hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Devaraj Das (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HADOOP-5285) JobTracker hangs for long periods of time
Date Fri, 20 Feb 2009 10:35:04 GMT

     [ https://issues.apache.org/jira/browse/HADOOP-5285?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Devaraj Das updated HADOOP-5285:
--------------------------------

    Attachment: 5285.patch

Attaching patch. The patch
1) Fixes obtainTaskCleanupTask to check whether the job got inited before it tries to take
a lock on JobInProgress
2) Moves the CleanupQueue class outside the TaskTracker to a generic class. The JobTracker
also uses that class for queueing up deletes during job finalization.
3) Moves the references to completedJobStore outside the block where the JobTracker is locked.

> JobTracker hangs for long periods of time
> -----------------------------------------
>
>                 Key: HADOOP-5285
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5285
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: mapred
>    Affects Versions: 0.20.0
>            Reporter: Vinod K V
>            Assignee: Devaraj Das
>            Priority: Blocker
>             Fix For: 0.20.0
>
>         Attachments: 5285.patch, trace.txt
>
>
> On one of the larger clusters of 2000 nodes, JT hanged quite often, sometimes for times
in the order of 10-15 minutes and once for one and a half hours(!). The stack trace shows
that JobInProgress.obtainTaskCleanupTask() is waiting for lock on JobInProgress object which
JobInProgress.initTasks() is holding for a long time waiting for DFS operations.

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


Mime
View raw message