[
https://issues.apache.org/jira/browse/HADOOP-3727?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12612141#action_12612141
]
Steve Loughran commented on HADOOP-3727:
----------------------------------------
One aspect of the taskCleanupThread is that it holds an implicit reference to
the TaskThread, so until the TaskCleanupThread is terminated, the TaskTracker
and its memory will be retained.
> TaskTracker directoryCleanupThread never gets terminated
> --------------------------------------------------------
>
> Key: HADOOP-3727
> URL: https://issues.apache.org/jira/browse/HADOOP-3727
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Affects Versions: 0.19.0
> Reporter: Steve Loughran
> Priority: Minor
>
> When a task tracker starts its work it calls startCleanupThreads() to create
> a directoryCleanupThread.
> only, that thread is never terminated; it runs for the life of the process.
> It should be terminated when the TaskTracker itself is terminated, presumably
> after it has done the last cleanup.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
