Add Linux mdraid health monitoring (#1750)

agent/emmc_linux.go

@@ -199,19 +199,6 @@ func readHexByteFile(path string) (uint8, bool) {
 	return b, ok
 }
 
-func readStringFile(path string) string {
-	content, _ := readStringFileOK(path)
-	return content
-}
-
-func readStringFileOK(path string) (string, bool) {
-	b, err := os.ReadFile(path)
-	if err != nil {
-		return "", false
-	}
-	return strings.TrimSpace(string(b)), true
-}
-
 func hasEmmcHealthFiles(deviceDir string) bool {
 	entries, err := os.ReadDir(deviceDir)
 	if err != nil {

agent/file_utils.go

@@ -0,0 +1,24 @@
+package agent
+
+import (
+	"os"
+	"strings"
+)
+
+func readStringFile(path string) string {
+	content, _ := readStringFileOK(path)
+	return content
+}
+
+func readStringFileOK(path string) (string, bool) {
+	b, err := os.ReadFile(path)
+	if err != nil {
+		return "", false
+	}
+	return strings.TrimSpace(string(b)), true
+}
+
+func fileExists(path string) bool {
+	_, err := os.Stat(path)
+	return err == nil
+}

agent/mdraid_linux.go

@@ -0,0 +1,239 @@
+//go:build linux
+
+package agent
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"github.com/henrygd/beszel/internal/entities/smart"
+)
+
+// mdraidSysfsRoot is a test hook; production value is "/sys".
+var mdraidSysfsRoot = "/sys"
+
+type mdraidHealth struct {
+	name          string
+	level         string
+	arrayState    string
+	degraded      uint64
+	raidDisks     uint64
+	syncAction    string
+	syncCompleted string
+	syncSpeed     string
+	mismatchCnt   uint64
+	capacity      uint64
+}
+
+func scanMdraidDevices() []*DeviceInfo {
+	blockDir := filepath.Join(mdraidSysfsRoot, "block")
+	entries, err := os.ReadDir(blockDir)
+	if err != nil {
+		return nil
+	}
+
+	devices := make([]*DeviceInfo, 0, 2)
+	for _, ent := range entries {
+		name := ent.Name()
+		if !isMdraidBlockName(name) {
+			continue
+		}
+		mdDir := filepath.Join(blockDir, name, "md")
+		if !fileExists(filepath.Join(mdDir, "array_state")) {
+			continue
+		}
+
+		devPath := filepath.Join("/dev", name)
+		devices = append(devices, &DeviceInfo{
+			Name:     devPath,
+			Type:     "mdraid",
+			InfoName: devPath + " [mdraid]",
+			Protocol: "MD",
+		})
+	}
+
+	return devices
+}
+
+func (sm *SmartManager) collectMdraidHealth(deviceInfo *DeviceInfo) (bool, error) {
+	if deviceInfo == nil || deviceInfo.Name == "" {
+		return false, nil
+	}
+
+	base := filepath.Base(deviceInfo.Name)
+	if !isMdraidBlockName(base) && !strings.EqualFold(deviceInfo.Type, "mdraid") {
+		return false, nil
+	}
+
+	health, ok := readMdraidHealth(base)
+	if !ok {
+		return false, nil
+	}
+
+	deviceInfo.Type = "mdraid"
+	key := fmt.Sprintf("mdraid:%s", base)
+	status := mdraidSmartStatus(health)
+
+	attrs := make([]*smart.SmartAttribute, 0, 10)
+	if health.arrayState != "" {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "ArrayState", RawString: health.arrayState})
+	}
+	if health.level != "" {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "RaidLevel", RawString: health.level})
+	}
+	if health.raidDisks > 0 {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "RaidDisks", RawValue: health.raidDisks})
+	}
+	if health.degraded > 0 {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "Degraded", RawValue: health.degraded})
+	}
+	if health.syncAction != "" {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "SyncAction", RawString: health.syncAction})
+	}
+	if health.syncCompleted != "" {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "SyncCompleted", RawString: health.syncCompleted})
+	}
+	if health.syncSpeed != "" {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "SyncSpeed", RawString: health.syncSpeed})
