Running jupyter notebook remotely in a docker swarm cluster
summary
Claims a concrete fix: run a full browser plus Jupyter inside a Docker Swarm container and proxy access via an HTTP gateway so long-running cells keep output when the user disconnects. Implements images with Firefox, a VNC server, conda environments, a proxy network and registry, and uses Docker Swarm and Apache Guacamole to expose a browser inside container. It works with small UI lag and awkward copy‑paste but preserves kernels and enables screen‑share collaboration.
AI-written summary · about
chapters (12)
- 0:00Remote Jupyter in Swarm
- 1:25Why workaround needed
- 2:52Jupyter Notebook architecture
- 8:20Cluster requirements
- 13:26Swarm setup overview
- 15:14Docker basics
- 17:22Networks and services in Swarm
- 18:33Custom Jupyter container
- 21:46Remote desktop demo
- 24:15Workaround pros and cons
- 26:49Future improvements
- 31:48Q&A
related talks
36mHow to Roll your own Dockerized Jupyter Environment
01h 36mDeploying Jupyter Notebooks for students and researchers
22mDock Your Jupyter Notebook
01h 11mLearning to use Docker for development
33mIntegrating Jupyter Notebooks into your Infrastructure
29mLaunch Jupyter to the Cloud: an example of using Docker and Terraform
31mBuilding an interactive training environment using JupyterHub
37mHigher order infrastructure: from Docker basics to cluster management