Skip to content

Latest commit

 

History

History
92 lines (67 loc) · 3.26 KB

DEPLOYMENT.md

File metadata and controls

92 lines (67 loc) · 3.26 KB
Raw

Deploy tljh_repo2docker as a hub-managed service

A guide to help you add tljh_repo2docker service to a JupyterHub deployment from scratch.

Note

In this guide, we assume you have experience with setting up JupyterHub on Kubernetes. We will use Zero to JupyterHub as the base JupyterHub deployment.

Create a new JupyterHub docker image.

Since we are going to use tljh_repo2docker as a hub-managed service, the python package needs to be installed into the base JupyterHub image. We will build a custom image based on the zero-to-jupyterhub image. Here is a minimal docker file to build the custom image:

FROM quay.io/jupyterhub/k8s-hub:3.3.7

USER root
RUN python3 -m pip install "tljh-repo2docker>=2.0.0a1"

USER jovyan

Then you can build and push the image to the registry of your choice:

docker build . -t k8s-hub-tljh:xxx

Installing JupyterHub and tljh_repo2docker with Helm chart.

With a Kubernetes cluster available and Helm installed, we can install our custom JupyterHub image in the Kubernetes cluster using the JupyterHub Helm chart. You can find a chart template at example/tljh_r2d.

Using local build-backend via repo2docker.

To build images using the local docker engine, modify the value.yaml file with the following content:

# values.yaml
jupyterhub:
  enabled: true
  hub:
    image:
      name: k8s-hub-tljh
      tag: xxx
    extraFiles:
      my_jupyterhub_config:
        mountPath: /usr/local/etc/jupyterhub/jupyterhub_config.d/jupyterhub_config.py

Now install the chart with the JupyterHub config from jupyterhub_config_local.py:

helm upgrade --install tljh . --wait --set-file jupyterhub.hub.extraFiles.my_jupyterhub_config.stringData=./jupyterhub_config_local.py

Using binderhub service as build-backend.

To use binderhub as the build-backend, you need to deploy binderhub service and config tljh-repo2docker to use this service. Here is an example of the configuration:

#values.yaml

jupyterhub:
  enabled: true
  hub:
    image:
      name: k8s-hub-binhderhub-tljh
      tag: xxx
    config:
      BinderSpawner:
        auth_enabled: true
    extraFiles:
      my_jupyterhub_config:
        mountPath: /usr/local/etc/jupyterhub/jupyterhub_config.d/jupyterhub_config.py

binderhub-service:
  config:
    BinderHub:
      base_url: /services/binder
      use_registry: true
      image_prefix: ''
      enable_api_only_mode: true

  buildPodsRegistryCredentials:
    server: 'https://index.docker.io/v1/' # Set image registry for pushing images

Now install the chart with the JupyterHub config from jupyterhub_config_binderhub.py and with image registry credentials passed via command line:

helm upgrade --install tljh . --wait --set-file jupyterhub.hub.extraFiles.my_jupyterhub_config.stringData=./jupyterhub_config_binderhub.py --set binderhub-service.buildPodsRegistryCredentials.password=xxx --set binderhub-service.buildPodsRegistryCredentials.username=xxx

Now you should have a working instance of JupyterHub with tljh_repo2docker service deployed.