hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Uma Maheswara Rao G (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HDFS-3068) RemoteBlockReader2 fails when using SocksSocketFactory
Date Wed, 03 Apr 2013 05:03:16 GMT

    [ https://issues.apache.org/jira/browse/HDFS-3068?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13620613#comment-13620613
] 

Uma Maheswara Rao G commented on HDFS-3068:
-------------------------------------------

@Tom, How about fall back to old block reader if the underlaying socket input stream does
not have readabale channel? Because RemoteBlockReader2 needed channel based inputstreams to
deal with direct buffers. Also we can warn about this while falling back to old reader.
                
> RemoteBlockReader2 fails when using SocksSocketFactory 
> -------------------------------------------------------
>
>                 Key: HDFS-3068
>                 URL: https://issues.apache.org/jira/browse/HDFS-3068
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: hdfs-client
>    Affects Versions: 0.23.1
>            Reporter: Tom White
>
> When hadoop.rpc.socket.factory.class.default is set to org.apache.hadoop.net.SocksSocketFactory,
HDFS file reads fail with errors like
> {noformat}
> Socket Socket[addr=/10.12.185.132,port=50010,localport=55216] does not have an associated
Channel.
> {noformat}
> The workaround is to set dfs.client.use.legacy.blockreader=true to use the old implementation
of RemoteBlockReader. RemoteBlockReader should not be removed until this bug is fixed.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message