Cannot Display HDFS Subdirectories when Retrieving Schema from a Metadata Repository HDFS Connection

Talend Version        6.3.1


Additional Versions  
Product Big Data
Component Studio
Problem Description

A Hadoop cluster is kerberized. In the cluster, there is a directory, for example /user/talendhdp/test that belongs to the user talendhdp and group hdfs.


The permission rights to the test directory are set with no other group permission rights:

rwxrwx--- talendhdp hdfs test (hdfs dfs -chmod 770 /user/talendhdp/test


After creating a metadata Hadoop cluster in a Talend Studio 6.3.1 repository, the problem is that when retrieving a schema with the related metadata HDFS connection from Talend Studio, the /user/talendhdp/test directory is not displayed.


Using Talend Studio 6.4.1, this problem does not occur.

Problem root cause

The issue is that the username is anonymous for the metadata HDFS connection in Talend Studio, and it cannot be changed adequately.

Solution or Workaround

The solution consists of updating the HDFS configuration item file corresponding to the HDFS connection created from the Hadoop Cluster connection automatically. In this example, the connection name is QA208_CDH58_HDFS.


Edit the talend_studio_workspace\LOCAL_PROJECT\metadata\hadoop\hdfs\QA208_CDH58_HDFS_0.1.item file to add a userName="" attribute after the relativeHadoopClusterId attribute.


For example:

<?xml version="1.0" encoding="UTF-8"?>
<TalendHDFS:HDFSConnection xmi:version="2.0" xmlns:xmi=""
xmi:id="_NT-pwJlEEeeY3O8FApgGZg" name="QA208_CDH58_HDFS" label="QA208_CDH58_HDFS" 
userName="" rowSeparator="&quot;\n&quot;" fieldSeparator="&quot;;&quot;"/>
JIRA ticket number  
Version history
Revision #:
19 of 19
Last update:
‎02-24-2019 11:17 PM
Updated by:
Labels (3)
Tags (1)