From ad320706554d6ecb5469a2ee327fcc51c638cb66 Mon Sep 17 00:00:00 2001 From: Nachiket Roy Date: Sun, 12 Jul 2026 13:41:04 +0530 Subject: [PATCH] docs: regenerate metrics guide to include GPU and platform metrics Signed-off-by: Nachiket Roy --- docs/kepler/design/metrics.md | 198 ++++++++++++++++++++++++++++++++++ 1 file changed, 198 insertions(+) diff --git a/docs/kepler/design/metrics.md b/docs/kepler/design/metrics.md index 7d728ddb..b0bf8cff 100644 --- a/docs/kepler/design/metrics.md +++ b/docs/kepler/design/metrics.md @@ -95,6 +95,90 @@ These metrics provide energy and power information at the node level. - **Constant Labels**: - `node_name` +#### kepler_node_gpu_active_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu in active state at node level in joules +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_active_watts + +- **Type**: GAUGE +- **Description**: GPU active power (total - idle) in watts +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_idle_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu in idle state at node level in joules +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_idle_watts + +- **Type**: GAUGE +- **Description**: GPU idle power (auto-detected minimum) in watts +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu at node level in joules +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_watts + +- **Type**: GAUGE +- **Description**: Total GPU power consumption in watts +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + +#### kepler_node_gpu_info + +- **Type**: GAUGE +- **Description**: GPU device information for mapping index to UUID/name +- **Labels**: + - `gpu` + - `gpu_uuid` + - `gpu_name` + - `vendor` +- **Constant Labels**: + - `node_name` + ### Container Metrics These metrics provide energy and power information for containers. @@ -113,6 +197,19 @@ These metrics provide energy and power information for containers. - **Constant Labels**: - `node_name` +#### kepler_container_cpu_seconds_total + +- **Type**: COUNTER +- **Description**: Total user and system time of cpu at container level in seconds +- **Labels**: + - `container_id` + - `container_name` + - `runtime` + - `state` + - `pod_id` +- **Constant Labels**: + - `node_name` + #### kepler_container_cpu_watts - **Type**: GAUGE @@ -127,6 +224,32 @@ These metrics provide energy and power information for containers. - **Constant Labels**: - `node_name` +#### kepler_container_gpu_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu at container level in joules +- **Labels**: + - `container_id` + - `container_name` + - `runtime` + - `state` + - `pod_id` +- **Constant Labels**: + - `node_name` + +#### kepler_container_gpu_watts + +- **Type**: GAUGE +- **Description**: Power consumption of gpu at container level in watts +- **Labels**: + - `container_id` + - `container_name` + - `runtime` + - `state` + - `pod_id` +- **Constant Labels**: + - `node_name` + ### Process Metrics These metrics provide energy and power information for individual processes. @@ -177,6 +300,36 @@ These metrics provide energy and power information for individual processes. - **Constant Labels**: - `node_name` +#### kepler_process_gpu_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu at process level in joules +- **Labels**: + - `pid` + - `comm` + - `exe` + - `type` + - `state` + - `container_id` + - `vm_id` +- **Constant Labels**: + - `node_name` + +#### kepler_process_gpu_watts + +- **Type**: GAUGE +- **Description**: Power consumption of gpu at process level in watts +- **Labels**: + - `pid` + - `comm` + - `exe` + - `type` + - `state` + - `container_id` + - `vm_id` +- **Constant Labels**: + - `node_name` + ### Virtual Machine Metrics These metrics provide energy and power information for virtual machines. @@ -237,6 +390,30 @@ These metrics provide energy and power information for pods. - **Constant Labels**: - `node_name` +#### kepler_pod_gpu_joules_total + +- **Type**: COUNTER +- **Description**: Energy consumption of gpu at pod level in joules +- **Labels**: + - `pod_id` + - `pod_name` + - `pod_namespace` + - `state` +- **Constant Labels**: + - `node_name` + +#### kepler_pod_gpu_watts + +- **Type**: GAUGE +- **Description**: Power consumption of gpu at pod level in watts +- **Labels**: + - `pod_id` + - `pod_name` + - `pod_namespace` + - `state` +- **Constant Labels**: + - `node_name` + ### Other Metrics Additional metrics provided by Kepler. @@ -252,6 +429,27 @@ Additional metrics provided by Kepler. - `version` - `goversion` +## Experimental Metrics + +⚠️ **Warning**: The following metrics are experimental and may change or be removed in future versions. They are provided for early testing and feedback purposes. + +### Platform Power Metrics + +These experimental metrics provide platform-level power information from BMC sources (e.g., Redfish). Enable the experimental Redfish feature to collect these metrics. + +#### kepler_platform_watts + +- **Type**: GAUGE +- **Description**: Current platform power in watts from BMC (PowerSubsystem or deprecated Power API) +- **Labels**: + - `source` + - `node_name` + - `bmc_id` + - `chassis_id` + - `source_id` + - `source_name` + - `source_type` + --- This documentation was automatically generated by the gen-metric-docs tool.