I have been using using Jupyter server with CA 11.1.2, two issues (maybe you can say it is same one) troubled my setup:
1. ca_jupyter_server-instance:latest docker container exit automatically by itself;
2. when users trying to connect jupyter server kernel again using notebook, it will try to start new container (due to auto-exit) and it will take long time to start/connect, and in most situations it will fail;

Anyone else experienced these? I have Jupyter server installed on Linux CentOS 7.  

Is there anyway to make the "ca_jupyter_server" container stays up without exiting just like JupyterHub and JupyterViewer?  In addition, the connecting to Jupyter server seems quite slow in compare with connnecting with local jupyter server, any tuning needed?

------------------------------
Kevin Wang
------------------------------

#CognosAnalyticswithWatson

Hi Kevin, sorry for the late reply.

For 1), there is a service that will automatically shut down user containers that have been idle for over an hour. Unfortunately, this is not configurable right now, but there shouldn't be any consequence to the removal. Opening a new or existing notebook will re-create the container.

For 2), it looks like you're trying to restart the kernel without first closing the notebook. For the server to come back up, you would need to close the notebook and open a new or existing notebook. 

As far as I know, no tuning is needed. The latencies are due to network speed and redirecting requests from the browser -> Cognos -> Jupyter host -> Docker. 

Feel free to post any error logs if connection problems persist. 

Regards,
Rob

------------------------------
ROBERT MCAULEY
------------------------------