[
https://issues.apache.org/jira/browse/HDFS-16791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17616160#comment-17616160
]
ASF GitHub Bot commented on HDFS-16791:
---------------------------------------
mccormickt12 commented on code in PR #4967:
URL: https://github.com/apache/hadoop/pull/4967#discussion_r992950901
##########
hadoop-hdfs-project/hadoop-hdfs-client/src/main/proto/ClientNamenodeProtocol.proto:
##########
@@ -428,6 +428,14 @@ message GetPreferredBlockSizeResponseProto {
message GetSlowDatanodeReportRequestProto {
}
+message GetEnclosingRootRequestProto {
+ required string filename = 1;
Review Comment:
sounds good. Ill update these to `optional` since it is best practice for
proto going forward and hadoop doesn't seem to be super opinionated about this.
> New API for enclosing root path for a file
> ------------------------------------------
>
> Key: HDFS-16791
> URL: https://issues.apache.org/jira/browse/HDFS-16791
> Project: Hadoop HDFS
> Issue Type: Improvement
> Reporter: Tom McCormick
> Assignee: Tom McCormick
> Priority: Major
> Labels: pull-request-available
>
> At LinkedIn we run many HDFS volumes that are federated by either
> ViewFilesystem or Router Based Federation. As our number of hdfs volumes
> grows, we have a growing need to migrate data seemlessly across volumes.
> Many frameworks have a notion of staging or temp directories, but those
> directories often live in random locations. We want an API getEnclosingRoot,
> which provides the root path a file or dataset.
> In ViewFilesystem / Router Based Federation, the enclosingRoot will be the
> mount point.
> We will also take into account other restrictions for renames like
> encryptions zones.
> If there are several paths (a mount point and an encryption zone), we will
> return the longer path
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
