fix: Use historical records to average disk usage for extra disk alerts (#1801)

- Introduced a new test file `alerts_disk_test.go` to validate the behavior of disk alerts using historical data for extra filesystems.
- Enhanced the `HandleSystemAlerts` function to correctly calculate disk usage for extra filesystems based on historical records.
- Updated the `SystemAlertStats` struct to include `ExtraFs` for tracking additional filesystem statistics.

agent/gpu_amd_linux.go

@@ -33,8 +33,8 @@ func (gm *GPUManager) hasAmdSysfs() bool {
 		return false
 	}
 	for _, vendorPath := range cards {
-		vendor, err := os.ReadFile(vendorPath)
-		if err == nil && strings.TrimSpace(string(vendor)) == "0x1002" {
+		vendor, err := utils.ReadStringFileLimited(vendorPath, 64)
+		if err == nil && vendor == "0x1002" {
 			return true
 		}
 	}
@@ -88,12 +88,11 @@ func (gm *GPUManager) collectAmdStats() error {
 
 // isAmdGpu checks whether a DRM card path belongs to AMD vendor ID 0x1002.
 func isAmdGpu(cardPath string) bool {
-	vendorPath := filepath.Join(cardPath, "device/vendor")
-	vendor, err := os.ReadFile(vendorPath)
+	vendor, err := utils.ReadStringFileLimited(filepath.Join(cardPath, "device/vendor"), 64)
 	if err != nil {
 		return false
 	}
-	return strings.TrimSpace(string(vendor)) == "0x1002"
+	return vendor == "0x1002"
 }
 
 // updateAmdGpuData reads GPU metrics from sysfs and updates the GPU data map.
@@ -155,11 +154,11 @@ func (gm *GPUManager) updateAmdGpuData(cardPath string) bool {
 
 // readSysfsFloat reads and parses a numeric value from a sysfs file.
 func readSysfsFloat(path string) (float64, error) {
-	val, err := os.ReadFile(path)
+	val, err := utils.ReadStringFileLimited(path, 64)
 	if err != nil {
 		return 0, err
 	}
-	return strconv.ParseFloat(strings.TrimSpace(string(val)), 64)
+	return strconv.ParseFloat(val, 64)
 }
 
 // normalizeHexID normalizes hex IDs by trimming spaces, lowercasing, and dropping 0x.
@@ -274,16 +273,16 @@ func cacheMissingAmdgpuName(deviceID, revisionID string) {
 // Falls back to showing the raw device ID if not found in the lookup table.
 func getAmdGpuName(devicePath string) string {
 	// Try product_name first (works for some enterprise GPUs)
-	if prod, err := os.ReadFile(filepath.Join(devicePath, "product_name")); err == nil {
-		return strings.TrimSpace(string(prod))
+	if prod, err := utils.ReadStringFileLimited(filepath.Join(devicePath, "product_name"), 128); err == nil {
+		return prod
 	}
 
 	// Read PCI device ID and look it up
-	if deviceID, err := os.ReadFile(filepath.Join(devicePath, "device")); err == nil {
-		id := normalizeHexID(string(deviceID))
+	if deviceID, err := utils.ReadStringFileLimited(filepath.Join(devicePath, "device"), 64); err == nil {
+		id := normalizeHexID(deviceID)
 		revision := ""
-		if revBytes, revErr := os.ReadFile(filepath.Join(devicePath, "revision")); revErr == nil {
-			revision = normalizeHexID(string(revBytes))
+		if rev, revErr := utils.ReadStringFileLimited(filepath.Join(devicePath, "revision"), 64); revErr == nil {
+			revision = normalizeHexID(rev)
 		}
 
 		if name, found, done := getCachedAmdgpuName(id, revision); found {

agent/mdraid_linux.go

@@ -170,11 +170,18 @@ func mdraidSmartStatus(health mdraidHealth) string {
 	case "inactive", "faulty", "broken", "stopped":
 		return "FAILED"
 	}
+	// During rebuild/recovery, arrays are often temporarily degraded; report as
+	// warning instead of hard failure while synchronization is in progress.
+	syncAction := strings.ToLower(strings.TrimSpace(health.syncAction))
+	switch syncAction {
+	case "resync", "recover", "reshape":
+		return "WARNING"
+	}
 	if health.degraded > 0 {
 		return "FAILED"
 	}
-	switch strings.ToLower(strings.TrimSpace(health.syncAction)) {
-	case "resync", "recover", "reshape", "check", "repair":
+	switch syncAction {
+	case "check", "repair":
 		return "WARNING"
 	}
 	switch state {

