[
https://issues.apache.org/jira/browse/HDFS-16011?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17341707#comment-17341707
]
zhu commented on HDFS-16011:
----------------------------
[~ayushtkn] Thanks for your comment.
The ViewFS that supports nested mounting will match the mount point recently
according to the path, if it can be matched, it will return to
targetFileSystem, if it can’t match, it will return to InternalDirFs.
*For example, the following:*
*mount point*
/a/b -> /a/b
/a/b/c/d -> /a/b/c/d
*resolve：*
/a/b (/a/b targetFileSystem)
/a/b/c (/a/b targetFileSystem)
/a (/a InternalDirFs)
I think there is no need to change the processing logic of getListing() and
getContensummary(). This ViewFs nested installation has been used in our
internal version for some time, and these two methods have also been verified.
Indeed, renaming needs improvement. For target file systems with the same
permissions, rename should be used. I'll add test cases if it works.
> HDFS: Support viewfs nested mount
> ---------------------------------
>
> Key: HDFS-16011
> URL: https://issues.apache.org/jira/browse/HDFS-16011
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs, viewfs
> Affects Versions: 3.2.2, 3.3.1
> Reporter: zhu
> Assignee: zhu
> Priority: Major
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> The current viewfs does not support nested mount points. The current viewfs
> does not support nested mount points.
> *E.g:*
> 1./home/ => /home/
> 2./home/work => /home/work
> If mount point 1 is loaded, mount point 2 cannot be added, and the following
> exception will be thrown when loading 2.
> {code:java}
> throw new FileAlreadyExistsException("Path " + nextInode.fullPath +
> " already exists as link");
> {code}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
