[
https://issues.apache.org/jira/browse/HADOOP-4967?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12665066#action_12665066
]
Amareshwari Sriramadasu commented on HADOOP-4967:
-------------------------------------------------
Patch looks fine.
> Inconsistent state in JVM manager
> ---------------------------------
>
> Key: HADOOP-4967
> URL: https://issues.apache.org/jira/browse/HADOOP-4967
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Reporter: Amareshwari Sriramadasu
> Assignee: Devaraj Das
> Attachments: 4967.patch
>
>
> The following is the scenario:
> -> TT got a LaunchTaskAction. And It found the slot free. It started the
> TaskRunner
> -> TT received KillJobActtion. It did a runner.kill().
> But, runner.kill() didn't remove JVM Runner from jvmIdToRunner , as it is not
> polulated yet.
> It released the slots.
> -> TaskRunner thread spawned jvm and populated jvmIdToRunner.
> -> The next task getting launched on the slot finds it as busy.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
