Hadoop: Error in configuring object Hadoop: Error in configuring object hadoop hadoop

Hadoop: Error in configuring object


I got this to work as follows:

I'm running in local mode from my hadoop base directory, hadoop-1.0.0 with an input subdirectory under it, and I get the same error you do.

I edited the failing java file to get it to log the path instead of the filename, rebuilt it ("ant binary"), and reran it. It was looking for the file in the directory I was running from. I have no idea if it was looking in the hadoop base dir or the execution dir.

...so I made a symbolic link in the directory I run terasort in pointing to the real file in the input directory.

It's a cheap hack, but it works.

- Tim.


The problem was occurring because i was deploying the job on an NFS. I changed the hadoop.tmp.dir to point it to a local file system(/tmp) and the problem disappeared in a jiffy.


Have you setup to run in pseudo distributed mode (or a real cluster)? Unless you configure Hadoop, it will run in local job runner mode (as libs inside a single process) - Terasort does NOT work in LocalJobRunner mode. Look for the word LocalJobRunner in the output to check.

Here is a link to setup HDFS, SSH and rsync: http://hadoop.apache.org/docs/r1.1.1/single_node_setup.html#PseudoDistributed