[
https://issues.apache.org/jira/browse/HADOOP-19572?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17953174#comment-17953174
]
ASF GitHub Bot commented on HADOOP-19572:
-----------------------------------------
anujmodi2021 commented on code in PR #7698:
URL: https://github.com/apache/hadoop/pull/7698#discussion_r2100427030
##########
hadoop-tools/hadoop-azure/src/main/java/org/apache/hadoop/fs/azurebfs/services/AbfsBlobClient.java:
##########
@@ -400,37 +398,46 @@ public ListResponseData listPath(final String
relativePath, final boolean recurs
listResponseData.setOp(retryListOp);
}
}
+ return listResponseData;
+ }
- if (isEmptyListResults(listResponseData) && is404CheckRequired) {
+ /**
+ * Post-processing of the list operation on Blob endpoint.
+ * There are two client handing to be done on list output.
+ * 1. Empty List returned on server could potentially mean path is a file.
+ * 2. There can be duplicates returned from the server for explicit
non-empty directory.
+ * @param relativePath relative path to be listed.
+ * @param fileStatuses list of file statuses returned from the server.
+ * @param tracingContext tracing context to trace server calls.
+ * @param uri URI to be used for path conversion.
+ * @return rectified list of file statuses.
+ * @throws AzureBlobFileSystemException if any failure occurs.
+ */
+ @Override
+ public List<FileStatus> postListProcessing(String relativePath,
List<FileStatus> fileStatuses,
+ TracingContext tracingContext, URI uri) throws
AzureBlobFileSystemException {
+ List<FileStatus> rectifiedFileStatuses = new ArrayList<>();
+ if (fileStatuses.isEmpty() && !ROOT_PATH.equals(relativePath)) {
// If the list operation returns no paths, we need to check if the path
is a file.
// If it is a file, we need to return the file in the list.
+ // If it is a directory or root path, we need to return an empty list.
// If it is a non-existing path, we need to throw a
FileNotFoundException.
- if (relativePath.equals(ROOT_PATH)) {
- // Root Always exists as directory. It can be an empty listing.
- return listResponseData;
- }
AbfsRestOperation pathStatus = this.getPathStatus(relativePath,
tracingContext, null, false);
BlobListResultSchema listResultSchema =
getListResultSchemaFromPathStatus(relativePath, pathStatus);
- LOG.debug("ListBlob attempted on a file path. Returning file status.");
- List<VersionedFileStatus> fileStatusList = new ArrayList<>();
+ LOG.debug("ListStatus attempted on a file path {}. Returning file
status.", relativePath);
Review Comment:
This is fine as originally ListStatus was attempted on file path.
> ABFS: [FnsOverBlob] Empty Page Issue on Subsequent ListBlob call
> ----------------------------------------------------------------
>
> Key: HADOOP-19572
> URL: https://issues.apache.org/jira/browse/HADOOP-19572
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs/azure
> Affects Versions: 3.5.0, 3.4.1
> Reporter: Anuj Modi
> Priority: Major
> Labels: pull-request-available
>
> We came across a new behavior from server where ListBlob call can return
> empty list even after returning a next marker(continuation token) from
> previous list call.
> This is to handle that case and do not infer listing to be incomplete.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
