Skip to content

feat(vGPUmonitor): add missing GPU hardware health and power metrics #2466

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
devGPP23 wants to merge 1 commit into from
Closed

feat(vGPUmonitor): add missing GPU hardware health and power metrics #2466

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
104 changes: 104 additions & 0 deletions cmd/vGPUmonitor/metrics.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,6 +70,36 @@ var (
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

hostGPUTemperatureDesc = prometheus.NewDesc(
"hami_host_gpu_temperature_celsius",
"GPU Temperature in Celsius",
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

hostGPUPowerUsageDesc = prometheus.NewDesc(
"hami_host_gpu_power_usage_milliwatts",
"GPU Power Usage in milliwatts",
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

hostGPUPowerLimitDesc = prometheus.NewDesc(
"hami_host_gpu_power_limit_milliwatts",
"GPU Power Limit in milliwatts",
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

hostGPUMemoryUtilizationDesc = prometheus.NewDesc(
"hami_host_gpu_memory_utilization_ratio",
"GPU Memory Bandwidth Utilization ratio (0-100)",
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

hostGPUMemoryTotalDesc = prometheus.NewDesc(
"hami_host_gpu_memory_total_bytes",
"GPU Total Memory in bytes",
[]string{"device_index", "device_uuid", "device_type"}, nil,
)

ctrvGPUdesc = prometheus.NewDesc(
"hami_vgpu_memory_used_bytes",
"vGPU device memory usage in bytes",
Expand Down Expand Up @@ -188,6 +218,11 @@ func sendLegacyMetric(ch chan<- prometheus.Metric, desc *prometheus.Desc, valueT
// These descriptors are used by the Prometheus registry to register the metrics.
func (cc ClusterManagerCollector) Describe(ch chan<- *prometheus.Desc) {
ch <- hostGPUdesc
ch <- hostGPUTemperatureDesc
ch <- hostGPUPowerUsageDesc
ch <- hostGPUPowerLimitDesc
ch <- hostGPUMemoryUtilizationDesc
ch <- hostGPUMemoryTotalDesc
ch <- ctrvGPUdesc
ch <- ctrvGPUlimitdesc
ch <- hostGPUUtilizationdesc
Expand Down Expand Up @@ -288,6 +323,10 @@ func (cc ClusterManagerCollector) collectGPUDeviceMetrics(ch chan<- prometheus.M
return err
}

if err := cc.collectGPUHealthMetrics(ch, hdev, index); err != nil {
return err
}

return nil
}

Expand Down Expand Up @@ -320,6 +359,13 @@ func (cc ClusterManagerCollector) collectGPUMemoryMetrics(ch chan<- prometheus.M
fmt.Sprint(index), uuid, deviceName,
)

ch <- prometheus.MustNewConstMetric(
hostGPUMemoryTotalDesc,
prometheus.GaugeValue,
float64(memory.Total),
fmt.Sprint(index), uuid, deviceName,
)

sendLegacyMetric(ch, legacyHostGPUdesc, prometheus.GaugeValue, float64(memory.Used),
fmt.Sprint(index), uuid, deviceName,
)
Expand Down Expand Up @@ -352,13 +398,71 @@ func (cc ClusterManagerCollector) collectGPUUtilizationMetrics(ch chan<- prometh
fmt.Sprint(index), uuid, deviceName,
)

ch <- prometheus.MustNewConstMetric(
hostGPUMemoryUtilizationDesc,
prometheus.GaugeValue,
float64(util.Memory),
fmt.Sprint(index), uuid, deviceName,
)

sendLegacyMetric(ch, legacyHostGPUUtilizationdesc, prometheus.GaugeValue, float64(util.Gpu),
fmt.Sprint(index), uuid, deviceName,
)

return nil
}

func (cc ClusterManagerCollector) collectGPUHealthMetrics(ch chan<- prometheus.Metric, hdev nvml.Device, index int) error {
uuid, nvret := hdev.GetUUID()
if nvret != nvml.SUCCESS {
return fmt.Errorf("nvml GetUUID err: %s", nvml.ErrorString(nvret))
}

deviceName, nvret := hdev.GetName()
if nvret != nvml.SUCCESS {
return fmt.Errorf("nvml GetName err: %s", nvml.ErrorString(nvret))
}
deviceName = "NVIDIA-" + deviceName

temp, ret := hdev.GetTemperature(nvml.TEMPERATURE_GPU)
if ret == nvml.SUCCESS {
ch <- prometheus.MustNewConstMetric(
hostGPUTemperatureDesc,
prometheus.GaugeValue,
float64(temp),
fmt.Sprint(index), uuid, deviceName,
)
} else if ret != nvml.ERROR_NOT_SUPPORTED {
klog.V(4).Infof("nvml get temperature error for device %d, ret=%d", index, ret)
}

power, ret := hdev.GetPowerUsage()
if ret == nvml.SUCCESS {
ch <- prometheus.MustNewConstMetric(
hostGPUPowerUsageDesc,
prometheus.GaugeValue,
float64(power),
fmt.Sprint(index), uuid, deviceName,
)
} else if ret != nvml.ERROR_NOT_SUPPORTED {
klog.V(4).Infof("nvml get power usage error for device %d, ret=%d", index, ret)
}

limit, ret := hdev.GetPowerManagementLimit()
if ret == nvml.SUCCESS {
ch <- prometheus.MustNewConstMetric(
hostGPUPowerLimitDesc,
prometheus.GaugeValue,
float64(limit),
fmt.Sprint(index), uuid, deviceName,
)
} else if ret != nvml.ERROR_NOT_SUPPORTED {
klog.V(4).Infof("nvml get power limit error for device %d, ret=%d", index, ret)
}

return nil
}

func (cc ClusterManagerCollector) collectPodAndContainerInfo(ch chan<- prometheus.Metric) error {
nodeName := os.Getenv(util.NodeNameEnvName)
if nodeName == "" {
Expand Down
Loading