[
https://issues.apache.org/jira/browse/HDFS-13254?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16393992#comment-16393992
]
Weiwei Wu commented on HDFS-13254:
----------------------------------
[~ywskycn] Thank you for your detailed explanation. The two solutions you
provided can indeed avoid this problem.
[~elgoiri], I want the following scenario:
Use RBF as a file system for users and provide the same API interface as the
original HDFS.
For example, give the Spark user a directory /spark with mount points
/spark ==> 1->/spark
/spark/pathA ==> 2->/spark
/spark/pathB ==> 3->/spark
The user does not need to know the mount ns corresponding to each directory. It
only needs to change the access path in the original program to hdfs://ns-fed/
and the program can run.
There are several advantages to this:
1. When the underlying ns change, the user program does not need to be modified.
2. The user program does not need to be modified when the user program is
migrated to other rbf clusters.
> RBF: Cannot mv/cp file cross namespace
> --------------------------------------
>
> Key: HDFS-13254
> URL: https://issues.apache.org/jira/browse/HDFS-13254
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Weiwei Wu
> Priority: Major
>
> When I try to mv a file from a namespace to another, the client return an
> error.
>
> Do we have any plan to support cp/mv file cross namespace?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
