In order to use HDFS for input or output, 2 additional steps are required: - Copy the new core-site.xml to "/usr/lib/hadoop/conf" - eg "cp /mnt/opt/hadoop-infinite/mapreduce/hadoop/core-site.xml /usr/lib/hadoop/conf/"
- Ensure a tomcat user directory exists, eg: "runuser - hdfs -c 'hadoop fs -mkdir /user/tomcat && hadoop fs -chown tomcat:hdfs /user/tomcat;'"
- (This will get created the first time a job runs - but you may need to create it first to start copying input data in - note this is only needed for jobs that read in external HDFS data - most jobs read from the Infinit.e datastore)
|