Networking issue with opennebula k8s service

Binder
1

Hi all, and thanks to the team for your amazing job!

So we are trying to install binderhub on bare-metal cluster, we already successfully spawned one with openstack but we needed to change our provider for better hardware and support. Our new provider uses opennebula and we are using this service to spawn a k8s cluster : https://docs.opennebula.io/appliances/service/kubernetes.html

k8s: 1.18
helm: 2.16.9

However we have networking issues after installing binderhub:

  1. user-scheduler pods are crashing :
    Readiness probe failed: Get http://10.244.6.11:8081/jupyter/hub/health: dial tcp 10.244.6.11:8081: connect: connection refused
  2. hub-pod is also crashing:
    Liveness probe failed: Get http://10.244.5.8:10251/healthz: net/http: request canceled (Client.Timeout exceeded while awaiting headers)

I do not have experience on networking, but I am just wondering why he is trying to read 10.244.X when the IP from calico is 10.244.0.0/16.
Do you have any other thoughts how I could debug that ?

Thank you,

2

I gave up using the opennebula k8s service, but I still have SSL issues with a bare metal cluster. If any devs has deep knowledge on k8s/binderhub networking, please let me know it would be appreciated.