hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hairong Kuang (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-5554) DataNodeCluster should create blocks with the same generation stamp as the blocks created in CreateEditsLog
Date Mon, 23 Mar 2009 23:33:50 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-5554?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12688458#action_12688458
] 

Hairong Kuang commented on HADOOP-5554:
---------------------------------------

I have tested the patch manually.

> DataNodeCluster should create blocks with the same generation stamp as the blocks created
in CreateEditsLog
> -----------------------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-5554
>                 URL: https://issues.apache.org/jira/browse/HADOOP-5554
>             Project: Hadoop Core
>          Issue Type: Bug
>          Components: test
>    Affects Versions: 0.19.2
>            Reporter: Hairong Kuang
>            Assignee: Hairong Kuang
>             Fix For: 0.19.2, 0.20.0, 0.21.0
>
>         Attachments: blockGS.patch
>
>
> HADOOP-5384 makes DataNodeCluster to create blocks with generation stamp Block#GRANDFATHER_GENERATION_STAMP(0)
so simuated datanodes do not crash NameNode any more. But there is still a problem. CreateEditLogs
creates blocks with generation stamp GenerationStamp#FIRST_VALID_STAMP (1000). Because of
the generation stamp mismatch, all injected blocks are marked as invalid when NameNode processes
block reports.

-- 
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