jupyterhub/jupyterhub

Ask quick dev questions about JupyterHub, the multi-user server for Jupyter notebooks. Use discourse.jupyter.org for user questions, support, and discussion.

Erik Sundell

@consideRatio

I described the new kind of failure in https://github.com/jupyterhub/jupyterhub/issues/3258#issuecomment-729622785.

(new as in I had not reported reported it before, but I think I've seen it in the past)

I think we see these issues with Github actions that we didn't with travis because of different performance on their servers, making certain things fail by chance more often by the systematic difference in hardware etc.

Tiemen Schuijbroek

@Tiemen_gitlab

Anybody got any idea how to debug the Gitlab OAuthenticator on a Kubernetes setup?

Tiemen Schuijbroek

@Tiemen_gitlab

Check, this appears to be a mismatch between the Helmchart and the current config keys in JupyterHub itself. Although:

auth:
  whitelist:
    users:

is still in the Helmchart, the current recommended way seems to be to change c.Authenticator.allowed_users in the additional Hub config.

carstenpohllhind

@carstenpohllhind

Hi! I have a question for the kubespawner parameter called self.user_options In the documentation of the kubespawner doc it says, that the result of that form is given to MY spawner. I must admit, which spawner is meant by that?

Jeremiah Evans

@zjevans

Hey, I'm trying to find setting in the kubespawner that would allow me to proxy additional ports from the spawned notebook container

Alex

@pennycoders

Hello everyone. Can I set limits per profile?

I am trying to follow this guide: https://zero-to-jupyterhub.readthedocs.io/en/latest/

Nick Levandoski

@metric-chicken

@pennycoders yes. If by limits you mean ram, storage, etc.

Min RK

@minrk

@zjevans you might be interested in jupyter-server-proxy

magnuhho

@magnuhho

Hi. I am setting up Jupyterhub on k8s 18.10 using helm, chart version 0.8.2 = jupyterhub 0.9.6. The hub goes into a crash loop, trying to fetch proxy routes fail. ( Proxy: Fetching GET http://10.233.63.110:8001/api/routes gives HTTP 599). Googled alot, can't seem to solve it. Hints?

magnuhho

@magnuhho

Installing with the 0.9.0 chart, as of https://zero-to-jupyterhub.readthedocs.io/en/latest/jupyterhub/installation.html gives the same error

config.yaml consists of the proxy.secretToken, db.type: sqlite and singleuser.storage.type: none to avoid pvcs

magnuhho

@magnuhho

running jupyterhub with --log-level=10 reveals the url, Proxy: Fetching GET http://10.233.1.215:8001/api/routes
, which is the proxy-api service.

The proxy-api service endpoint address/port seem correct, and configurable-http-proxy is running as expected in the proxy pod

Dhruv Kumar

@darbaga

Hi. I'm visually impaired (use a screen reader) and the terminal doesn't render right in the notebook app. Is there a way to get access to the actual terminal through ssh? (I use TLJH, which seems to have its own conda install which isn't manipulatable globally.)

magnuhho

@magnuhho

I changed operating systems for the Kuberneters nodes from Fedora CoreOS to CentOS, and things just started working.

Angus Hollands

@agoose77

@magnuhho yuvipanda has implemented an SSH server for JupyterHub. The GitHub repository is here: https://github.com/yuvipanda/jupyterhub-ssh

I have never tried it myself.

Joshua Cook

@joshuacook

Am doing vanilla install of JHub on GCP K8s per the docs. The process times out at the helm upgrade step. Have tried the suggestion of increasing the timeout time (to as much as 120 min) and the process still times out. Any suggestions?

Huy Chau

@huychau

hi all, I am using Jupyterhub K8s (0.7.0) and got the error

[I 2020-11-25 07:44:25.781 JupyterHub app:1849] Hub API listening on http://0.0.0.0:8081/hub/
[I 2020-11-25 07:44:25.781 JupyterHub app:1851] Private Hub API connect url http://172.21.15.127:8081/hub/
[I 2020-11-25 07:44:25.781 JupyterHub app:1864] Not starting proxy
[D 2020-11-25 07:44:25.781 JupyterHub proxy:296] Fetching routes to check
[D 2020-11-25 07:44:25.782 JupyterHub proxy:678] Proxy: Fetching GET http://172.21.15.222:8001/api/routes
[E 2020-11-25 07:44:45.784 JupyterHub app:1952]
    Traceback (most recent call last):
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/app.py", line 1950, in launch_instance_async
        await self.start()
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/app.py", line 1895, in start
        await self.proxy.check_routes(self.users, self._service_map)
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/proxy.py", line 57, in locked_method
        return await method(*args, **kwargs)
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/proxy.py", line 297, in check_routes
        routes = await self.get_all_routes()
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/proxy.py", line 725, in get_all_routes
        resp = await self.api_request('', client=client)
      File "/usr/local/lib/python3.6/dist-packages/jupyterhub/proxy.py", line 686, in api_request
        result = await client.fetch(req)
    tornado.curl_httpclient.CurlError: HTTP 599: Connection timed out after 20001 milliseconds

[D 2020-11-25 07:44:45.786 JupyterHub application:647] Exiting application: jupyterhub

Please help me on it, thanks?

