hadoop-yarn-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Devaraj K (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (YARN-3826) Race condition in ResourceTrackerService: potential wrong diagnostics messages
Date Fri, 19 Jun 2015 11:03:00 GMT

    [ https://issues.apache.org/jira/browse/YARN-3826?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14593318#comment-14593318
] 

Devaraj K commented on YARN-3826:
---------------------------------

Thanks [~chengbing.liu] for the patch.

bq. Since we are calling setDiagnosticsMessage in nodeHeartbeat, which can be called concurrently,
the static resync and shutdown may have wrong diagnostics messages in some cases.
I see diagnostics message is setting into resync/shutdown response objects before returning
them. Can you give more details about a case where race condition could occur?

bq. these static members can hardly save any memory, since the normal heartbeat responses
are created for each heartbeat.
I agree on this.

> Race condition in ResourceTrackerService: potential wrong diagnostics messages
> ------------------------------------------------------------------------------
>
>                 Key: YARN-3826
>                 URL: https://issues.apache.org/jira/browse/YARN-3826
>             Project: Hadoop YARN
>          Issue Type: Bug
>          Components: resourcemanager
>    Affects Versions: 2.7.0
>            Reporter: Chengbing Liu
>            Assignee: Chengbing Liu
>         Attachments: YARN-3826.01.patch
>
>
> Since we are calling {{setDiagnosticsMessage}} in {{nodeHeartbeat}}, which can be called
concurrently, the static {{resync}} and {{shutdown}} may have wrong diagnostics messages in
some cases.
> On the other side, these static members can hardly save any memory, since the normal
heartbeat responses are created for each heartbeat.



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

Mime
View raw message