Issue during installation of LSP (Containers in Nublado failing to spawn)

Support Rubin Science Platform
,
#1

Hello,

I’m in the process of trying to deploy an instance of the LSP on an Openstack / Kubernetes cluster, and I’m encountering some issues and was wondering if someone can point me in the right direction.
The resources & documentation that I’m using are from the master branch of https://github.com/lsst-sqre/lsp-deploy which uses the Helm charts from here: https://github.com/lsst-sqre/charts .

Initially I was running into some issues with the Fileserver & Nublado, which I got around temporarily using a different NFS Docker image. However the issue that I’m now seeing, is at the point of spawning notebooks from the list of images (lsstsqre/sciplat-lab). Specifically, I’ve tried a launching notebooks from a few different images including the “Recommended” one, all of which seem to fail, with the most common exception being the following (shows up in browser):

“Redirect loop detected. Notebook has jupyterhub version [x.y.z], but the Hub expects 1.1.0. Try installing jupyterhub==1.1.0 in the user environment if you continue to have problems.”

I’m using a Proxy server, and was initially thinking it may also be an issue with my configuration, but haven’t narrowed it down to anything specific yet.

I’m wondering if this may be an issue with version miss-matches between what the Helm chart configuration uses and what is in Docker Hub? Is there anything obvious that I may be missing or doing wrong, such as using the wrong versions of documentation or configuration files? Has anyone else has encountered similar problems?

Thanks in advance,

Stelios Voutsinas
University of Edinburgh

#2

Hi, what version of Kubernetes are you running?

#3

Hello,
I’m using Kubernetes version 1.17.3.
(Helm version is 2.16.3)

#4

I have not tested with anything past 1.16. I don’t expect it to make a difference. It’s working fine here, but that may be with environment overrides you don’t have. What is the image for jupyterhub you’re using? lsstsqre/sciplat-hub:0.18.6 is current and is working for me.

#5

Stelio, do you want an invite to a slack channel so we can help you more directly? Drop me an email address.

#6

For Jupyterhub I am using the “latest” lsstsqre/sciplat-hub image, which “docker inspect” shows as being version: 0.18.7.

#7

I’m on the LSSTC slack channel already (Username: Stelios Voutsinas), should I post on any specific channels or would it be best to direct message someone?

#8

For closure, we sorted Stelios out on Slack, he’s good now.