s2i-custom-notebook

Image Template to build custom notebook images for JupyterHub.

List of available custom images

s2i Thoth Images

Image	Status
s2i-custom py3.6
s2i-custom py3.8
s2i-custom elyra-py3.8
s2i-custom py3.8 Red Hat internal

Usage

Set up project repository

1. In your project repository, create a directory named notebooks. (The contents of this directory will be copied to the resulting image.)
2. Add your notebooks and any other required data to the notebooks directory.
3. Add the required Python dependencies for the notebooks in a Pipfile and also lock the dependencies in a Pipfile.lock. (All dependencies will be installed during image build time)
4. Choose the suitable base image based on the project requirement from the List of available custom images.
5. Build the image.

How to build the image

Locally

Prerequisites

• s2i: We assume the majority of users will be using linux and recommend using the the instructions here to install s2i if needed.
• Docker: If you are using Fedora please use these instructions to install docker. If you are on RHEL8 you can run into issues with docker, please follow the additional documentation that can be found here.

Build

You can build the image locally using s2i, some environment variables are required you can provide them via --environment-file= docs/env or explicit each using --env, by executing the following command in the root of your repository.

$ s2i build . quay.io/thoth-station/s2i-custom-notebook:latest MY-CUSTOM-IMAGE:latest

You can also build the image directly using your repository url.

s2i build my.repo.url.git quay.io/thoth-station/s2i-custom-notebook:latest MY-CUSTOM-IMAGE:latest

Be sure to replace MY-CUSTOM-IMAGE above with the name you intend to use for your image.

You might need to update your deployment of JupyterHub for the newly built/imported image to be available for use.

Running the image Locally

After building the image, you can run it locally using the following command:

$ docker run -p 8080:8080 MY-CUSTOM-IMAGE:latest start-singleuser.sh --ip="0.0.0.0" --port=8080

If you prefer to use podman, you can using the following command:

$ podman run -p 8080:8080 docker-daemon:MY-CUSTOM-IMAGE:latest start-singleuser.sh --ip="0.0.0.0" --port=8080

Since the s2i notebook builder relies on docker, if using podman, make sure to prepend your image name with docker-daemon: so that podman checks the correct local registry.

How to add the image to Open Data Hub/JupyterHub

• Once you have pushed your image to an image registry that is accessible by your JupyterHub instance, you can use the ImageStream template in this repository at openshift/oc-imagestream.yaml.

---
apiVersion: image.openshift.io/v1
kind: ImageStream
metadata:
  labels:
    opendatahub.io/notebook-image: "true" # <-- setting it true makes it visible to ODH/JupyterHub
    opendatahub.io/notebook-image-name: "My Notebook Image with Python 3.6"
  name: custom-notebook # <-- Update the ImageStream name here
spec:
  lookupPolicy:
    local: true
  tags:
  - from:
      kind: DockerImage
      name: quay.io/thoth-station/s2i-custom-notebook:latest # <-- Change to your container Image with tag
    importPolicy:
      scheduled: true
    name: "latest"

• After updating the template file with your image details, you can run the following command to create a ImageStream in your JupyterHub Namespace:

oc apply -f openshift/oc-imagestream.yaml -n MY_JUPYTERHuB_NAMESPACE

In OpenDataHub v0.5 or lower, the ImageStream name should have the following format: ``s2i-XXXX-notebook``

• Once the ImageStream has been created, you might need to restart the JupyterHub pod for it see the newly added ImageStream.

How to build the custom base image

Instead of using the pre-built version of the custom notebook, you can build the custom notebook from source code. we follow overlay based method in s2i-custom-notebook build. A tool Thamos is used for the installation of python stacks. Details about the tool can be found at Thamos Documentation

Example for building python3.6 based custom base image:

• Build python36 from the overlay/python36

podman build -t s2i-custom-notebook -f overlays/python36/Dockerfile .

• Build python38 from the overlay/python38

podman build -t s2i-custom-py38-notebook -f overlays/python38/Dockerfile .

• Build python38 from the overlay/python38-internal

podman build -t s2i-custom-py38-internal-notebook -f overlays/python38-internal/Dockerfile .

• Build elyra-custom from the overlay/elyra

podman build -t s2i-elyra-custom-notebook -f overlays/elyra/Dockerfile .