xnat/jupyterhub

The xnat/jupyter image contains a preconfigured JupyterHub for running with XNAT.

To connect XNAT with JupyterHub, a jupyterhub_config.py file has been created that

1. Authenticates a user with XNAT
2. Configures the JupyterHub Docker SwarmSpawner to mount XNAT data

This repo contains a Dockerfile, docker-compose.yml, and docker-stack.yml file for the xnat/jupyterhub image. The xnat-jupyter-plugin must be installed in your XNAT. The xnat/jupyterhub image must also be running on a manager node within a Docker Swarm in order to spawn Jupyter servers as Docker containers. Note that there is one xnat/jupyterhub instance per XNAT.

See the XNAT JupyterHub Plugin Wiki for the latest documentation on this plugin.

Running the xnat/jupyterhub Image

These are the steps for deploying the xnat/jupyterhub image alongside an existing XNAT. For convenience, the xnat-docker-compose repository has a branch with JupyterHub added as a service alongside XNAT.

1. Clone repo into a directory.

   cd /opt
   git clone https://bitbucket.org/xnat-containers/xnat-jupyterhub.git jupyterhub

2. Configure the docker-stack.yml environmental variables.

   It is critical to correctly configure the UID and GID of JupyterHub (JH) and the single-user Jupyter notebook (NB) containers. In general, the Tomcat/XNAT UID is used as the UID for JH and the NB containers. The Tomcat/XNAT GID is used as the GID for the NB containers only. The GID for JH is the group id of the docker group.

   Get the id of the tomcat/xnat user

   $ id tomcat
    uid=54(tomcat) gid=54(tomcat) groups=54(tomcat),992(docker)

   You can also find the gid of the docker socket with

   $ cat /etc/group | grep docker
   docker:x:992:tomcat

   Tomcat, JH and the NB containers will share this UID (54). Tomcat and the NB containers will share the same GID (54) while JH will be a member of the docker group (992).

   If you have a domain name for this server we will also set the JH_XNAT_URL environmental variable. This environmental variable is used by JupyterHub to communicate with your XNAT.

   # docker-stack.yml
   
   # ...
   
   environment:
     JH_NETWORK: *networkname
     JH_UID: 54
     JH_GID: 992
     NB_UID: 54
     NB_GID: 54
     JH_START_TIMEOUT: 180
     JH_XNAT_URL: http://172.17.0.1 OR https://your.xnat.org
     JH_XNAT_SERVICE_TOKEN: # create a service token for XNAT to talk to JH
     JH_XNAT_USERNAME: jupyterhub
     JH_XNAT_PASSWORD: # create a password for JH to talk to XNAT

   You will also need to create a service token for XNAT to use with JupyterHub and create a password for JupyterHub to use with XNAT. Supplies these to JupyterHub here. Save these for later, you will need them to configure the JupyterHub plugin in the XNAT UI.

3. Initiate a Docker swarm if you are not already on one. Otherwise, make sure you are on a master node.

   docker swarm init

4. Start JupyterHub on a master node:

   docker stack deploy -c docker-stack.yml jupyterhub

   JupyterHub should be available at http://localhost:8000/jupyterhub (or your domain name). See notes below for configuring your reverse proxy if you can't find JupyterHub.

5. Pull single-user container images:

   docker image pull xnat/datascience-notebook:latest

6. Login to XNAT and configure the JupyterHub plugin. See the documentation for all the details on how to configure the plugin. You will need the service token and password you created in step 2. In XNAT the JupyterHub service account needs to be enabled and the password set, and the JupyterHub URL, service token, and path translation need to be set.

7. To view and inspect JupyterHub and a running single-user notebook container:

   docker service ls
   docker service logs 9hzw7gu0mo79
   docker service inspect 9hzw7gu0mo79     

8. Stop JupyterHub

   docker stack rm jupyterhub

Using a reverse proxy

JupyterHub and XNAT should live behind the same reverse proxy with XNAT at / and JupyterHub at /jupyterhub. The xnat-docker-compose repo has an example of having XNAT and JupyterHub behind the same reverse proxy. This is based on the JupyterHub documentation.

