After some discussion about how best to manage multiple applications on a cluster it was decided that in addition to using application context classpaths configured per namespace that we would separate the iterator jar files into their own HDFS directories. The convention is /accumulo/classpath/APP_NAME