+	}
+	if health.mismatchCnt > 0 {
+		attrs = append(attrs, &smart.SmartAttribute{Name: "MismatchCount", RawValue: health.mismatchCnt})
+	}
+
+	sm.Lock()
+	defer sm.Unlock()
+
+	if _, exists := sm.SmartDataMap[key]; !exists {
+		sm.SmartDataMap[key] = &smart.SmartData{}
+	}
+
+	data := sm.SmartDataMap[key]
+	data.ModelName = "Linux MD RAID"
+	if health.level != "" {
+		data.ModelName = "Linux MD RAID (" + health.level + ")"
+	}
+	data.SerialNumber = ""
+	data.FirmwareVersion = ""
+	data.Capacity = health.capacity
+	data.Temperature = 0
+	data.SmartStatus = status
+	data.DiskName = filepath.Join("/dev", base)
+	data.DiskType = "mdraid"
+	data.Attributes = attrs
+	sm.SmartDataMap[key] = data
+
+	return true, nil
+}
+
+func readMdraidHealth(blockName string) (mdraidHealth, bool) {
+	var out mdraidHealth
+
+	if !isMdraidBlockName(blockName) {
+		return out, false
+	}
+
+	mdDir := filepath.Join(mdraidSysfsRoot, "block", blockName, "md")
+	arrayState, okState := readStringFileOK(filepath.Join(mdDir, "array_state"))
+	if !okState {
+		return out, false
+	}
+
+	out.name = blockName
+	out.arrayState = arrayState
+	out.level = readStringFile(filepath.Join(mdDir, "level"))
+	out.syncAction = readStringFile(filepath.Join(mdDir, "sync_action"))
+	out.syncCompleted = readStringFile(filepath.Join(mdDir, "sync_completed"))
+	out.syncSpeed = readStringFile(filepath.Join(mdDir, "sync_speed"))
+
+	if val, ok := readUintFile(filepath.Join(mdDir, "raid_disks")); ok {
+		out.raidDisks = val
+	}
+	if val, ok := readUintFile(filepath.Join(mdDir, "degraded")); ok {
+		out.degraded = val
+	}
+	if val, ok := readUintFile(filepath.Join(mdDir, "mismatch_cnt")); ok {
+		out.mismatchCnt = val
+	}
+
+	if capBytes, ok := readMdraidBlockCapacityBytes(blockName, mdraidSysfsRoot); ok {
+		out.capacity = capBytes
+	}
+
+	return out, true
+}
+
+func mdraidSmartStatus(health mdraidHealth) string {
+	state := strings.ToLower(strings.TrimSpace(health.arrayState))
+	switch state {
+	case "inactive", "faulty", "broken", "stopped":
+		return "FAILED"
+	}
+	if health.degraded > 0 {
+		return "FAILED"
+	}
+
+	switch strings.ToLower(strings.TrimSpace(health.syncAction)) {
+	case "resync", "recover", "reshape", "check", "repair":
+		return "WARNING"
+	}
+
+	if state == "clean" || state == "active" || state == "readonly" {
+		return "PASSED"
+	}
+	return "UNKNOWN"
+}
+
+func isMdraidBlockName(name string) bool {
+	if !strings.HasPrefix(name, "md") {
+		return false
+	}
+	suffix := strings.TrimPrefix(name, "md")
+	if suffix == "" {
+		return false
+	}
+	for _, c := range suffix {
+		if c < '0' || c > '9' {
+			return false
+		}
+	}
+	return true
+}
+
+func readUintFile(path string) (uint64, bool) {
+	raw, ok := readStringFileOK(path)
+	if !ok {
+		return 0, false
+	}
+	parsed, err := strconv.ParseUint(strings.TrimSpace(raw), 10, 64)
+	if err != nil {
+		return 0, false
+	}
+	return parsed, true
+}
+
+func readMdraidBlockCapacityBytes(blockName, root string) (uint64, bool) {
+	sizePath := filepath.Join(root, "block", blockName, "size")
+	lbsPath := filepath.Join(root, "block", blockName, "queue", "logical_block_size")
+
+	sizeStr, ok := readStringFileOK(sizePath)
+	if !ok {
+		return 0, false
+	}
+	sectors, err := strconv.ParseUint(strings.TrimSpace(sizeStr), 10, 64)
+	if err != nil || sectors == 0 {
+		return 0, false
+	}
+
+	lbsStr, ok := readStringFileOK(lbsPath)
+	logicalBlockSize := uint64(512)
+	if ok {
+		if parsed, err := strconv.ParseUint(strings.TrimSpace(lbsStr), 10, 64); err == nil && parsed > 0 {
+			logicalBlockSize = parsed
+		}
+	}
+
+	return sectors * logicalBlockSize, true
+}

