Mailing-List: contact mapreduce-issues-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: mapreduce-issues@hadoop.apache.org
Message-ID: <1319597136.1259743340807.JavaMail.jira@brutus>
Date: Wed, 2 Dec 2009 08:42:20 +0000 (UTC)
From: "Vinod K V (JIRA)" <jira@apache.org>
To: mapreduce-issues@hadoop.apache.org
Subject: [jira] Commented: (MAPREDUCE-1119) When tasks fail to report
 status, show tasks's stack dump before killing
In-Reply-To: <44290688.1255733731270.JavaMail.jira@brutus>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/MAPREDUCE-1119?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12784707#action_12784707 ] 

Vinod K V commented on MAPREDUCE-1119:
--------------------------------------

bq. I am not very sure we want to use the same configuration property for sleeping after dump-stack
bq. So I think the delay is necessary. As for adding a new config item.. maybe wait til someone makes an issue of this? Is there any performance implication here?
Given that the delay is necessary, it is only about the config item - new or the same one. Thinking a bit about it, i am fine with keeping the same configuration because that kind of gives the same time for thread-dump as the time that is given for the process to clean itself up. I know it isn't a very concrete reason but it works for me for now :)

Gone through your latest patch. It looks good now. +1 from my side.

I'll run it through Hudson once again to be sure about the test-cases. May be  {{TestGridmixSubmission}} and {{TestJobHistory.testDoneFolderOnHDFS()}} failures are timing related.

After Hudson blesses, can you ask someone to commit this? Thanks for being patient throughout!

> When tasks fail to report status, show tasks's stack dump before killing
> ------------------------------------------------------------------------
>
>                 Key: MAPREDUCE-1119
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-1119
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>          Components: tasktracker
>    Affects Versions: 0.22.0
>            Reporter: Todd Lipcon
>            Assignee: Aaron Kimball
>         Attachments: MAPREDUCE-1119.2.patch, MAPREDUCE-1119.3.patch, MAPREDUCE-1119.4.patch, MAPREDUCE-1119.5.patch, MAPREDUCE-1119.6.patch, MAPREDUCE-1119.patch
>
>
> When the TT kills tasks that haven't reported status, it should somehow gather a stack dump for the task. This could be done either by sending a SIGQUIT (so the dump ends up in stdout) or perhaps something like JDI to gather the stack directly from Java. This may be somewhat tricky since the child may be running as another user (so the SIGQUIT would have to go through LinuxTaskController). This feature would make debugging these kinds of failures much easier, especially if we could somehow get it into the TaskDiagnostic message

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.