Below is an example of configuring haproxy:

# haproxy.cfg example

# ...
# ...
# ...

frontend appserver
        bind 0.0.0.0:80
        bind 0.0.0.0:443 ssl crt /etc/ssl/certs/devcert.pem
        http-response replace-value Location ^http://(.*)$ https://\1
        use_backend jupyterhub if { path_beg /jupyterhub } || { path_beg /jupyterhub/ }
        use_backend web_servers if { path_beg / }
        default_backend web_servers

backend web_servers
        mode http
        balance roundrobin
        option forwardfor
        http-request set-header X-Forwarded-Port %[dst_port]
        http-response set-header location %[res.hdr(location),regsub(http://,https://)] if { status 301 302 }
        server web01 127.0.0.1:8080

backend jupyterhub
        mode http
        balance roundrobin
        option forwardfor
        http-request set-header X-Forwarded-Port %[dst_port]
        http-response set-header location %[res.hdr(location),regsub(http://,https://)] if { status 301 302 }
        server jhub01 127.0.0.1:8000

Arguments and Environmental Variables

Here's a summary of the arguments and environmental variables used in the JupyterHub dockerfile.

Variable	Description / Comments
JH_NETWORK	JupyterHub and the single-user Jupyter containers must run on the same Docker network. Used in jupyterhub_config.py.
JH_XNAT_URL	The URL of the XNAT to connect to. Required by XnatAuthenticator and xnat_pre_spawn_hook.
JH_XNAT_SERVICE_TOKEN	XNAT needs a service account within JupyterHub inorder to spawn servers. This is the token used by XNAT for the service account. Required by xnat_pre_spawn_hook.
JH_XNAT_USERNAME	JupyterHub needs an account with XNAT to retrieve the user options during the spawning process. Required by xnat_pre_spawn_hook.
JH_XNAT_PASSWORD	The password for JupyterHub's account on XNAT. Required by xnat_pre_spawn_hook.
JH_START_TIMEOUT	The amount of time JupyterHub should wait (in seconds) before deciding the single user container failed to start
JH_UID	The UID to run JupyterHub with.
JH_GID	The GID to run JupyterHub with. This group should have access to the Docker socket.
NB_UID	The UID to run the single-user Jupyter containers with. This should match the UID of the XNAT archive.
NB_GID	The GID to run the single-user Jupyter containers with. This should match the GID of the XNAT archive.
JH_TLS_CLIENT_CERT	The path to the client certificate for JupyterHub to use when connecting to XNAT.
JH_TLS_CLIENT_KEY	The path to the client key for JupyterHub to use when connecting to XNAT.
JH_TLS_CA_CERT	The path to the CA certificate for JupyterHub to use when connecting to XNAT.
JH_TLS_VERIFY	Whether or not JupyterHub should verify the XNAT certificate.

Running on Kubernetes

The 'xnat-jupyterhub-chart' directory contains a Helm chart which deploys JupyterHub for an XNAT instance on Kubernetes. The chart is based on the Zero to JupyterHub chart. The chart contains a values.yaml file which contains the default values needed to deploy JupyterHub for an XNAT. A postgres database is also deployed for JupyterHub. This chart presumes that the XNAT instance is deployed in the same Kubernetes namespace as JupyterHub. This chart, andrewl4/xnat-skaffold, was used to deploy XNAT when developing this chart. The PV and PVC for the user workspaces are created by the XNAT deployment. Be sure the JupyterHub plugin is installed in the XNAT instance.

The chart can be deployed with the following command:

helm repo add bitnami https://charts.bitnami.com/bitnami
helm repo add jupyterhub https://jupyterhub.github.io/helm-chart
helm repo update
helm upgrade --install jupyterhub xnat-jupyterhub-chart/ -n xnat --create-namespace --values xnat-jupyterhub-chart/values.yaml
helm uninstall jupyterhub -n xnat

You may need to update the values.yaml file to align with your deployment and XNAT instance.

After deploying the chart, you will need to configure the JupyterHub plugin preferences in the XNAT UI. You will need to set the JupyterHub API URL to http://proxy-public/jupyterhub/hub/api. Don't forget to set the path translation preferences as well.