[ 
https://issues.apache.org/jira/browse/HDFS-16487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17498405#comment-17498405
 ] 

Ayush Saxena commented on HDFS-16487:
-------------------------------------

Couple of points here.
 * With HDFS-15554 in, We have a configurable option which will not allow you 
to have dangling mount points. So, if that is a concern the config can be 
turned on.
* Second, we have HDFS-13506 in progress, If we get that in, then whenever you 
call for AddMount entry, it will create the target destination. Not sure why it 
is on hold, can try to push that.
* Thirdly, I think a lot of such ideas are similar to ViewFs, If in case of 
ViewFs the mount entries target isn't available, we tend to do a listing on the 
parent. In think That InternalDirOfViewFs will also list it.
* One more,  say you have a mount entry /dir/subdir -> some target and you 
don't have any /dir mounted and no default namespace. Then also on listing /dir 
will get listed, and say if you have default namespace configured, then also 
/dir will get listed despite /dir not available on the default NS.

The most I feel, we can get HDFS-13506 in, else if this is creating a problem, 
then HDFS-15554 can solve it.

> RBF: getListing uses raw mount table points
> -------------------------------------------
>
>                 Key: HDFS-16487
>                 URL: https://issues.apache.org/jira/browse/HDFS-16487
>             Project: Hadoop HDFS
>          Issue Type: Bug
>            Reporter: Fengnan Li
>            Assignee: Aihua Xu
>            Priority: Major
>
> In getListing, the result is a union of subclusters results and mount points. 
> However these two are of different concepts and the latter one is something 
> Router internal. It is very possible that the actual path doesn't exist in 
> the dest HDFS yet. 
> Can we choose a different strategy that check each children mount point and 
> confirm there is the HDFS path in the dest cluster? If so, we can add it; 
> otherwise we should skip this mount because it confuses clients. (Clients 
> could directly create a subdir under a dangling mount point)



--
This message was sent by Atlassian Jira
(v8.20.1#820001)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org

Reply via email to