2
$\begingroup$

I download and installed CDH 5 package succesfully on a single linux node in pseudo-distributed Mode on my CentOS 6.5

Starting Hadoop and verifying it is Working Properly as in this link

I succesfully finished the following steps

Step 1: Format the NameNode.

Step 2: Start HDFS

Step 3: Create the /tmp Directory

Step 4: Create the MapReduce system directories:

Step 5: Verify the HDFS File Structure

Step 6: Start MapReduce

while following command in step 7 I get the following error.

Step 7: Create User Directories

$ sudo -u hdfs hadoop fs -mkdir -p /user/hadoopuser

mkdir: '/user/hadoopuser': No such file or directory

(where hadoopuser is my linux login username)

If I create the directory manually as /user/hadoopuser in the filesystem, it is not accepting.

How to success the step 7:?

Please provide the sloution to procced the remaining installation.

$\endgroup$
2
  • 2
    $\begingroup$ try sudo hadoop fs -ls -l / and see what comes back. Usually a permissions error comes back explicitly, but lets have a gander at what the root path looks like. I wonder if the file system was never formatted. $\endgroup$ Jul 3, 2014 at 17:39
  • 2
    $\begingroup$ This might be better at serverfault.com $\endgroup$
    – Sean Owen
    Jul 3, 2014 at 21:36

3 Answers 3

1
$\begingroup$

The simplest way to setup a user directory is to login to Hue as an admin and create a user account. This will create a directory in HDFS located at /user/yourusername with the correct permissions to work in.

$\endgroup$
0
$\begingroup$

Problem is resolved. I created the folders in terminal.

6b. hdfs dfs -mkdir /user hdfs dfs -mkdir /user/hduser

  1. Put this file to HDFS: hadoop fs -put file1.txt /user/hduser/file1.txt hadoop fs -put file.txt file.txt
$\endgroup$
0
$\begingroup$

You should create /user folder at first.

sudo -u hdfs hadoop fs -mkdir -p /user

$\endgroup$
1
  • $\begingroup$ The -p flag means that you don't have to create the parent directory first. $\endgroup$ Nov 9, 2015 at 22:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.