Hello All, We have applications in our data organization like Oracle Data Integrator and Jupyter Hub. Those set of applications require up to date jars, core-site.xml, hdfs-site xml etc to integrate properly with the cluster. Our Application Architects suggests we should install those applications' clients and/or agents on edge nodes of our cluster. I don't want to do this because: *I might need to deal with conflicting environment requirements. *I will have limited control over server resource consumption *Application level operations such as upgrades might have adverse affects on the cluster I am considering to offer mounting jars/conf directories of edge nodes on read only mode to application servers. Any suggestions and evaluations about both architectures or alternatives are more than welcome. Best regards ... View more