fix(agent): claim sensor names before filtering so excludes don't rename (#2645)

updateTemperatures detected duplicate SensorKeys by looking at the output
map, which only holds sensors that passed every filter. A sensor dropped by
SENSORS or SKIP_GPU therefore never claimed its name, so excluding the first
of two same-named sensors renamed the second one from nvme_sensor_2_4 to
nvme_sensor_2 and dropped it too, making it impossible to exclude just one.

Track handed-out names in a separate set that is populated before any filter
runs, so a sensor's key depends only on the hardware list.

agent/sensors.go

@@ -151,12 +151,24 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 	}
 
 	systemStats.Temperatures = make(map[string]float64, len(temps))
+	// claimed tracks every name handed out, including names of sensors that are
+	// dropped below, so a sensor's key never depends on which sensors were filtered
+	claimed := make(map[string]struct{}, len(temps))
 	for i, sensor := range temps {
 		// check for malformed strings on darwin (gopsutil/issues/1832)
 		if runtime.GOOS == "darwin" && !utf8.ValidString(sensor.SensorKey) {
 			continue
 		}
 
+		// claim a name before any filtering, so excluding a sensor does not
+		// rename the sensors after it
+		sensorName := sensor.SensorKey
+		if _, ok := claimed[sensorName]; ok {
+			// if key already claimed, append int to key
+			sensorName = sensorName + "_" + strconv.Itoa(i)
+		}
+		claimed[sensorName] = struct{}{}
+
 		// scale temperature
 		if sensor.Temperature != 0 && sensor.Temperature < 1 {
 			sensor.Temperature = scaleTemperature(sensor.Temperature)
@@ -165,11 +177,6 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 		if sensor.Temperature <= 0 || sensor.Temperature >= 200 {
 			continue
 		}
-		sensorName := sensor.SensorKey
-		if _, ok := systemStats.Temperatures[sensorName]; ok {
-			// if key already exists, append int to key
-			sensorName = sensorName + "_" + strconv.Itoa(i)
-		}
 		// skip if not in whitelist or blacklist
 		if !isValidSensor(sensorName, a.sensorConfig) {
 			continue

agent/sensors_test.go

@@ -798,3 +798,51 @@ func TestUpdateTemperaturesSkipGpu(t *testing.T) {
 	newAgent(false).updateTemperatures(stats)
 	assert.Len(t, stats.Temperatures, 2)
 }
+
+func TestUpdateTemperaturesExcludedSensorKeepsNames(t *testing.T) {
+	originalGetSensorTemps := getSensorTemps
+	t.Cleanup(func() {
+		getSensorTemps = originalGetSensorTemps
+	})
+	// two drives that report the same SensorKey, plus an unrelated sensor
+	getSensorTemps = func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+		return []sensors.TemperatureStat{
+			{SensorKey: "nvme_sensor_2", Temperature: 74.85},
+			{SensorKey: "nvme_sensor_2", Temperature: 41},
+			{SensorKey: "coretemp_core_0", Temperature: 55},
+		}, nil
+	}
+
+	newAgent := func(exclude string) *Agent {
+		config := map[string]struct{}{}
+		if exclude != "" {
+			config[exclude] = struct{}{}
+		}
+		return &Agent{
+			systemInfo: system.Info{},
+			sensorConfig: &SensorConfig{
+				context:     context.Background(),
+				timeout:     2 * time.Second,
+				sensors:     config,
+				isBlacklist: exclude != "",
+			},
+		}
+	}
+
+	// unfiltered: the second duplicate is suffixed with its index
+	stats := &system.Stats{}
+	newAgent("").updateTemperatures(stats)
+	assert.Equal(t, map[string]float64{
+		"nvme_sensor_2":   74.85,
+		"nvme_sensor_2_1": 41,
+		"coretemp_core_0": 55,
+	}, stats.Temperatures)
+
+	// excluding the first duplicate keeps the second one's name
+	stats = &system.Stats{}
+	newAgent("nvme_sensor_2").updateTemperatures(stats)
+	assert.Equal(t, map[string]float64{
+		"nvme_sensor_2_1": 41,
+		"coretemp_core_0": 55,
+	}, stats.Temperatures)
+}