You are viewing a plain text version of this content. The canonical link for it is here.

Posted to common-issues@hadoop.apache.org by "John Zhuge (JIRA)" <ji...@apache.org> on 2016/03/20 19:22:33 UTC

[jira] [Updated] (HADOOP-12946) Ensure one JVMPauseMonitor thread per JVM

     [ https://issues.apache.org/jira/browse/HADOOP-12946?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

John Zhuge updated HADOOP-12946:
--------------------------------
    Attachment: HADOOP-12946-001.patch

Patch 001:
* Make the thread object static.
* Use a reference count to track how many instances depend on the thread.
* Ensure the thread is started on the first start call.
* Ensure the thread is only stopped on the last stop call
* Move AbstractService specific test cases from TestJvmMetrics to a new
TestAbstractService class.
* Add JvmPauseMonitor test cases.

> Ensure one JVMPauseMonitor thread per JVM
> -----------------------------------------
>
>                 Key: HADOOP-12946
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12946
>             Project: Hadoop Common
>          Issue Type: Improvement
>            Reporter: John Zhuge
>            Assignee: John Zhuge
>            Priority: Minor
>         Attachments: HADOOP-12946-001.patch
>
>
> Ensure at most one JVMPauseMonitor thread is running per JVM when there are multiple JVMPauseMonitor instances, e.g., in mini clusters. This will prevent redundant GC pause log messages while still maintaining one monitor thread running.
> This is a different way to fix HADOOP-12855.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)