agent/mdraid_linux_test.go

@@ -0,0 +1,100 @@
+//go:build linux
+
+package agent
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/henrygd/beszel/internal/entities/smart"
+)
+
+func TestMdraidMockSysfsScanAndCollect(t *testing.T) {
+	tmp := t.TempDir()
+	prev := mdraidSysfsRoot
+	mdraidSysfsRoot = tmp
+	t.Cleanup(func() { mdraidSysfsRoot = prev })
+
+	mdDir := filepath.Join(tmp, "block", "md0", "md")
+	queueDir := filepath.Join(tmp, "block", "md0", "queue")
+	if err := os.MkdirAll(mdDir, 0o755); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.MkdirAll(queueDir, 0o755); err != nil {
+		t.Fatal(err)
+	}
+
+	write := func(path, content string) {
+		t.Helper()
+		if err := os.WriteFile(path, []byte(content), 0o644); err != nil {
+			t.Fatal(err)
+		}
+	}
+
+	write(filepath.Join(mdDir, "array_state"), "active\n")
+	write(filepath.Join(mdDir, "level"), "raid1\n")
+	write(filepath.Join(mdDir, "raid_disks"), "2\n")
+	write(filepath.Join(mdDir, "degraded"), "0\n")
+	write(filepath.Join(mdDir, "sync_action"), "resync\n")
+	write(filepath.Join(mdDir, "sync_completed"), "10%\n")
+	write(filepath.Join(mdDir, "sync_speed"), "100M\n")
+	write(filepath.Join(mdDir, "mismatch_cnt"), "0\n")
+	write(filepath.Join(queueDir, "logical_block_size"), "512\n")
+	write(filepath.Join(tmp, "block", "md0", "size"), "2048\n")
+
+	devs := scanMdraidDevices()
+	if len(devs) != 1 {
+		t.Fatalf("scanMdraidDevices() = %d devices, want 1", len(devs))
+	}
+	if devs[0].Name != "/dev/md0" || devs[0].Type != "mdraid" {
+		t.Fatalf("scanMdraidDevices()[0] = %+v, want Name=/dev/md0 Type=mdraid", devs[0])
+	}
+
+	sm := &SmartManager{SmartDataMap: map[string]*smart.SmartData{}}
+	ok, err := sm.collectMdraidHealth(devs[0])
+	if err != nil || !ok {
+		t.Fatalf("collectMdraidHealth() = (ok=%v, err=%v), want (true,nil)", ok, err)
+	}
+	if len(sm.SmartDataMap) != 1 {
+		t.Fatalf("SmartDataMap len=%d, want 1", len(sm.SmartDataMap))
+	}
+	var got *smart.SmartData
+	for _, v := range sm.SmartDataMap {
+		got = v
+		break
+	}
+	if got == nil {
+		t.Fatalf("SmartDataMap value nil")
+	}
+	if got.DiskType != "mdraid" || got.DiskName != "/dev/md0" {
+		t.Fatalf("disk fields = (type=%q name=%q), want (mdraid,/dev/md0)", got.DiskType, got.DiskName)
+	}
+	if got.SmartStatus != "WARNING" {
+		t.Fatalf("SmartStatus=%q, want WARNING", got.SmartStatus)
+	}
+	if got.ModelName == "" || got.Capacity == 0 {
+		t.Fatalf("identity fields = (model=%q cap=%d), want non-empty model and cap>0", got.ModelName, got.Capacity)
+	}
+	if len(got.Attributes) < 5 {
+		t.Fatalf("attributes len=%d, want >= 5", len(got.Attributes))
+	}
+}
+
+func TestMdraidSmartStatus(t *testing.T) {
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "inactive"}); got != "FAILED" {
+		t.Fatalf("mdraidSmartStatus(inactive) = %q, want FAILED", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "active", degraded: 1}); got != "FAILED" {
+		t.Fatalf("mdraidSmartStatus(degraded) = %q, want FAILED", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "active", syncAction: "recover"}); got != "WARNING" {
+		t.Fatalf("mdraidSmartStatus(recover) = %q, want WARNING", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean"}); got != "PASSED" {
+		t.Fatalf("mdraidSmartStatus(clean) = %q, want PASSED", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "unknown"}); got != "UNKNOWN" {
+		t.Fatalf("mdraidSmartStatus(unknown) = %q, want UNKNOWN", got)
+	}
+}

