Zero to Kubernetes: Helm install of JupyterHub fails

2
votes

I'm following the Zero to Kubernetes GKE installation guide.

I've successfully installed kubernetes and helm but when it comes to do the install:

helm upgrade --cleanup-on-fail \
  --install $RELEASE jupyterhub/jupyterhub \
  --namespace $NAMESPACE \
  --create-namespace \
  --version=0.9.0 \
  --values config.yaml

The error I keep getting is:

Release "jhub" does not exist. Installing it now.
Error: failed pre-install: timed out waiting for the condition

I've tried adding the flag --timeout 1200s as per the troubleshooting instructions on GKE but it still hangs up.

Any ideas on how to resolve this? If it continues to fail I'll probably just try deploying on Azure AD but ideally GKE should work.

(For reference I've actually done this deploy twice on GKE successfully but had to delete those clusters so I'm really not sure what the issue is.)

EDIT: I resolved the issue.

Further down the config.yaml file I somehow changed the tag from 1.0.7 which was the version registered via Jupyter to 1.1.0 and there was a conflict between my Jupyter registry and my JupyterHub config.

I'm still inexperienced with Jupyter & JupyterHub so apologies if this seemed obvious to others.

1
kubernetesgoogle-kubernetes-enginekubernetes-helmjupyterhub
try to check the events by running kubectl -n $NAMESPACE get events --sort-by='{.lastTimestamp}' - Yuri G.
I've just followed the how to you shared and everything works smoothly for me. Perhaps you are facing some network issue, try to list the deployment and verify if everything works fine. Use helm list --namespace jhub. Please let me know the result. - Mr.KoopaKiller
Thanks for the replies. The how to was totally fine it was me being imprecise and forgetful with my previous registries on Jupyter that were conflicting with JupyterHub. - gun.go

1 Answers

0
votes

As mentioned in comments:

The issue was solved by cleanning old configurations from the cluster.

All the information on posted guide is working fine.