[
https://issues.apache.org/jira/browse/HADOOP-5471?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12682547#action_12682547
]
Amareshwari Sriramadasu commented on HADOOP-5471:
-------------------------------------------------
Test failure TestCapacityScheduler.testClusterBlockingForLackOfMemory is not
related to the patch.
> SyncLogs thread in Child.java would update wrong file for a cleanup attempt,
> in some cases.
> -------------------------------------------------------------------------------------------
>
> Key: HADOOP-5471
> URL: https://issues.apache.org/jira/browse/HADOOP-5471
> Project: Hadoop Core
> Issue Type: Bug
> Components: mapred
> Affects Versions: 0.19.1
> Reporter: Amareshwari Sriramadasu
> Assignee: Amareshwari Sriramadasu
> Fix For: 0.19.2
>
> Attachments: patch-5471-1.txt, patch-5471.txt
>
>
> This happens in the following scenario:
> If jvm is launched for a cleanup attempt and getTask has not returned yet.
> thus, isCleanup value is not obtained yet. But, the SyncLogs thread would do
> a syncLogs with wrong isCleanup value (i.e. with wrong index file).
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