I must use this version and can not upgrade anymore. I also update the hub and proxy deployment strategy type to Recreate.

Antoine Toussaint

@AntoineToussaint

I am trying to install jupyterhub from latest helm chart (0.10.4) in a namespace that is istio enabled, the install stalls at:
client.go:544: [debug] hook-image-awaiter: Jobs active: 1, jobs failed: 0, jobs succeeded: 0
Error: failed pre-install: timed out waiting for the condition
anybody had the same issue and know how to fix it?

the install works well with istio auto injection disabled

Antoine Toussaint

@AntoineToussaint

note that a solution is to disable the prePuller hook

Joshua Cook

@joshuacook

I had the same issue

I do not have a solution

Antoine Toussaint

@AntoineToussaint

@joshuacook in the config, I disable the prePuller so at least the install finishes, now I have to figure out the proxy stuff but seems it is solved with https://github.com/splunk/jupyterhub-istio-proxy

Joshua Cook

@joshuacook

thanks :)

Antoine Toussaint

@AntoineToussaint

quick hack though, you can enable istio auto inject after the install and all jupyter pods come now with the sidecar which is what I wanted (for multi cluster with istio)

Joshua Cook

@joshuacook

how did you disable the prePuller in the config?

Ilya Trushchenko

@white-hat

:wave:
Have you ever encountered such a bug - non admin users are denied access to their own notebooks?

They are able to login to the Hub a create a notebook, but they cannot access it later

I'm using oauthenticator.generic.GenericOAuthenticator with Okta

Ilya Trushchenko

@white-hat

[D 2020-11-25 18:47:54.073 JupyterHub utils:220] Server at http://10.255.145.169:8888/user/ilya.t@preview-***.com/ responded with 302
[I 2020-11-25 18:47:54.073 JupyterHub base:893] User ilya.t@preview-***.com took 4.754 seconds to start
[I 2020-11-25 18:47:54.073 JupyterHub proxy:258] Adding user ilya.t@preview-***.com to proxy /user/ilya.t@preview-***.com/ => http://10.255.145.169:8888

it seems to add my user to the proxty, but still not allowing it to the notebook

Ilya Trushchenko

@white-hat

I can also see both API token and authorized application in "Token" tab for the user

Mehmet

@MhmtErsy

Hi, I have a problem with JupyterHub-YarnSpawner. When I launch a X core-Y GB server. I can't use more than 50% of the core. I have been using CDH platform with YARN that configured as Fair Scheduler.

1 reply

Is it could be related to a configuration that missed?

Antoine Toussaint

@AntoineToussaint

@joshuacook in config:
prePuller:
hook:
enabled: false

Bovigny

@cbovigny_gitlab

Hi everybody,
I have a question about the jupyterhub on k8s.

Why the application would like to be able to patch and/or update persistentvolume resources, in which situation would the application do that?

Thank you very much

Animesh

@anick16

Hi All, I am configuring Jupyterhub on my k8s cluster and stuck on the LDAP. It gives Invalid Password for all the users. I used helm and tried many combinations in the config.yaml. I tried AD config, it correctly identifies and errors out there is no entry for a non-exisitng account. However, returns Invalid Password for a correct user/pass. Need your help guys

--
proxy:
secretToken: "88888888888888888888888888888888"
ingress:
enabled: true
hosts:

- my-machine.org

auth:
admin:
users:

  - my_username

type: ldap
ldap:
server:
address: my.ad.server.com
ssl: true
dn:
templates:

    cn={username},ou=person,dc=my,dc=ad,dc=server,dc=com

Animesh

@anick16

Using helm version 0.9.0 for installation

Animesh

@anick16

@kuriakinzeng I am facing similar issue reg jupyterhub LDAP config as you posted earlier. Did you get any luck? I am stuck, it gives Invalid Password every time.
JupyterHub ldapauthenticator:379] Invalid password for user my_username

shreeishitagupta

@shreeishitagupta

I want to have audit logs of jupyterhub. Please let me know if you guys know anything.

ifquant

@ifquant

hi,all. i want to known jupyterhub could work with s3 for user storage? is there some example?

Min RK

@minrk

You can use s3contents to store user files in s3.

ifquant

@ifquant

thanks,@minrk. is there any method to configure kubespawner to support s3?

Min RK

@minrk

to do it at the kubernetes level to mount the bucket transparently as a volume, I think you want a fuse mount, e.g. this post for gcs: https://pliutau.com/mount-gcs-bucket-k8s/

otherwise, installing and configuring s3contents in the user pods ought to work

Where communities thrive

People

Repo info

Activity