hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Suresh Srinivas (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-1813) Hdfs Federation: Authentication using BlockToken in RPC to datanode fails.
Date Thu, 07 Apr 2011 17:36:05 GMT

    [ https://issues.apache.org/jira/browse/HDFS-1813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13017061#comment-13017061
] 

Suresh Srinivas commented on HDFS-1813:
---------------------------------------

+1 patch looks good. Now we are consistently setting block access token for the last block
reported in LocatedBlock. This may not be required for the completed file, however from consistency
point of view, your change makes sense.

Comments:
# Please add/modify existing unit test for checking token is set for last block.
# Please add/modify unit test to check for READ access.

> Hdfs Federation: Authentication using BlockToken in RPC to datanode fails.
> --------------------------------------------------------------------------
>
>                 Key: HDFS-1813
>                 URL: https://issues.apache.org/jira/browse/HDFS-1813
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: Jitendra Nath Pandey
>            Assignee: Jitendra Nath Pandey
>             Fix For: Federation Branch
>
>         Attachments: HDFS-1813.1.patch
>
>
>  Several issues are causing this problem
> 1. The last LocatedBlock returned by getBlockLocations doesn't have BlockToken.
> 2. The blockPoolTokenSecretManager is not initialized before created rpc server in datanode.
> 3. The getReplicaVisibleLength API in datanode expects WRITE permission in the block
token, but the block tokens are generated with read permission only in getBlockLocations at
namenode.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message