Actually getting logs from JupyterHub?

Hi, folks,

Having a major issue with some authentication and Error 500s, and can’t debug, because I can’t seem to access any logs. My config.yaml for the deployment has:

debug:
  enabled: true

which I thought was sufficient to get debug mode enabled? I’m running the 3.0.0 helm chart, and everything was working a couple of weeks ago on a two-year old version. Updating everything (new images for the single-user mode, updated helm, etc.), and now after OAuth, users are just getting immediate Error 500s. Which wasn’t happening to me, so I didn’t notice it to debug earlier - I was able to still login fine through all of this. And still am.

So what am I missing to get actual debug logs so I can try to track down what is happening? I’ve got: no messages in journalctl; no logs being written that I can find. I want the hub to dump errors somewhere, please.

Since you’re using the Helm chart the logs are handled by Kubernetes, so e.g. kubectl -n <namespace> logs deploy/hub should show the hub logs.

To see the logs for other pods you’ll probably need to use the name of the pod. See the commands in

3 Likes

Adding to what manics said, you can also look at events with kubctl -n namespace get events and kubectl -n namespace describe pod pod_name .

logs, events and describe are the three kubectl commands I use most to troubleshoot.

1 Like