I have installed splunk today on both a windows 7 64 server and a centos linux 32 bit machine. I have also installed installed version 122 of hadoop connect app from splunk as a tgz file.
I want to connect to a CDH5 Centos linux based hadoop cluster built with the CDH5 manager. I have set my hdfs uri, my java home and hadoop home as well as the name node htt port. When I click save on both windows and linux I get the error
Unable to connect to Hadoop cluster 'hdfs://hc2nn:8020/' with principal 'None': Invalid HADOOP_HOME. Cannot find Hadoop command under bin directory HADOOP_HOME='/opt/cloudera/parcels/CDH-5.1.3-1.cdh5.1.3.p0.12'.
I can see the hadoop command within the $HADOOP_HOME/bin directory and I have checked that the command works i.e. I can connect to hdfs and do a listing. I wondered whether anyone had seen this error before ?
I was running hunk from a linux host that didnt have hdfs access to the remote cluster. Rather than try and sort that access out I installed hunk on the cluster and it runs ok.
HADOOP_HOME should be set to a directory on the local machine where splunk is installed.
Seems like you are setting hadoop home to a directory on remote cluster where hadoop is installed?
HADOOP_HOME='/opt/cloudera/parcels/CDH-5.1.3-1.cdh5.1.3.p0.12'
Does the user Splunk is running as have permissions to access the $HADOOP_HOME/bin directory?