kokon191 commented on code in PR #7075:
URL: https://github.com/apache/hadoop/pull/7075#discussion_r1811706106
##########
hadoop-hdfs-project/hadoop-hdfs-rbf/src/main/java/org/apache/hadoop/hdfs/server/federation/router/RouterClientProtocol.java:
##########
@@ -1959,8 +1960,45 @@ public BatchedEntries<OpenFileEntry> listOpenFiles(long
prevId)
public BatchedEntries<OpenFileEntry> listOpenFiles(long prevId,
EnumSet<OpenFilesIterator.OpenFilesType> openFilesTypes, String path)
throws IOException {
- rpcServer.checkOperation(NameNode.OperationCategory.READ, false);
- return null;
+ rpcServer.checkOperation(NameNode.OperationCategory.READ, true);
+ List<RemoteLocation> locations = rpcServer.getLocationsForPath(path,
false, false);
+ RemoteMethod method =
+ new RemoteMethod("listOpenFiles", new Class<?>[] {long.class,
EnumSet.class, String.class},
+ prevId, openFilesTypes, new RemoteParam());
+ Map<RemoteLocation, BatchedEntries> results =
+ rpcClient.invokeConcurrent(locations, method, true, false, -1,
BatchedEntries.class);
+
+ // Get the largest inodeIds for each namespace, and the smallest inodeId
of them
+ // then ignore all entries above this id to keep a consistent prevId for
the next listOpenFiles
+ long minOfMax = Long.MAX_VALUE;
+ for (BatchedEntries nsEntries : results.values()) {
Review Comment:
I went with this option because this approach doesn't require changing the
current protocol for `listOpenFiles` (id, types, path), otherwise client needs
to be aware of the last id for each namespace. I think router `getListing` uses
the same logic
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
