[
https://issues.apache.org/jira/browse/YARN-573?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13632146#comment-13632146
]
Omkar Vinit Joshi commented on YARN-573:
----------------------------------------
We need to fix one more problem present in Private localization.
* Current behavior : LocalizerRunner is started when container makes first
request for the resource. It internally starts ContainerLocalizer which starts
localizing. However as soon as pending becomes empty ContainerLocalizer is
asked to kill itself ( LocalizerAction.DIE - check update method). Now if the
resource localization request comes little later than this then the request
will never be handled ( localized) as ContainerLocalizer is never restarted. At
present this is not occurring because first we are localizing resources one at
a time and events are sent really very fast to ContainerLocalizer.
* Expected behavior : So as soon as ContainerLocalizer is done localizing all
the resources (pending becomes empty); LocalizerRunner will remove itself from
the available localizers before asking ContainerLocalizer to kill itself.
> Shared data structures in Public Localizer and Private Localizer are not
> Thread safe.
> -------------------------------------------------------------------------------------
>
> Key: YARN-573
> URL: https://issues.apache.org/jira/browse/YARN-573
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Omkar Vinit Joshi
> Assignee: Omkar Vinit Joshi
>
> PublicLocalizer
> 1) pending accessed by addResource (part of event handling) and run method
> (as a part of PublicLocalizer.run() ).
> PrivateLocalizer
> 1) pending accessed by addResource (part of event handling) and
> findNextResource (i.remove()).
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
