How does the cull values get used in the jupyerhub helm chart

JupyterHub
1

First of all, everything is working and I am just puzzled at how it works.

According to https://github.com/jupyterhub/zero-to-jupyterhub-k8s/blob/f11f8ea8ea857917f59b7f9a1b79a9570e21e622/jupyterhub/values.yaml#L653-L668, I am able to specify my own cull values like

cull:
  timeout: 600
  every: 60
  users: true
  adminUsers: true

And when I installed the jupyterhub helm chart and from the hub pod (using kubectl debug), I can see the process

hub-77485c4b79-5d6ql:~# ps aux
PID   USER     TIME  COMMAND
    1 1000      0:00 tini -- jupyterhub --config /usr/local/etc/jupyterhub/jupyterhub_config.py --upgrade-db
    7 1000      0:40 {jupyterhub} /usr/local/bin/python /usr/local/bin/jupyterhub --config /usr/local/etc/jupyterhub/jupyterhub_config.py --upgrade-db
   11 1000      0:02 python3 -m jupyterhub_idle_culler --url=http://localhost:8081/hub/api --timeout=600 --cull-every=60 --concurrency=10 --cull-users
  427 root      0:00 bash
  436 root      0:00 ps aux

Process 11 is exactly what I expect to see with my customized parameters.

However, I am very confused on how these parameters get into my pod. I understand that we have a configmap called hub that got mounted as volumes in /usr/local/etc/jupyterhub/config/. By inspecting the data of either the configmap or the mounted folder, I cannot find the cull parameters. Furthermore, I even read the helm chart itself and I found no use of those cull parameters and also confirmed that no such reference exists in the generate manifest.

So how did my parameters get into the hub workload, seemingly magically? :thinking:

I would really appreciate some insights. This is very cool.

2

The config values are loaded by the jupyterhub_config.py file:

3

Thank you. The question really was how the get_config() loads the values.

I think I found out the answer: zero-to-jupyterhub-k8s/jupyterhub/files/hub/z2jh.py at f11f8ea8ea857917f59b7f9a1b79a9570e21e622 · jupyterhub/zero-to-jupyterhub-k8s · GitHub shows

    for source in ("secret/values.yaml", "existing-secret/values.yaml"):
        path = f"/usr/local/etc/jupyterhub/{source}"

Basically these are secrets mounted at /usr/local/etc/jupyterhub/secret/values.yaml would contain the configuration, instead of from the configmap.