with this change:
> 1. The Dispatcher thread won't be blocked by above FSDownload submission.
> Dispatcher thread handles most of time critical events at Node manager.
> 2. don't need synchronization on HashMap (pending).
> Because pending will be only accessed in PublicLocalizer thread.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
