Hi.
I’m getting a problem with my Jupyterhub deployment on k8s. No matter what image I’ve supplied it, it still boots up the default single image server.
Here is my config.yaml file:
proxy:
secretToken: <some_token>
singleuser:
image:
name: jupyter/datascience-notebook
tag: latest
# pullSecrets:
# - "regcred"
hub:
config:
Authenticator:
admin_users:
- <ome_users>
allowed_users:
- <some_users>
GitHubOAuthenticator:
client_id: "<some_id>"
client_secret: "<some_secret>"
oauth_callback_url: "<my_url>/hub/oauth_callback"
JupyterHub:
authenticator_class: github
However, when I update the hub with
helm upgrade --cleanup-on-fail \
--install jhub jupyterhub/jupyterhub \
--namespace=jhub \
--version=0.11.1 \
--values config.yaml
The subsequent singleuser sessions don’t have any of the datascience’s packages installed. If I instead uncomment the pullSecrets and use my own docker image as in the tutorial:
FROM jupyter/datascience-notebook:latest
# create a user, since we don't want to run as root
ARG JUPYTERHUB_VERSION=1.3.0
RUN pip3 install --no-cache \
jupyterhub==$JUPYTERHUB_VERSION
CMD ["jupyterhub-singleuser"]
The same problem persists. When I launch these pods outside of jupyterhub and inspect the installed conda environment, I’ve found that they contain the packages. So jupyterhub has a bug that is preventing it from using jupyter/datascience-notebook:latest image or else I have donfigured something wrong. Advice is appreciated. Thanks.