S.M.A.R.T support

beszel/internal/agent/agent.go

@@ -27,6 +27,7 @@ type Agent struct {
 	sensorsWhitelist map[string]struct{}        // List of sensors to monitor
 	systemInfo       system.Info                // Host system info
 	gpuManager       *GPUManager                // Manages GPU data
+	smartManager     *SmartManager              // Manages SMART data
 }
 
 func NewAgent() *Agent {
@@ -83,6 +84,12 @@ func NewAgent() *Agent {
 		agent.gpuManager = gm
 	}
 
+	if sm, err := NewSmartManager(); err != nil {
+		slog.Debug("SMART", "err", err)
+	} else {
+		agent.smartManager = sm
+	}
+
 	// if debugging, print stats
 	if agent.debug {
 		slog.Debug("Stats", "data", agent.gatherStats())

beszel/internal/agent/smart.go

@@ -0,0 +1,288 @@
+package agent
+
+import (
+	"beszel/internal/entities/smart"
+	"beszel/internal/entities/system"
+	"context"
+	"encoding/json"
+	"fmt"
+	"os/exec"
+	"reflect"
+	"sync"
+	"time"
+
+	"golang.org/x/exp/slog"
+)
+
+// SmartManager manages data collection for SMART devices
+type SmartManager struct {
+	SmartDataMap map[string]*system.SmartData
+	SmartDevices []*DeviceInfo
+	mutex        sync.Mutex
+}
+
+type scanOutput struct {
+	Devices []struct {
+		Name     string `json:"name"`
+		Type     string `json:"type"`
+		InfoName string `json:"info_name"`
+		Protocol string `json:"protocol"`
+	} `json:"devices"`
+}
+
+type DeviceInfo struct {
+	Name     string `json:"name"`
+	Type     string `json:"type"`
+	InfoName string `json:"info_name"`
+	Protocol string `json:"protocol"`
+}
+
+var errNoValidSmartData = fmt.Errorf("no valid GPU data found") // Error for missing data
+
+// Starts the SmartManager
+func (sm *SmartManager) Start() {
+	sm.SmartDataMap = make(map[string]*system.SmartData)
+	for {
+		sm.ScanDevices()
+		// TODO: add retry logic
+		for _, deviceInfo := range sm.SmartDevices {
+			err := sm.CollectSmart(deviceInfo)
+			if err != nil {
+				slog.Warn("smartctl failed, stopping", "err", err)
+				return
+			}
+		}
+	}
+}
+
+// GetCurrentData returns the current SMART data
+func (sm *SmartManager) GetCurrentData() map[string]system.SmartData {
+	sm.mutex.Lock()
+	defer sm.mutex.Unlock()
+	result := make(map[string]system.SmartData)
+	for key, value := range sm.SmartDataMap {
+		result[key] = *value
+	}
+	return result
+}
+
+// ScanDevices scans for SMART devices
+func (sm *SmartManager) ScanDevices() error {
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	defer cancel()
+
+	cmd := exec.CommandContext(ctx, "smartctl", "--scan", "-j")
+	output, err := cmd.Output()
+
+	if err != nil {
+		return err
+	}
+
+	hasValidData := sm.parseScan(output)
+	if !hasValidData {
+		return errNoValidSmartData
+	}
+	return nil
+}
+
+// CollectSmart collects SMART data for a device
+func (sm *SmartManager) CollectSmart(deviceInfo *DeviceInfo) error {
+	ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+	defer cancel()
+
+	cmd := exec.CommandContext(ctx, "smartctl", "--all", "-j", deviceInfo.Name)
+
+	output, err := cmd.Output()
+
+	if err != nil {
+		return err
+	}
+
+	hasValidData := false
+	if deviceInfo.Type == "scsi" {
+		hasValidData = sm.parseSmartForScsi(output)
+	} else if deviceInfo.Type == "nvme" {
+		hasValidData = sm.parseSmartForNvme(output)
+	}
+
+	if !hasValidData {
+		return errNoValidSmartData
+	}
+	return nil
+}
+
+// parseScan parses the output of smartctl --scan -j and updates the SmartDevices slice
+func (sm *SmartManager) parseScan(output []byte) bool {
+	sm.mutex.Lock()
+	defer sm.mutex.Unlock()
+
+	sm.SmartDevices = make([]*DeviceInfo, 0)
+	scan := &scanOutput{}
+
+	if err := json.Unmarshal(output, scan); err != nil {
+		fmt.Printf("Failed to parse JSON: %v\n", err)
+		return false
+	}
+
+	scannedDeviceNameMap := make(map[string]bool)
+
+	for _, device := range scan.Devices {
+		deviceInfo := &DeviceInfo{
+			Name:     device.Name,
+			Type:     device.Type,
+			InfoName: device.InfoName,
+			Protocol: device.Protocol,
+		}
+		sm.SmartDevices = append(sm.SmartDevices, deviceInfo)
+		scannedDeviceNameMap[device.Name] = true
+	}
+	// remove devices that are not in the scan
+	for key := range sm.SmartDataMap {
+		if _, ok := scannedDeviceNameMap[key]; !ok {
+			delete(sm.SmartDataMap, key)
+		}
+	}
+	devicesString := ""
+	for _, device := range sm.SmartDevices {
+		devicesString += device.Name + " "
+	}
+
+	return true
+}
+
+// parseSmartForScsi parses the output of smartctl --all -j /dev/sdX and updates the SmartDataMap
+func (sm *SmartManager) parseSmartForScsi(output []byte) bool {