agent/mdraid_linux_test.go

@@ -85,6 +85,9 @@ 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, syncAction: "recover"}); got != "WARNING" {
+		t.Fatalf("mdraidSmartStatus(degraded+recover) = %q, want WARNING", got)
+	}
 	if got := mdraidSmartStatus(mdraidHealth{arrayState: "active", degraded: 1}); got != "FAILED" {
 		t.Fatalf("mdraidSmartStatus(degraded) = %q, want FAILED", got)
 	}

agent/network.go

@@ -104,10 +104,7 @@ func (a *Agent) initializeNetIoStats() {
 	// get current network I/O stats and record valid interfaces
 	if netIO, err := psutilNet.IOCounters(true); err == nil {
 		for _, v := range netIO {
-			if nicsEnvExists && !isValidNic(v.Name, nicCfg) {
-				continue
-			}
-			if a.skipNetworkInterface(v) {
+			if skipNetworkInterface(v, nicCfg) {
 				continue
 			}
 			slog.Info("Detected network interface", "name", v.Name, "sent", v.BytesSent, "recv", v.BytesRecv)
@@ -243,7 +240,19 @@ func (a *Agent) applyNetworkTotals(
 	a.netIoStats[cacheTimeMs] = nis
 }
 
-func (a *Agent) skipNetworkInterface(v psutilNet.IOCountersStat) bool {
+// skipNetworkInterface returns true if the network interface should be ignored.
+func skipNetworkInterface(v psutilNet.IOCountersStat, nicCfg *NicConfig) bool {
+	if nicCfg != nil {
+		if !isValidNic(v.Name, nicCfg) {
+			return true
+		}
+		// In whitelist mode, we honor explicit inclusion without auto-filtering.
+		if !nicCfg.isBlacklist {
+			return false
+		}
+		// In blacklist mode, still apply the auto-filter below.
+	}
+
 	switch {
 	case strings.HasPrefix(v.Name, "lo"),
 		strings.HasPrefix(v.Name, "docker"),

agent/network_test.go

@@ -261,6 +261,39 @@ func TestNewNicConfig(t *testing.T) {
 		})
 	}
 }
+func TestSkipNetworkInterface(t *testing.T) {
+	tests := []struct {
+		name       string
+		nic        psutilNet.IOCountersStat
+		nicCfg     *NicConfig
+		expectSkip bool
+	}{
+		{"loopback lo", psutilNet.IOCountersStat{Name: "lo", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"loopback lo0", psutilNet.IOCountersStat{Name: "lo0", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"docker prefix", psutilNet.IOCountersStat{Name: "docker0", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"br- prefix", psutilNet.IOCountersStat{Name: "br-lan", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"veth prefix", psutilNet.IOCountersStat{Name: "veth0abc", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"bond prefix", psutilNet.IOCountersStat{Name: "bond0", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"cali prefix", psutilNet.IOCountersStat{Name: "cali1234", BytesSent: 100, BytesRecv: 100}, nil, true},
+		{"zero BytesRecv", psutilNet.IOCountersStat{Name: "eth0", BytesSent: 100, BytesRecv: 0}, nil, true},
+		{"zero BytesSent", psutilNet.IOCountersStat{Name: "eth0", BytesSent: 0, BytesRecv: 100}, nil, true},
+		{"both zero", psutilNet.IOCountersStat{Name: "eth0", BytesSent: 0, BytesRecv: 0}, nil, true},
+		{"normal eth0", psutilNet.IOCountersStat{Name: "eth0", BytesSent: 100, BytesRecv: 200}, nil, false},
+		{"normal wlan0", psutilNet.IOCountersStat{Name: "wlan0", BytesSent: 1, BytesRecv: 1}, nil, false},
+		{"whitelist overrides skip (docker)", psutilNet.IOCountersStat{Name: "docker0", BytesSent: 100, BytesRecv: 100}, newNicConfig("docker0"), false},
+		{"whitelist overrides skip (lo)", psutilNet.IOCountersStat{Name: "lo", BytesSent: 100, BytesRecv: 100}, newNicConfig("lo"), false},
+		{"whitelist exclusion", psutilNet.IOCountersStat{Name: "eth1", BytesSent: 100, BytesRecv: 100}, newNicConfig("eth0"), true},
+		{"blacklist skip lo", psutilNet.IOCountersStat{Name: "lo", BytesSent: 100, BytesRecv: 100}, newNicConfig("-eth0"), true},
+		{"blacklist explicit eth0", psutilNet.IOCountersStat{Name: "eth0", BytesSent: 100, BytesRecv: 100}, newNicConfig("-eth0"), true},
+		{"blacklist allow eth1", psutilNet.IOCountersStat{Name: "eth1", BytesSent: 100, BytesRecv: 100}, newNicConfig("-eth0"), false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			assert.Equal(t, tt.expectSkip, skipNetworkInterface(tt.nic, tt.nicCfg))
+		})
+	}
+}
+
 func TestEnsureNetworkInterfacesMap(t *testing.T) {
 	var a Agent
 	var stats system.Stats

agent/utils/utils.go

@@ -1,6 +1,7 @@
 package utils
 
 import (
+	"io"
 	"math"
 	"os"
 	"strconv"
@@ -50,6 +51,23 @@ func ReadStringFileOK(path string) (string, bool) {
 	return strings.TrimSpace(string(b)), true
 }
 
