hub: add ExpiryMap.UpdateExpiration and sync SMART fetch intervals (#1800)

- Update smartFetchMap expiration when agent smart interval changes
- Prevent background SMART fetching before initial system details are
loaded
- Add buffer to SMART fetch timing check
- Get rid of unnecessary pointers in expirymap

Makefile

@@ -51,7 +51,6 @@ clean:
 lint:
 	golangci-lint run
 
-test: export GOEXPERIMENT=synctest
 test:
 	go test -tags=testing ./...
 

agent/agent.go

@@ -6,7 +6,6 @@ package agent
 
 import (
 	"log/slog"
-	"os"
 	"strings"
 	"sync"
 	"time"
@@ -14,6 +13,7 @@ import (
 	"github.com/gliderlabs/ssh"
 	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/agent/deltatracker"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/common"
 	"github.com/henrygd/beszel/internal/entities/system"
 	gossh "golang.org/x/crypto/ssh"
@@ -68,11 +68,11 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 		slog.Info("Data directory", "path", agent.dataDir)
 	}
 
-	agent.memCalc, _ = GetEnv("MEM_CALC")
+	agent.memCalc, _ = utils.GetEnv("MEM_CALC")
 	agent.sensorConfig = agent.newSensorConfig()
 
 	// Parse disk usage cache duration (e.g., "15m", "1h") to avoid waking sleeping disks
-	if diskUsageCache, exists := GetEnv("DISK_USAGE_CACHE"); exists {
+	if diskUsageCache, exists := utils.GetEnv("DISK_USAGE_CACHE"); exists {
 		if duration, err := time.ParseDuration(diskUsageCache); err == nil {
 			agent.diskUsageCacheDuration = duration
 			slog.Info("DISK_USAGE_CACHE", "duration", duration)
@@ -82,7 +82,7 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 	}
 
 	// Set up slog with a log level determined by the LOG_LEVEL env var
-	if logLevelStr, exists := GetEnv("LOG_LEVEL"); exists {
+	if logLevelStr, exists := utils.GetEnv("LOG_LEVEL"); exists {
 		switch strings.ToLower(logLevelStr) {
 		case "debug":
 			agent.debug = true
@@ -103,7 +103,7 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 	agent.refreshSystemDetails()
 
 	// SMART_INTERVAL env var to update smart data at this interval
-	if smartIntervalEnv, exists := GetEnv("SMART_INTERVAL"); exists {
+	if smartIntervalEnv, exists := utils.GetEnv("SMART_INTERVAL"); exists {
 		if duration, err := time.ParseDuration(smartIntervalEnv); err == nil && duration > 0 {
 			agent.systemDetails.SmartInterval = duration
 			slog.Info("SMART_INTERVAL", "duration", duration)
@@ -148,15 +148,6 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 	return agent, nil
 }
 
-// GetEnv retrieves an environment variable with a "BESZEL_AGENT_" prefix, or falls back to the unprefixed key.
-func GetEnv(key string) (value string, exists bool) {
-	if value, exists = os.LookupEnv("BESZEL_AGENT_" + key); exists {
-		return value, exists
-	}
-	// Fallback to the old unprefixed key
-	return os.LookupEnv(key)
-}
-
 func (a *Agent) gatherStats(options common.DataRequestOptions) *system.CombinedData {
 	a.Lock()
 	defer a.Unlock()
@@ -213,7 +204,7 @@ func (a *Agent) gatherStats(options common.DataRequestOptions) *system.CombinedD
 			data.Stats.ExtraFs[key] = stats
 			// Add percentages to Info struct for dashboard
 			if stats.DiskTotal > 0 {
-				pct := twoDecimals((stats.DiskUsed / stats.DiskTotal) * 100)
+				pct := utils.TwoDecimals((stats.DiskUsed / stats.DiskTotal) * 100)
 				data.Info.ExtraFsPct[key] = pct
 			}
 		}

agent/agent_cache_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/agent_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/client.go

@@ -14,6 +14,7 @@ import (
 	"time"
 
 	"github.com/henrygd/beszel"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/common"
 
 	"github.com/fxamacker/cbor/v2"
@@ -43,7 +44,7 @@ type WebSocketClient struct {
 // newWebSocketClient creates a new WebSocket client for the given agent.
 // It reads configuration from environment variables and validates the hub URL.
 func newWebSocketClient(agent *Agent) (client *WebSocketClient, err error) {
-	hubURLStr, exists := GetEnv("HUB_URL")
+	hubURLStr, exists := utils.GetEnv("HUB_URL")
 	if !exists {
 		return nil, errors.New("HUB_URL environment variable not set")
 	}
@@ -72,12 +73,12 @@ func newWebSocketClient(agent *Agent) (client *WebSocketClient, err error) {
 // If neither is set, it returns an error.
 func getToken() (string, error) {
 	// get token from env var
-	token, _ := GetEnv("TOKEN")
+	token, _ := utils.GetEnv("TOKEN")
 	if token != "" {
 		return token, nil
 	}
 	// get token from file
-	tokenFile, _ := GetEnv("TOKEN_FILE")
+	tokenFile, _ := utils.GetEnv("TOKEN_FILE")
 	if tokenFile == "" {
 		return "", errors.New("must set TOKEN or TOKEN_FILE")
 	}
@@ -197,7 +198,7 @@ func (client *WebSocketClient) handleAuthChallenge(msg *common.HubRequest[cbor.R
 	}
 
