JupyterHub is failing due "didn't respond in 30 seconds" and "Back-off restarting failed container"

11/14/2019

I'm trying to run JupyterHub locally with Kind and Helm 3.

To launch it:

kind create cluster

RELEASE=jhub
NAMESPACE=jhub
kubectl create namespace ${NAMESPACE}
helm upgrade --install ${RELEASE} jupyterhub/jupyterhub \
  --namespace ${NAMESPACE}  \
  --version=0.8.2 \
  --values config.yaml

kubectl -n jhub port-forward service/proxy-public 8000:80

config.yaml is pretty minimal:

proxy:                                                                                                                                                        
  secretToken: "token_goes_here" 
  service:
    type: ClusterIP
hub:
  uid: 0
  fsGid: 0
  extraConfig:
    myConfig.py: |
      c.JupyterHub.spawner_class = 'kubespawner.KubeSpawner'

http://localhost:8000/ works beautifully, I log in and the Hub is trying to launch my single user image. But it is failing with:

Server requested 2019-11-14 14:59:50+00:00 [Normal] Successfully assigned jhub/jupyter-user1 to kind-control-plane

2019-11-14 14:59:51+00:00 [Normal] Container image "jupyterhub/k8s-singleuser-sample:0.8.2" already present on machine

2019-11-14 14:59:51+00:00 [Normal] Created container notebook Spawn failed: Server at http://10.244.0.10:8888/user/user1/ didn't respond in 30 seconds

And sometimes, I'm getting:

[Warning] "Back-off restarting failed container"

-- Chiron
jupyter
jupyter-notebook
kubernetes-helm

1 Answer

11/19/2019

This could be solved by adding the following to your config.yaml in order to make it working locally:

c.Spawner.args = ['--allow-root']
-- Chiron
Source: StackOverflow