nmfs-opensci / container-images Goto Github PK

Dear Hub Champion,

If you are not using custom-built images for your JupyterHub, then 2i2c has already taken action to secure your hub. Please disregard the rest of this message.

We would like you to be aware of a potential vulnerability for JupyterHubs. If you are using a custom image with jupyter-server-proxy installed, then please take action to secure your hub. The affected versions are <=4.1.0 and <=3.2.2 and may be pulled in as a dependency of other packages.

Recommended actions

If your custom image is based on an upstream community image, then update your base image to the latest version
If your custom image is using pip, conda or similar, then you may need to explicitly pin all of your packages to versions compatible with patched versions of jupyter-server-proxy
Once you have updated and re-built your image, test that it is indeed using a patched version >=4.1.1, >=3.2.3 of jupyter-server-proxy.
See the security advisory on GitHub for full details and instructions on how to check for this vulnerability GHSA-w3vc-fx9p-wp4v.

Optional: A note on upgrading JupyterHub

You may also want to take this opportunity to upgrade your custom image to JupyterHub version >=4.1.0 to address a separate JupyterHub vulnerability GHSA-7r3h-4ph8-w38g. You may experience XSRF and 403 bugs in JupyterHub versions 4.1.0 – 4.1.4, therefore we recommend

upgrading JupyterHub to >=4.1.5
upgrading nbgitpuller to >=1.2.1 (if using)
upgrading jupyterhub-singleuser to the latest version (if using conda/mamba)
You are receiving this email because you are noted as a 'technical contact' for your community. If you do not wish to receive such emails or there is someone else in your organization who should be receiving this kind of email, please let me know at [email protected]. Thank you!

Best wishes,
Jenny Wong
2i2c

https://stackoverflow.com/questions/31976367/how-do-i-delete-refresh-available-kernels-for-ipython-jupyter-notebook-v4-0

Major hack to only show the snakemake and qiime2 kernels
jupyterhub/jupyterhub#2759

customize the launcher https://jupyter-app-launcher.readthedocs.io/en/latest/

• NESDIC training courses https://www.star.nesdis.noaa.gov/atmospheric-composition-training/software_jupyter_notebook_set-up.php
• https://www.star.nesdis.noaa.gov/atmospheric-composition-training/training_AMS_Short_Course_2023.php
• https://www.star.nesdis.noaa.gov/atmospheric-composition-training/training_AMS_Short_Course_2022.php
• https://www.star.nesdis.noaa.gov/atmospheric-composition-training/training_GSFC_ASTG_2023.php#trainingAgenda

https://www.nesdis.noaa.gov/events/ams-course-making-beautiful-images-of-noaa-satellite-data-using-python
[email protected]

error

git command not found - please ensure you have Git > 2 installed

https://towardsdatascience.com/create-your-own-gpu-accelerated-yupyter-notebook-server-with-google-colab-using-docker-2fa14900bab5

Codespace Idea

Just an idea to possibly help streamline codespace environment generation and setup

Link: https://github.com/MichaelAkridge-NOAA/codespace-generator

• This will dynamically generate basic codespace development container configurations (devcontainer.json) for each container image listed in a GitHub organization(nmfs-opensci)'s package registry.
• Proof of Concept. Still needs to be updated, expanded and configured to enhance codespace generation and to make it work with each image type/base

• https://2i2c.org/blog/2024/jupyterhub-binderhub-gesis/

Re our repo2docker convo for building dockerfiles for jupyterhub. See how the CryoCloud 2i2c hub does this.
GitHub action to build: https://github.com/CryoInTheCloud/hub-image/blob/main/.github/workflows/build.yaml
Build files: https://github.com/CryoInTheCloud/hub-image/tree/main
There are GitHub Actions for other tasks too like making the binder badge.

Goal images that open reasonably well in

• codespaces
• jupyterhubs of various sorts
• docker run on laptop and VM
• podman run
• binder

Have following features

• rstudio
• jupyterlab and notebook
• vscode
• easy enough way to open up those IDE
• IDEs work as expected/familiar
• terminal bash.
• terminal preset up with conda (no conda init needed)
• rstudio python env mimics whatever is default on image
• [ ]

Can layer. might be better
https://forum.qiime2.org/t/docker-image-with-qiime-and-jupyter/7206/2