[
https://issues.apache.org/jira/browse/MAPREDUCE-2928?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13112973#comment-13112973
]
Eli Collins commented on MAPREDUCE-2928:
----------------------------------------
The motivation for tracking numFailures is that the TT re-init code just wants
to know if there have been any failures since the last time it checked
directories, this can easily be determined by seeing if the total number of
failures has changed. The isDiskFailed method is stateful, eg you have to know
whether you've called it previously (which is why TT#initialize calls it and
ignores the return value), ie it tells you if there have been any failures
since the last time *isDiskFailed* was called, not if there have been any
failures since the disks were checked. Make sense? It be might more clear if
checkDirs returned the failure count but it throws DiskErrors itself so that
doesn't work.
> MR-2413 improvements
> --------------------
>
> Key: MAPREDUCE-2928
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2928
> Project: Hadoop Map/Reduce
> Issue Type: Sub-task
> Components: tasktracker
> Reporter: Eli Collins
> Assignee: Eli Collins
> Fix For: 0.20.205.0
>
> Attachments: mapreduce-2928-1.patch, mapreduce-2928-2.patch
>
>
> Tracks improvements to MR-2413. See [this
> comment|https://issues.apache.org/jira/browse/MAPREDUCE-2413?focusedCommentId=13095073&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13095073].
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
