I'm trying to run wordcount using Pentaho Mapreduce step. I have integrated Pentaho on my local windows machine with Cloudera Quickstart VM. I have copied the conf files from VM to cdh58 plugin and made necessary changes in conf files by replacing all url's pointing quickstart.cloudera to ip address of the VM.
Test cluster result was successful, however sometime zookeeper and oozie test fails

I was able to successfully load data into hdfs. I'm having issues with the PMR job. PMR step fails immediately without even loading pdi jar to the hdfs during the first run.

However the odd part of the error is PMR is pointing ResourceManager to localhost:8032 where actual resource manager is quickstartVM_IP:8032 (eg: I don't understand why PMR is pointing to local host when cluster in PDI is configured with any IP.
Also test results for Hadoop cluster are also very inconsistent which shows me as success in some cases and failure in some cases with no change in configs

Can some please help me out on this issue.