To use Pan or Kitchen on a Hadoop cluster, you must configure Pentaho to run transformations and jobs with either the PDI client or the Pentaho Server. However, these configurations are not needed if your PDI client is connected to the Pentaho Repository. To use Pan and Kitchen from a repository directly on the Pentaho Server, you must create the named cluster definition in the server's repository. See Connecting to a Hadoop cluster with the PDI client for information on creating that connection.
Note: If a user starts the PDI client and the Pentaho Server on the same platform, the
cluster configuration files in the /home/<user>/.pentaho/metastore
directory are overwritten. To avoid this issue, use the same cluster connection names on
both the PDI client
host and the Pentaho Server
host.