agent/mdraid_stub.go

@@ -0,0 +1,11 @@
+//go:build !linux
+
+package agent
+
+func scanMdraidDevices() []*DeviceInfo {
+	return nil
+}
+
+func (sm *SmartManager) collectMdraidHealth(deviceInfo *DeviceInfo) (bool, error) {
+	return false, nil
+}

agent/smart.go

@@ -199,6 +199,13 @@ func (sm *SmartManager) ScanDevices(force bool) error {
 		hasValidScan = true
 	}
 
+	// Add Linux mdraid arrays by reading sysfs health fields. This does not
+	// require smartctl and does not scan the whole device.
+	if raidDevices := scanMdraidDevices(); len(raidDevices) > 0 {
+		scannedDevices = append(scannedDevices, raidDevices...)
+		hasValidScan = true
+	}
+
 	finalDevices := mergeDeviceLists(currentDevices, scannedDevices, configuredDevices)
 	finalDevices = sm.filterExcludedDevices(finalDevices)
 	sm.updateSmartDevices(finalDevices)
@@ -450,6 +457,12 @@ func (sm *SmartManager) CollectSmart(deviceInfo *DeviceInfo) error {
 		return errNoValidSmartData
 	}
 
+	// mdraid health is not exposed via SMART; Linux exposes array state in sysfs.
+	if deviceInfo != nil {
+		if ok, err := sm.collectMdraidHealth(deviceInfo); ok {
+			return err
+		}
+	}
 	// eMMC health is not exposed via SMART on Linux, but the kernel provides
 	// wear / EOL indicators via sysfs. Prefer that path when available.
 	if deviceInfo != nil {
@@ -1146,9 +1159,11 @@ func NewSmartManager() (*SmartManager, error) {
 	slog.Debug("smartctl", "path", path, "err", err)
 	if err != nil {
 		// Keep the previous fail-fast behavior unless this Linux host exposes
-		// eMMC health via sysfs, in which case smartctl is optional.
-		if runtime.GOOS == "linux" && len(scanEmmcDevices()) > 0 {
-			return sm, nil
+		// eMMC or mdraid health via sysfs, in which case smartctl is optional.
+		if runtime.GOOS == "linux" {
+			if len(scanEmmcDevices()) > 0 || len(scanMdraidDevices()) > 0 {
+				return sm, nil
+			}
 		}
 		return nil, err
 	}

readme.md

@@ -51,7 +51,7 @@ The [quick start guide](https://beszel.dev/guide/getting-started) and other docu
 - **GPU usage / power draw** - Nvidia, AMD, and Intel.
 - **Battery** - Host system battery charge.
 - **Containers** - Status and metrics of all running Docker / Podman containers.
-- **S.M.A.R.T.** - Host system disk health (includes eMMC wear/EOL via Linux sysfs when available).
+- **S.M.A.R.T.** - Host system disk health (includes eMMC wear/EOL and Linux mdraid array health via sysfs when available).
 
 ## Help and discussion