Globalize autotuned system power source

audit/internal/platform/sat_fan_stress.go

@@ -43,17 +43,22 @@ type GPUStressMetric struct {
 
 // FanStressRow is one second-interval telemetry sample covering all monitored dimensions.
 type FanStressRow struct {
-	TimestampUTC string
-	ElapsedSec   float64
-	Phase        string // "baseline", "load1", "pause", "load2", "cooldown"
-	GPUs         []GPUStressMetric
-	Fans         []FanReading
-	CPUMaxTempC  float64 // highest CPU temperature from ipmitool / sensors
-	SysPowerW    float64 // DCMI system power reading
+	TimestampUTC   string
+	ElapsedSec     float64
+	Phase          string // "baseline", "load1", "pause", "load2", "cooldown"
+	GPUs           []GPUStressMetric
+	Fans           []FanReading
+	CPUMaxTempC    float64 // highest CPU temperature from ipmitool / sensors
+	SysPowerW      float64
+	SysPowerSource string
+	SysPowerMode   string
 }
 
 type cachedPowerReading struct {
 	Value     float64
+	Source    string
+	Mode      string
+	Reason    string
 	UpdatedAt time.Time
 }
 
@@ -278,7 +283,7 @@ func sampleFanStressRow(gpuIndices []int, phase string, elapsed float64) FanStre
 	row.GPUs = sampleGPUStressMetrics(gpuIndices)
 	row.Fans, _ = sampleFanSpeeds()
 	row.CPUMaxTempC = sampleCPUMaxTemp()
-	row.SysPowerW = sampleSystemPower()
+	row.SysPowerW, row.SysPowerSource, row.SysPowerMode = sampleSystemPowerResolved()
 	return row
 }
 
@@ -763,19 +768,19 @@ func sampleCPUTempViaSensors() float64 {
 	return max
 }
 
-// sampleSystemPower reads system power draw via DCMI.
-func sampleSystemPower() float64 {
+// sampleSystemPowerResolved reads system power via the global autotune source,
+// falling back to the historical heuristic before autotune or when degraded.
+func sampleSystemPowerResolved() (float64, string, string) {
 	now := time.Now()
-	current := 0.0
-	out, err := exec.Command("ipmitool", "dcmi", "power", "reading").Output()
-	if err == nil {
-		current = parseDCMIPowerReading(string(out))
-	}
+	current, decision, err := SampleSystemPowerResolved("")
 	systemPowerCacheMu.Lock()
 	defer systemPowerCacheMu.Unlock()
-	value, updated := effectiveSystemPowerReading(systemPowerCache, current, now)
+	if err != nil {
+		current = 0
+	}
+	value, updated := effectiveSystemPowerReading(systemPowerCache, current, decision.EffectiveSource, decision.Mode, decision.Reason, now)
 	systemPowerCache = updated
-	return value
+	return value, updated.Source, updated.Mode
 }
 
 // parseDCMIPowerReading extracts the instantaneous power reading from ipmitool dcmi output.
@@ -798,9 +803,9 @@ func parseDCMIPowerReading(raw string) float64 {
 	return 0
 }
 
-func effectiveSystemPowerReading(cache cachedPowerReading, current float64, now time.Time) (float64, cachedPowerReading) {
+func effectiveSystemPowerReading(cache cachedPowerReading, current float64, source, mode, reason string, now time.Time) (float64, cachedPowerReading) {
 	if current > 0 {
-		cache = cachedPowerReading{Value: current, UpdatedAt: now}
+		cache = cachedPowerReading{Value: current, Source: source, Mode: mode, Reason: reason, UpdatedAt: now}
 		return current, cache
 	}
 	if cache.Value > 0 && !cache.UpdatedAt.IsZero() && now.Sub(cache.UpdatedAt) <= systemPowerHoldTTL {