+// ReadStringFileLimited reads a file into a string with a maximum size (in bytes) to avoid
+// allocating large buffers and potential panics with pseudo-files when the size is misreported.
+func ReadStringFileLimited(path string, maxSize int) (string, error) {
+	f, err := os.Open(path)
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+
+	buf := make([]byte, maxSize)
+	n, err := f.Read(buf)
+	if err != nil && err != io.EOF {
+		return "", err
+	}
+	return strings.TrimSpace(string(buf[:n])), nil
+}
+
 // FileExists reports whether the given path exists.
 func FileExists(path string) bool {
 	_, err := os.Stat(path)

go.mod

@@ -1,6 +1,6 @@
 module github.com/henrygd/beszel
 
-go 1.26.0
+go 1.26.1
 
 require (
 	github.com/blang/semver v3.5.1+incompatible

internal/alerts/alerts.go

@@ -40,16 +40,22 @@ type UserNotificationSettings struct {
 	Webhooks []string `json:"webhooks"`
 }
 
+type SystemAlertFsStats struct {
+	DiskTotal float64 `json:"d"`
+	DiskUsed  float64 `json:"du"`
+}
+
 type SystemAlertStats struct {
-	Cpu          float64                       `json:"cpu"`
-	Mem          float64                       `json:"mp"`
-	Disk         float64                       `json:"dp"`
-	NetSent      float64                       `json:"ns"`
-	NetRecv      float64                       `json:"nr"`
-	GPU          map[string]SystemAlertGPUData `json:"g"`
-	Temperatures map[string]float32            `json:"t"`
-	LoadAvg      [3]float64                    `json:"la"`
-	Battery      [2]uint8                      `json:"bat"`
+	Cpu          float64                        `json:"cpu"`
+	Mem          float64                        `json:"mp"`
+	Disk         float64                        `json:"dp"`
+	NetSent      float64                        `json:"ns"`
+	NetRecv      float64                        `json:"nr"`
+	GPU          map[string]SystemAlertGPUData  `json:"g"`
+	Temperatures map[string]float32             `json:"t"`
+	LoadAvg      [3]float64                     `json:"la"`
+	Battery      [2]uint8                       `json:"bat"`
+	ExtraFs      map[string]SystemAlertFsStats  `json:"efs"`
 }
 
