[
https://issues.apache.org/jira/browse/HADOOP-14354?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15983912#comment-15983912
]
Inigo Goiri commented on HADOOP-14354:
--------------------------------------
In the NodeManager there are two threads that indirectly read values from
{{SysInfo}}:
* {{NodeResourceMonitor}}
* {{ContainersMonitor}}
We have seen some cases where values get to weird values once in a while.
The reason is that once in a while two threads try to get a value and both get
into {{refreshIfNeeded()}}.
Even if the refresh was thread safe, there is no point on two threads trying to
refresh these values at the same time.
> SysInfoWindows is not thread safe
> ---------------------------------
>
> Key: HADOOP-14354
> URL: https://issues.apache.org/jira/browse/HADOOP-14354
> Project: Hadoop Common
> Issue Type: Bug
> Reporter: Inigo Goiri
> Assignee: Inigo Goiri
> Attachments: HADOOP-14354-000.patch
>
>
> Multiple threads can trigger {{refreshIfNeeded()}} in {{SysInfoWindows}} at
> the same time and cause undesired behaviors.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
