v1.0-gamma

Summary of all changes merged for the gamma release

New Features

Carbon Intensity API Integration

Real-time carbon intensity values are now fetched from the carbonintensity.org.uk API for UK-based clusters, replacing the static config. value. A generic API service layer was introduced, which can be used to plug-in other regional carbon intensity sources. The default config. value remains as a fallback for unsupported regions (non-UK regions). Per-job and per-day emission records are stored in the database along with day-wise average carbon intensity values.

New plots added

• Carbon footprint of Failed Jobs

  A new plot shows the carbon footprint of failed jobs and how it compares to that of successful jobs. This replaces the failed job count plot from previous versions. Available at both user and institutional levels.

• GPU usage

  Day-wise GPU core-hours have been added to the Resource usage section, giving users and institutions visibility into GPU consumption over time.

• Energy charts

  A pie chart and bar chart have been added to visualise how energy usage is distributed across GPU, CPU, memory, and other sources. Available at both user and institutional levels.

Database overwrite protection

A skip_if_db_existsflag can now be set in config.yaml. This enables the user to skip database deletion and creation in case the target database is already setup. Furthermore, a the script now performs a check to see whether the target database exists or not - if it does, a warning is displayed.

UI changes

The dashboard has been refreshed with a more coherent colour scheme. The introduction section is now collapsible (and collapsed by default) to allow users to access the graphs immediately. Section order has been revised to surface carbon footprint and energy usage first, with CPU, GPU, memory, and job graphs consolidated into a ‘resource usage’ section. Textual content throughout has also been reviewed and updated for clarity.

Bug Fixes

Zero-value memory fields no longer cause errors

Two separate edge cases where NeededMemX and ReqMem being zero caused runtime errors have been resolved.

Add-users script attribute error fixed

GADashboardAddUsers previously did not call the parent __init__, leaving db_pass and grafana_pass uninitialised. Fixed by calling super().__init__() and removing the redundant duplicate initialisation.

Configuration and Infrastructure

Config file format changed from .txt to .yaml

The dashboard configuration file is now .yaml instead of .txt. Validation logic, the install workflow, the config template, and the sample config file have all been updated. The postcode field has been added to cluster_info.yaml to support the carbon intensity API lookup.

License updated

The LICENSE file has been revised to include program details and copyright information.

Beta - revision 2

beta version of the dashboard - revision 2

This second beta version essentially includes cosmetic changes on the Grafana dashboard:

• Use Grafana light theme by default
• Reorder and rename the sections in the dashboard
• The sections in the dashboard are now collapsible
• Hide the variables at the top of the pag

What's Changed

• Update User dashboard, including the reordering of the sections by @ens-lgil in #77
• Fix JSON dashboard and its import module by @ens-lgil in #78
• Add code to set the default theme when setting up Grafana by @ens-lgil in #79
• Minor cosmetic harmonizations on the dashboard by @ens-lgil in #80
• Merge branch "gamma-frontend" by @ens-lgil in #81
• Add missing grafana_ga.organization import by @ens-lgil in #82
• Prepare code for new demo branch by @ens-lgil in #83

Full Changelog: v1.0-beta...v1.0-beta-2

Beta

beta version of the dashboard

This is a fully functioning version of the dashboard, ready for a first round of external validation.