9 Releases 9 Tags

RSS Feed

• v0.2.0 a226f0191c

  Compare

  v0.2.0 — Metrics page, sidebar label fix, remove app bar badge Stable

  cpfarhood released this 2026-02-19 02:23:46 +00:00 | 81 commits to main since this release

  What's Changed

  Added

  • Metrics page (/intel-gpu/metrics): real-time GPU power draw (W) and TDP sourced from node-exporter's i915 hwmon metrics via kube-prometheus-stack Prometheus. Queries rate(node_hwmon_energy_joule_total[5m]) joined with node_hwmon_chip_names{chip_name="i915"} for per-GPU current power; node_hwmon_power_max_watt for TDP. Auto-discovers Prometheus service in the monitoring namespace.

  Changed

  • Sidebar parent label changed from Intel GPU → intel-gpu

  Removed

  • App bar health badge (colored info bubble in top navigation)

  Data Source

  The Metrics page requires kube-prometheus-stack with node-exporter enabled (the default). No additional exporters needed — the i915 kernel driver already exposes hwmon sensors that node-exporter scrapes automatically on Intel discrete GPU nodes.

  Metrics shown per GPU chip:

  • Current Power (W) — derived from rate(node_hwmon_energy_joule_total[5m])
  • TDP (W) — from node_hwmon_power_max_watt
  • Power bar showing % of TDP used

  Downloads

  • headlamp-intel-gpu-plugin-0.2.0.tar.gz
    9.3 KiB

    2026-02-19 02:37:32 +00:00
  • Source Code (ZIP)
  • Source Code (TAR.GZ)