Configure compute quotas - 2i2c Infrastructure Guide

This guide explains how to enable and configure compute quotas using the jupyterhub-usage-quotas system.

How to enable¶

The entry 12-conditional-usage-quotas under the jupyterhub.hub.extraConfig key in the basehub/values.yaml file conditionally enables the JupyterHub service running the usage dashboard. This service is opt-in by adding

basehub:
  jupyterhub:
    custom:
      usageQuotas:
        enabled: true

to a hub’s values.yaml file. The service is currently opt out for all Cloudbank hub’s due to issues with CI/CD deployment for this cluster, however this deployment strategy will be updated when this issue is fixed.

The quota manager itself is a library is unconditionally installed in all hub images through the helm-charts/images/hub/requirements.txt file.

Add a compute quota policy¶

See JupyterHub Usage Quotas -- Policy configuration.

Enable usage quota dashboard components¶

In the usage quota dashboard, the Compute component is disabled by default since compute quota policies do not exist by default. You can enable this component for each individual hub with a compute policy applied.

config/clusters/$CLUSTER_NAME/$HUB_NAME.values.yaml

hub:
  config:
    UsageViewer:
      enable_compute: true

Grafana Dashboards¶

The general grafonnet code to define the usage quotas Grafana dashboard can be found in /dashboards/usage-quotas.jsonnet. See Manually deploy dashboards for how to deploy this to communities with compute usage quotas enabled.