 type SystemAlertGPUData struct {

internal/alerts/alerts_disk_test.go

@@ -0,0 +1,155 @@
+//go:build testing
+
+package alerts_test
+
+import (
+	"encoding/json"
+	"testing"
+	"time"
+
+	"github.com/henrygd/beszel/internal/entities/system"
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/tools/types"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// TestDiskAlertExtraFsMultiMinute tests that multi-minute disk alerts correctly use
+// historical per-minute values for extra (non-root) filesystems, not the current live snapshot.
+func TestDiskAlertExtraFsMultiMinute(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+	require.NoError(t, err)
+	systemRecord := systems[0]
+
+	// Disk alert: threshold 80%, min=2 (requires historical averaging)
+	diskAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "Disk",
+		"system": systemRecord.Id,
+		"user":   user.Id,
+		"value":  80, // threshold: 80%
+		"min":    2,  // 2 minutes - requires historical averaging
+	})
+	require.NoError(t, err)
+	assert.False(t, diskAlert.GetBool("triggered"), "Alert should not be triggered initially")
+
+	am := hub.GetAlertManager()
+	now := time.Now().UTC()
+
+	extraFsHigh := map[string]*system.FsStats{
+		"/mnt/data": {DiskTotal: 1000, DiskUsed: 920}, // 92% - above threshold
+	}
+
+	// Insert 4 historical records spread over 3 minutes (same pattern as battery tests).
+	// The oldest record must predate (now - 2min) so the alert time window is valid.
+	recordTimes := []time.Duration{
+		-180 * time.Second, // 3 min ago - anchors oldest record before alert.time
+		-90 * time.Second,
+		-60 * time.Second,
+		-30 * time.Second,
+	}
+
+	for _, offset := range recordTimes {
+		stats := system.Stats{
+			DiskPct: 30, // root disk at 30% - below threshold
+			ExtraFs: extraFsHigh,
+		}
+		statsJSON, _ := json.Marshal(stats)
+
+		recordTime := now.Add(offset)
+		record, err := beszelTests.CreateRecord(hub, "system_stats", map[string]any{
+			"system": systemRecord.Id,
+			"type":   "1m",
+			"stats":  string(statsJSON),
+		})
+		require.NoError(t, err)
+		record.SetRaw("created", recordTime.Format(types.DefaultDateLayout))
+		err = hub.SaveNoValidate(record)
+		require.NoError(t, err)
+	}
+
+	combinedDataHigh := &system.CombinedData{
+		Stats: system.Stats{
+			DiskPct: 30,
+			ExtraFs: extraFsHigh,
+		},
+		Info: system.Info{
+			DiskPct: 30,
+		},
+	}
+
+	systemRecord.Set("updated", now)
+	err = hub.SaveNoValidate(systemRecord)
+	require.NoError(t, err)
+
+	err = am.HandleSystemAlerts(systemRecord, combinedDataHigh)
+	require.NoError(t, err)
+
+	time.Sleep(20 * time.Millisecond)
+
+	diskAlert, err = hub.FindFirstRecordByFilter("alerts", "id={:id}", dbx.Params{"id": diskAlert.Id})
+	require.NoError(t, err)
+	assert.True(t, diskAlert.GetBool("triggered"),
+		"Alert SHOULD be triggered when extra disk average (92%%) exceeds threshold (80%%)")
+
+	// --- Resolution: extra disk drops to 50%, alert should resolve ---
+
+	extraFsLow := map[string]*system.FsStats{
+		"/mnt/data": {DiskTotal: 1000, DiskUsed: 500}, // 50% - below threshold
+	}
+
+	newNow := now.Add(2 * time.Minute)
+	recordTimesLow := []time.Duration{
+		-180 * time.Second,
+		-90 * time.Second,
+		-60 * time.Second,
+		-30 * time.Second,
+	}
+
+	for _, offset := range recordTimesLow {
+		stats := system.Stats{
+			DiskPct: 30,
+			ExtraFs: extraFsLow,
+		}
+		statsJSON, _ := json.Marshal(stats)
+
+		recordTime := newNow.Add(offset)
+		record, err := beszelTests.CreateRecord(hub, "system_stats", map[string]any{
+			"system": systemRecord.Id,
+			"type":   "1m",
+			"stats":  string(statsJSON),
+		})
+		require.NoError(t, err)
+		record.SetRaw("created", recordTime.Format(types.DefaultDateLayout))
+		err = hub.SaveNoValidate(record)
+		require.NoError(t, err)
+	}
+
+	combinedDataLow := &system.CombinedData{
+		Stats: system.Stats{
+			DiskPct: 30,
+			ExtraFs: extraFsLow,
+		},
+		Info: system.Info{
+			DiskPct: 30,
+		},
+	}
+
+	systemRecord.Set("updated", newNow)
+	err = hub.SaveNoValidate(systemRecord)
+	require.NoError(t, err)
+
+	err = am.HandleSystemAlerts(systemRecord, combinedDataLow)
+	require.NoError(t, err)
+
+	time.Sleep(20 * time.Millisecond)
+
+	diskAlert, err = hub.FindFirstRecordByFilter("alerts", "id={:id}", dbx.Params{"id": diskAlert.Id})
+	require.NoError(t, err)
+	assert.False(t, diskAlert.GetBool("triggered"),
+		"Alert should be resolved when extra disk average (50%%) drops below threshold (80%%)")
+}

internal/alerts/alerts_system.go

@@ -195,19 +195,21 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 				alert.val += stats.NetSent + stats.NetRecv
 			case "Disk":
 				if alert.mapSums == nil {
-					alert.mapSums = make(map[string]float32, len(data.Stats.ExtraFs)+1)
+					alert.mapSums = make(map[string]float32, len(stats.ExtraFs)+1)
 				}
 				// add root disk
 				if _, ok := alert.mapSums["root"]; !ok {
 					alert.mapSums["root"] = 0.0
 				}
 				alert.mapSums["root"] += float32(stats.Disk)
-				// add extra disks
-				for key, fs := range data.Stats.ExtraFs {
-					if _, ok := alert.mapSums[key]; !ok {
-						alert.mapSums[key] = 0.0
+				// add extra disks from historical record
+				for key, fs := range stats.ExtraFs {
+					if fs.DiskTotal > 0 {
+						if _, ok := alert.mapSums[key]; !ok {
+							alert.mapSums[key] = 0.0
+						}
+						alert.mapSums[key] += float32(fs.DiskUsed / fs.DiskTotal * 100)
 					}
-					alert.mapSums[key] += float32(fs.DiskUsed / fs.DiskTotal * 100)
 				}
 			case "Temperature":
 				if alert.mapSums == nil {