Not beeing able to spawn notebook servers with jupyterhub (original) (raw)

I was trying to configure a docker environment where one of my services would be a jupyterhub with pre-customised images. I am using traefik as a reverse proxy because when deploying I will use a domain instead of localhost. Here are my docker-composer and jupyterhub_config.py

version: “3.8”

services:
traefik:
image: traefik:v3.0
command:
- “–api.insecure=true”
- “–entrypoints.websecure.address=:443”
- “–providers.docker=true”
- “–providers.file.filename=/etc/traefik_dynamic.yml”
- “–providers.file.watch=true”
ports:
- “443:443”
- “8888:8888”
volumes:
- “/var/run/docker.sock:/var/run/docker.sock”
- “./certs:/certs”
- “./traefik_dynamic.yml:/etc/traefik_dynamic.yml”
networks:
- lepocolab-net
labels:
- “traefik.enable=true”

lepocolab:
build: ./build/lepocolab
image: lepocolab/hub
container_name: lepocolab
environment:
- CONFIGPROXY_AUTH_TOKEN=dummytoken
volumes:
- ./conf:/srv/jupyterhub
- ./shared:/home/jovyan/shared
- /var/run/docker.sock:/var/run/docker.sock
restart: always
labels:
- “traefik.enable=true”
- “traefik.http.routers.jupyterhub.rule=Host(localhost) && PathPrefix(/colab)”
- “traefik.http.routers.jupyterhub.entrypoints=websecure”
- “traefik.http.routers.jupyterhub.tls=true”
- “traefik.http.services.jupyterhub.loadbalancer.server.port=8888”
networks:
- lepocolab-net

networks:
lepocolab-net:
driver: bridge

---------------------------

CONFIGURAÇÃO GERAL DO JUPYTERHUB

---------------------------

c.JupyterHub.bind_url = “http://0.0.0.0:8888/colab”
c.JupyterHub.cleanup_servers = False
c.JupyterHub.proxy_api_url = “https://localhost/colab/hub/api”
c.JupyterHub.proxy_auth_token = os.environ[‘CONFIGPROXY_AUTH_TOKEN’]
c.DockerSpawner.network_name = ‘lepocolab_lepocolab-net’ # nome da rede Docker

---------------------------

AUTENTICAÇÃO

---------------------------

c.JupyterHub.authenticator_class = “dummy”
c.DummyAuthenticator.password = “123”

---------------------------

DOCKERSPAWNER

---------------------------

c.JupyterHub.spawner_class = “dockerspawner.DockerSpawner”

c.DockerSpawner.allowed_images = [
‘lepocolab/datascience-notebook:ubuntu-24.04’,
‘lepocolab/pytorch-notebook:ubuntu-24.04’,
‘lepocolab/tensorflow-notebook:ubuntu-24.04’,
]

c.DockerSpawner.http_timeout = 300 # Aumentar o tempo de espera para 60 segundos

c.JupyterHub.log_level = ‘DEBUG’

c.DockerSpawner.volumes = {
‘/home/user/jupyterhub/users/{username}/work’: ‘/home/{username}/work’,
}

c.JupyterHubSingleUser.check_hub_version = False
c.DockerSpawner.api_url = None # Impede a comunicação com a API do Hub

c.DockerSpawner.port = 8000
c.DockerSpawner.remove = True
c.DockerSpawner.extra_create_kwargs = {
‘user’: ‘root’
}
c.DockerSpawner.extra_host_config = {
‘mem_limit’: ‘1G’,
‘memswap_limit’: ‘1G’
}

---------------------------

SERVIÇO DE LIMPEZA

---------------------------

c.JupyterHub.services = [
{
‘name’: ‘idle-culler’,
‘admin’: True,
‘command’: [
sys.executable,
‘-m’, ‘jupyterhub_idle_culler’,
‘–timeout=604800’,
],
},
]

manics April 20, 2025, 3:44pm 2

We’ve got an example of deploying JupyterHub with Docker Compose in

Can you try starting from that example, and iteratively add in your customisations until it breaks? When you get to that point please share your debug logs from JupyterHub. Thanks!