You are viewing a plain text version of this content. The canonical link for it is here.
Posted to mapreduce-issues@hadoop.apache.org by "Harsh J (JIRA)" <ji...@apache.org> on 2012/07/07 19:34:34 UTC
[jira] [Commented] (MAPREDUCE-78) tasks should not run on nodes
where they were previously lost
[ https://issues.apache.org/jira/browse/MAPREDUCE-78?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13408725#comment-13408725 ]
Harsh J commented on MAPREDUCE-78:
----------------------------------
Owen,
I wonder if this was happening cause the TT connected with a different port each time, and the checks of host took the whole TT ID (including its client port) instead of just the hostname?
> tasks should not run on nodes where they were previously lost
> -------------------------------------------------------------
>
> Key: MAPREDUCE-78
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-78
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Reporter: Owen O'Malley
>
> We had a case where a task tracker was getting lost every 15 minutes and the same reduce kept getting scheduled on it (15 times). Tasks that are killed should not reschedule on the node where they were previously killed.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira