Zero-to-jupyterhub-k8s: "504: Gateway Time-out" when accessing the proxy-public External-IP from Linux Ubuntu 16.04 machine

[UsDAnDreS]

Contents of config.yaml

I've been trying to install the Kubernettes and JupyterHub via my Linux Ubuntu 16.04 machine.

Having followed the instructions on "zero-to-jupyterhub.readthedocs.io" all the way from

• creating a Kubernetes cluster on Google Cloud (https://zero-to-jupyterhub.readthedocs.io/en/latest/google/step-zero-gcp.html),

• to "Setting up JupyterHub" page (https://zero-to-jupyterhub.readthedocs.io/en/latest/setup-jupyterhub.html)

  I ended up with a seemingly up and running JupyterHub and its corresponding proxy-public External IP, but once I try it in the browser, I get a "504: Gateway time-out" (images attached).

Anyone know what might be the issue?

[consideRatio]

Thank you for writing an issue! You made it very easy for me to grasp the situation with the screenshots and all, I appreciate it!

What does it say if you write kubectl get pods --namespace=uh-test-jupyter ?

When you ran the following command, did you write --version=v0.6 ?

helm install jupyterhub/jupyterhub \
    --version=v0.6 \
    --name=<YOUR-RELEASE-NAME> \
    --namespace=<YOUR-NAMESPACE> \
    -f config.yaml`

[UsDAnDreS]

Dear condeRatio, thanks for such swift response!

What does it say if you write kubectl get pods --namespace=uh-test-jupyter ?

When you ran the following command, did you write --version=v0.6 ?
helm install jupyterhub/jupyterhub
--version=v0.6
--name=
--namespace=
-f config.yaml`

Yes, I ran it precisely like that, as was mentioned in (https://zero-to-jupyterhub.readthedocs.io/en/latest/setup-jupyterhub.html)

On another note - when I did all of these steps in Google Cloud Console (setting up the Kubernetes cluster + Helm + JupyterHub), it worked fine and I could connect to the External IP. But it was tough for me to keep working in that Cloud Console (configuring Jupyter, the environment, files, etc etc), so I'm trying to do it directly from my Linux machine

[consideRatio]

Aha!!!

I notice that the hub Pod hub-8cd589cb9-cg7lc (controlled by a ReplicaSet hub-8cd589cb9, controlled by a Deployment hub) is Pending, I wonder why? It is probably because no Node (hardware) has resources enough to take on the job of running the Pod.

Can you run kubectl get nodes and also kubectl describe nodes?

[consideRatio]

Did you choose a n1-standard-1 machine perhaps? I actually use a n1-standard-1 machine to contain my hub and proxy etc, but I have optimized some stuff (key: reduce the hub and proxy requested CPU resources) allowing that in the config.yaml.

gcloud container clusters create <YOUR-CLUSTER> \
    --num-nodes=3 \
    --machine-type=n1-standard-2 \
    --zone=us-central1-b`

[UsDAnDreS]

@consideRatio You are absolutely on the money, and I am an absolute liar.

I completely forgot that I shrunk my "num-nodes" from 3 down to 1, and the "machine-type" from n1-standard-2 to n1-standard-1, because it was charging money at a considerable rate with previous settings (albeit I haven't even started doing anything, as I'm still extremely new to the whole thing).

I had (and still do) nothing besides "proxy: secretToken ..." two lines in my 'config.yaml', and yet to understand how to adjust it in order to optimize stuff as you said.

Thanks so much!

[consideRatio]

@UsDAnDreS I'm happy to have helped you! I opened #717, I hope to address z2jh guides pricey suggestions and perhaps optimize it to be more cost efficient by default.

I appreciate your clear communication and the screenshots! It reduces a lot of time for me as a reader!