What our organization did is we have base project that we created that we get everyone the pull from that has everything set up. It is really not ideal but it was the best we could come up with given our control over the product. You are correct that the .jupyter dir does not exist until jupyter-notebooks is run and we could not stick it in any directory to get copied on the docker build. We actually had upgraded over the weekend for other reasons and i tested out the auto kill for jupyter. If you are a CLOUDERA rep or a rep reads this please understand that the main reason our organization chose CDSW is because of this functionality and the ability to edit the docker build. We have been through Anaconda Enterprise, IBM Watson studio, and even tried to run JupyterHub. Our organization has 150+ data scientists/ analysts and they are all irresponsible when it comes to stopping their sessions. Anaconda was by far the worst performing product/company to work with for support. Now with CDSW we can propagate project configuration/ tutorial scripts/ spark and hive config from the top down to all projects via the docker build and we love it. Below are the contents of jupyter-notebook-config.py located in .juypter, it worked for us. , c.NotebookApp.shutdown_no_activity_timeout = 3600 c.MappingKernelManager.cull_idle_timeout = 2600 ... View more