Container images for use with Coder
We repackage several of the Jupyter Docker Stacks images with necessary changes to work with Coder.
This repo is configured to build all of the container images specified in the GitHub Actions .github/workflows/build.yml file. These builds can be viewed at Build Jupyter Coder Images. The resulting container images should be linked on the repo home page in the packages section to the right.
Builds can be triggered manually on the repo's actions tab, via pushing an update to the repo under the images/jupyter directory or an update to the workflow yaml file itself.
The jupyter tag should be given a value of latest or a build date like 2025-11-10 on the respective images from quay.io/organization/jupyter.
Either build trigger will default to latest if unset.
- Manually triggered builds will prompt for a jupyter_tag value
- Automatic builds will use the value of the JUPYTER_TAG variable specified in the repo Settings → Secrets and variables → Actions → Variables
Note: There are some images which have a specific cuda version which may need to have the
cuda12portion of their tag updated directly in the action yaml file. I.E. the PyTorch notebook.
Some images may need to be built locally due to container image size limitations in GitHub Actions. As of writing, this limitation is currently 10GB.
In order to build the Jupyter Docker Stack-based images locally, we need to specify the build target of notebook-image, i.e.:
docker build . -f Dockerfile -t [your-dockerhub-username]/[image-name]:v[x.x] --target notebook-imageOnce the container has been built, it should be tested on a dev instance of coder prior to being retagged as an official image.
Execute the following to retag a locally built image, replacing the notebook name:
docker image tag [container-id] ghcr.io/sdsu-research-ci/coder-images/[notebook-name]:[jupyter_tag]- I.E.
docker image tag ea984ac26018 ghcr.io/sdsu-research-ci/coder-images/minimal-notebook:2025-11-10