+	data := &smart.SmartInfoForSata{}
+
+	if err := json.Unmarshal(output, &data); err != nil {
+		return false
+	}
+
+	sm.mutex.Lock()
+	defer sm.mutex.Unlock()
+
+	// get device name (e.g. /dev/sda)
+	deviceName := data.Device.Name
+
+	// if device does not exist in SmartDataMap, initialize it
+	if _, ok := sm.SmartDataMap[deviceName]; !ok {
+		sm.SmartDataMap[deviceName] = &system.SmartData{}
+	}
+
+	// update SmartData
+	smartData := sm.SmartDataMap[deviceName]
+	smartData.ModelFamily = data.ModelFamily
+	smartData.ModelName = data.ModelName
+	smartData.SerialNumber = data.SerialNumber
+	smartData.FirmwareVersion = data.FirmwareVersion
+	smartData.Capacity = data.UserCapacity.Bytes
+	if data.SmartStatus.Passed {
+		smartData.SmartStatus = "PASSED"
+	} else {
+		smartData.SmartStatus = "FAILED"
+	}
+	smartData.DiskName = deviceName
+	smartData.DiskType = data.Device.Type
+
+	// update SmartAttributes
+	smartData.Attributes = make([]*system.SmartAttribute, 0, len(data.AtaSmartAttributes.Table))
+	for _, attr := range data.AtaSmartAttributes.Table {
+		smartAttr := &system.SmartAttribute{
+			Id:         attr.ID,
+			Name:       attr.Name,
+			Value:      attr.Value,
+			Worst:      attr.Worst,
+			Threshold:  attr.Thresh,
+			RawValue:   attr.Raw.Value,
+			RawString:  attr.Raw.String,
+			Flags:      attr.Flags.String,
+			WhenFailed: attr.WhenFailed,
+		}
+		smartData.Attributes = append(smartData.Attributes, smartAttr)
+	}
+	smartData.Temperature = data.Temperature.Current
+	sm.SmartDataMap[deviceName] = smartData
+
+	return true
+}
+
+// parseSmartForNvme parses the output of smartctl --all -j /dev/nvmeX and updates the SmartDataMap
+func (sm *SmartManager) parseSmartForNvme(output []byte) bool {
+	data := &smart.SmartInfoForNvme{}
+
+	if err := json.Unmarshal(output, &data); err != nil {
+		return false
+	}
+
+	sm.mutex.Lock()
+	defer sm.mutex.Unlock()
+
+	// get device name (e.g. /dev/nvme0)
+	deviceName := data.Device.Name
+
+	// if device does not exist in SmartDataMap, initialize it
+	if _, ok := sm.SmartDataMap[deviceName]; !ok {
+		sm.SmartDataMap[deviceName] = &system.SmartData{}
+	}
+
+	// update SmartData
+	smartData := sm.SmartDataMap[deviceName]
+	smartData.ModelName = data.ModelName
+	smartData.SerialNumber = data.SerialNumber
+	smartData.FirmwareVersion = data.FirmwareVersion
+	smartData.Capacity = data.UserCapacity.Bytes
+	if data.SmartStatus.Passed {
+		smartData.SmartStatus = "PASSED"
+	} else {
+		smartData.SmartStatus = "FAILED"
+	}
+	smartData.DiskName = deviceName
+	smartData.DiskType = data.Device.Type
+
+	v := reflect.ValueOf(data.NVMeSmartHealthInformationLog)
+	t := v.Type()
+	smartData.Attributes = make([]*system.SmartAttribute, 0, v.NumField())
+
+	// nvme attributes does not follow the same format as ata attributes,
+	// so we have to manually iterate over the fields abd update SmartAttributes
+	for i := 0; i < v.NumField(); i++ {
+		field := t.Field(i)
+		value := v.Field(i)
+		key := field.Name
+		val := value.Interface()
+		// drop non int values
+		if _, ok := val.(int); !ok {
+			continue
+		}
+		smartAttr := &system.SmartAttribute{
+			Name:  key,
+			Value: val.(int),
+		}
+		smartData.Attributes = append(smartData.Attributes, smartAttr)
+	}
+	smartData.Temperature = data.NVMeSmartHealthInformationLog.Temperature
+
+	sm.SmartDataMap[deviceName] = smartData
+
+	return true
+}
+
+// detectSmartctl checks if smartctl is installed, returns an error if not
+func (sm *SmartManager) detectSmartctl() error {
+	if _, err := exec.LookPath("smartctl"); err == nil {
+		return nil
+	}
+	return fmt.Errorf("no smartctl found - install smartctl")
+}
+
+// NewGPUManager creates and initializes a new GPUManager
+func NewSmartManager() (*SmartManager, error) {
+	var sm SmartManager
+	if err := sm.detectSmartctl(); err != nil {
+		return nil, err
+	}
+
+	go sm.Start()
+
+	return &sm, nil
+}

beszel/internal/agent/system.go

@@ -211,6 +211,17 @@ func (a *Agent) getSystemStats() system.Stats {
 			}
 		}
 	}
+	if a.smartManager != nil {
+		if smartData := a.smartManager.GetCurrentData(); len(smartData) > 0 {
+			systemStats.SmartData = smartData
+			if systemStats.Temperatures == nil {
+				systemStats.Temperatures = make(map[string]float64, len(a.smartManager.SmartDataMap))
+			}
+			for key, value := range a.smartManager.SmartDataMap {
+				systemStats.Temperatures[key] = float64(value.Temperature)
+			}
+		}
+	}
 
 	// update base system info
 	a.systemInfo.Cpu = systemStats.Cpu