SSL on Jupyterhub

JupyterHub Zero to JupyterHub on Kubernetes
1

Hi,

I’ve been following this guide on setting https on my deployed instance (on Kubernetes). Do I have to buy and setup a domain before the Set up automatic HTTPS stage?
What if I want HTTPS without getting a domain?
I did try using the IP and domain name of the load balancer but that produced an SSL_ERROR_INTERNAL_ERROR_ALERT error.

  • I’m running my cluster on AWS, I first deployed it without any https and then tried to add my IP as the domain but without success.

Any thoughts?

2

With keys that I’ve generated locally with openssl it works but stating it’s not secure.

3

You’ll need a domain that points to the IP of your cluster before you can setup a https certificate.

1 Like
4

I have a subdomain: Cluster.mydomain.io and I’ve put a dns record pointing to the dynamic IP of the public-proxy, I’ve also checked it propogated via dig +trace and I can see the A record to the IP.
When I’m updating this block:

proxy:
  https:
    enabled: true
    hosts:
      - cluster.mydomain.io
    letsencrypt:
      contactEmail: someEmail

I’m getting an error: SSL_ERROR_INTERNAL_ERROR_ALERT
I didn’t see anything on the hub/P.proxy logs while running on debugging.

5

PS, I’ve allocated an EIP in my AWS account and allocated it within the cluster’s vpc (which the EKS sits), and the A record I created is routing to this IP. I’ve also attached the configuration in the yaml:

proxy:
  service:
    loadBalancerIP: SOME_IP
6

Where are you seeing this error?

7

When I’m trying to load the subdomain in a firefox browser.
I did try running get events on the cluster and found out this:

Error syncing load balancer: failed to ensure load balancer: LoadBalancerIP cannot be specified for AWS ELB
13m         Normal    LoadbalancerIP           service/proxy-public                   -> 52.44.some_IP
13m         Normal    ScalingReplicaSet        deployment/proxy                       Scaled down replica set proxy-665d98df47 to 0
13m         Normal    ScalingReplicaSet        deployment/proxy                       Scaled up replica set proxy-7f4585474 to 1
2m47s       Normal    NoPods                   poddisruptionbudget/user-placeholder   No matching pods found
8

Alright, I was able to solve it.
here’s what I had to do if someone hits the same issue:

I was running on EKS within a VPC + an elastic IP.

I configured first the ip within the config.yaml like the docs says.
I then moved the dns into Route 53.
I had to create an A record to map from my subdomain to the ELB that was deployed through an alias. Once that propagated (I’ve waited ~10 minutes), I’ve added the subdomain on the https section and updated the chart, it worked fine!

Thanks for the help!

1 Like