Ask quick dev questions about mybinder.org, BinderHub, and repo2docker. Use discourse.jupyter.org for user questions, support, and discussion.
pip -r requirements.txt. Also, I want to use automatically generated requirement.txt from Python virtual environment. In this case, there is some error that there are no python libraries. Is there a way to pass when there is no library? I found some solutions to pass it using customed python code with requirements.txt, but I couldn't find a way to use MyBinder configuration options.
Cannot assign requested address . Any ideas or debugging tips for this?
Thanks!
(also, @sgibson91 , is there a way of retrieving the former hubs minutes? I couldn't find a link to the previous pages on the current hackmd. Thanks!)
4 replies
config:
BinderHub:
auth_enabled: true
use_registry: true
image_prefix: dockerhubprefix/binder-prod-
hub_url: https://binderhub.mycloud/
jupyterhub:
cull:
# don't cull authenticated users
users: False
custom:
binderauth_enabled: true
hub:
config:
GitHubOAuthenticator:
client_id: redacted
client_secret: redacted
oauth_callback_url: https://binderhub.mycloud/hub/oauth_callback
admin_users:
- my-github-username
allowed_users:
- someoneelse
scope:
- read:user
JupyterHub:
authenticator_class: github
redirectToServer: false
services:
binder:
oauth_no_confirm: true
oauth_redirect_uri: "http://binderhub.mycloud/oauth_callback"
oauth_client_id: "binder-oauth-client-test"
singleuser:
# to make notebook servers aware of hub
cmd: jupyterhub-singleuser
#auth: {}
ingress:
enabled: true
annotations:
kubernetes.io/ingress.class: "nginx-jupyterhub"
nginx.ingress.kubernetes.io/proxy-body-size: 200m
hosts:
- binderhub.mycloudWhen I visit binderhub, I'm shown a binderhub-branded 404 page. Looking at pod logs, I see this:
[I 210427 02:37:18 log:140] 302 GET / -> https://binderhub.mycloud/hub/api/oauth2/authorize?client_id=binder-oauth-client-test&redirect_uri=http%3A%2F%2Fbinderhub.mycloud%2Foauth_callback&response_type=code&state=[secret] (@10.128.5.123) 1.68ms
[W 210427 02:37:18 log:140] 404 GET /hub/api/oauth2/authorize?client_id=binder-oauth-client-test&redirect_uri=http%3A%2F%2Fbinderhub.mycloud%2Foauth_callback&response_type=code&state=[secret] (@10.128.5.123) 10.79msAny idea what's going on?
Hi all, I've got jupyterhub running and then I started binderhub with:
python3 -m binderhub -f testing/local-binder-k8s-hub/binderhub_config.pybut getting
[E 210503 19:16:47 builder:317] Tornado HTTP Timeout error: Failed to get image manifest for binder-2dexamples-2ddemo-2djulia-28add9:39a18c401b854f472ea72d9dedd7b11988ef289b
Traceback (most recent call last):
File "/home/aacurtis/cdao-dv-k8s/binderhub/binderhub/builder.py", line 313, in get
image_manifest = await self.registry.get_image_manifest(*'/'.join(image_name.split('/')[-2:]).split(':', 1))
File "/home/aacurtis/cdao-dv-k8s/binderhub/binderhub/registry.py", line 198, in get_image_manifest
auth_resp = await client.fetch(auth_req)
tornado.httpclient.HTTPClientError: HTTP 401: UnauthorizedCan someone tip me off to what I need to do to fix this? I assume jupyterhub isn't letting binderhub in?
Hi! I am trying to setup binderhub on already existing kubernetes cluster, but unfortunately I get
User "developerssso:EKSGetTokenAuth" cannot get resource "clusterroles" in API group "rbac.authorization.k8s.io" at the cluster scopeMy guess is that it is due to https://github.com/jupyterhub/binderhub/blob/c7cbf63e27b09d223035f8bdda2dab189ad826d1/helm-chart/binderhub/templates/rbac.yaml#L49
Is there a way to turn it off as I would preferrably not give such authorisation to it?
I have tried helm install binder jupyterhub/binderhub --version=0.2.0-n563.h72d30fb --namespace=data -f secret.yaml -f config.yaml --set imageCleaner.enabled=false but I get the same issue
If that's the wrong place for such question, sorry!
@Antobiotics are you sure you want to workaround not having access to manipulate Cluster wide k8s resources? It will have many drawbacks, not just the first error you run into and workaround in BinderHub, but also in the JupyterHub Helm chart acting as a dependency.
I suggest you look into granting your k8s user (perhaps coupled to an AWS user somehow) that you are acting with sufficient permissions when working with helm and kubectl.
binderhub:
imageCleaner:
enabled: false
jupyterhub:
scheduling:
userScheduler:
enabled: false
Hi folks! Does anyone know why a binder build would be replacing CPython with pypy??? In this PR for a work project we found in the build logs that the mamba update command at build time is updating cpython to pypy. That then causes problems loading scipy...
For now I'm brute-forcing a workaround by manually updating scipy itself, which seems to do the trick of reverting back to CPython, but I'm puzzled by this behavior. Any ideas would be appreciated.
pypy string in some of the package names and it all made sense. B/c the error we were getting didn't mention pypy at all, just not being able to import scipy due to a failure deep in a C callback API (which in retrospect makes sense would be a pypy bug :)
BTW, in the past three days I've been seeing a lot of
Found built image, launching...
Launching server...
Launch attempt 1 failed, retrying...Sometimes it resolves eventually (long wait), sometimes it just times out altogether...
Is anyone else seeing the same? I looked at the status page but it shows all green and "no data" on the plots, so I'm not sure what's going on...
I want to change the server to jupyter-server for a particular repo. I've tried using the start file
#!/usr/bin/env bash
echo "${@}"
exec jupyter lab "${@:3}"but it seems that something is going wrong and Binder is failing to launch the built image server.
When I use r2d locally, I have to set --no-browser but otherwise it works. Any ideas?
- Do you think automatically unpacking zip files on build of zenodo uploads would be a good idea?
- If yes, in which repository should I open an issue about it (I imagine this could be implemented on different levels)?
