Managing Notebooks

Monitor, access, restart, and delete Jupyter notebook instances from the Notebooks page.

Opening a Notebook

Click the Open button in the notebook row, or navigate directly to:

https://<domain>/jupyter/<name>/

The platform redirects you to the auto-login URL, which includes the Jupyter token and opens JupyterLab without requiring manual authentication.

Monitoring Resource Usage

The Usage column shows real-time CPU and memory consumption for each notebook:

• CPU — displayed in millicores (e.g., 245m = 0.245 cores)
• Memory — displayed in MB (e.g., 512 MB)

These metrics update when the page is refreshed. They require metrics-server to be installed.

To see historical trends, you can integrate with Prometheus/Grafana using standard Kubernetes pod metrics.

Pod Status

The Status badge shows the current state of the notebook pod:

Status	Meaning
Running	Notebook is ready and accessible
Starting	Pod is initializing or image is being pulled
Pending	Waiting for cluster resources (CPU/memory/GPU)
Failed	Pod crashed — check pod logs for details

Restarting a Notebook

If a notebook becomes unresponsive, restart the pod via kubectl:

kubectl rollout restart deployment/<notebook-name> -n apps

Or delete the pod and let the Deployment controller recreate it:

kubectl delete pod -l app=<notebook-name> -n apps

Deleting a Notebook

1. Click the Delete button in the notebook row
2. Confirm the deletion

Caution

Deleting a notebook removes the Deployment and Service but does not delete the PersistentVolumeClaim. Notebook data is preserved. To also remove data, manually delete the PVC:

kubectl delete pvc <notebook-name>-data -n apps

Viewing Logs

To troubleshoot a failing notebook:

kubectl logs -l app=<notebook-name> -n apps --tail=100