 	if authRequest.NeedSysInfo {
-		response.Name, _ = GetEnv("SYSTEM_NAME")
+		response.Name, _ = utils.GetEnv("SYSTEM_NAME")
 		response.Hostname = client.agent.systemDetails.Hostname
 		serverAddr := client.agent.connectionManager.serverOptions.Addr
 		_, response.Port, _ = net.SplitHostPort(serverAddr)

agent/client_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/connection_manager.go

@@ -1,9 +1,9 @@
 package agent
 
 import (
+	"context"
 	"errors"
 	"log/slog"
-	"os"
 	"os/signal"
 	"syscall"
 	"time"
@@ -91,8 +91,8 @@ func (c *ConnectionManager) Start(serverOptions ServerOptions) error {
 	c.eventChan = make(chan ConnectionEvent, 1)
 
 	// signal handling for shutdown
-	sigChan := make(chan os.Signal, 1)
-	signal.Notify(sigChan, syscall.SIGINT, syscall.SIGTERM)
+	sigCtx, stopSignals := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
+	defer stopSignals()
 
 	c.startWsTicker()
 	c.connect()
@@ -109,8 +109,8 @@ func (c *ConnectionManager) Start(serverOptions ServerOptions) error {
 			_ = c.startWebSocketConnection()
 		case <-healthTicker:
 			_ = health.Update()
-		case <-sigChan:
-			slog.Info("Shutting down")
+		case <-sigCtx.Done():
+			slog.Info("Shutting down", "cause", context.Cause(sigCtx))
 			_ = c.agent.StopServer()
 			c.closeWebSocket()
 			return health.CleanUp()

agent/connection_manager_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/cpu.go

@@ -14,10 +14,10 @@ var lastPerCoreCpuTimes = make(map[uint16][]cpu.TimesStat)
 // init initializes the CPU monitoring by storing the initial CPU times
 // for the default 60-second cache interval.
 func init() {
-	if times, err := cpu.Times(false); err == nil {
+	if times, err := cpu.Times(false); err == nil && len(times) > 0 {
 		lastCpuTimes[60000] = times[0]
 	}
-	if perCoreTimes, err := cpu.Times(true); err == nil {
+	if perCoreTimes, err := cpu.Times(true); err == nil && len(perCoreTimes) > 0 {
 		lastPerCoreCpuTimes[60000] = perCoreTimes
 	}
 }
@@ -89,10 +89,7 @@ func getPerCoreCpuUsage(cacheTimeMs uint16) (system.Uint8Slice, error) {
 	lastTimes := lastPerCoreCpuTimes[cacheTimeMs]
 
 	// Limit to the number of cores available in both samples
-	length := len(perCoreTimes)
-	if len(lastTimes) < length {
-		length = len(lastTimes)
-	}
+	length := min(len(lastTimes), len(perCoreTimes))
 
 	usage := make([]uint8, length)
 	for i := 0; i < length; i++ {

agent/data_dir.go

@@ -6,6 +6,8 @@ import (
 	"os"
 	"path/filepath"
 	"runtime"
+
+	"github.com/henrygd/beszel/agent/utils"
 )
 
 // GetDataDir returns the path to the data directory for the agent and an error
@@ -16,7 +18,7 @@ func GetDataDir(dataDirs ...string) (string, error) {
 		return testDataDirs(dataDirs)
 	}
 
-	dataDir, _ := GetEnv("DATA_DIR")
+	dataDir, _ := utils.GetEnv("DATA_DIR")
 	if dataDir != "" {
 		dataDirs = append(dataDirs, dataDir)
 	}

agent/data_dir_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/disk.go

@@ -8,6 +8,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 
 	"github.com/shirou/gopsutil/v4/disk"
@@ -37,7 +38,7 @@ func isDockerSpecialMountpoint(mountpoint string) bool {
 
 // Sets up the filesystems to monitor for disk usage and I/O.
 func (a *Agent) initializeDiskInfo() {
-	filesystem, _ := GetEnv("FILESYSTEM")
+	filesystem, _ := utils.GetEnv("FILESYSTEM")
 	efPath := "/extra-filesystems"
 	hasRoot := false
 	isWindows := runtime.GOOS == "windows"
@@ -78,14 +79,21 @@ func (a *Agent) initializeDiskInfo() {
 		if _, exists := a.fsStats[key]; !exists {
 			if root {
 				slog.Info("Detected root device", "name", key)
-				// Check if root device is in /proc/diskstats. Do not guess a
-				// fallback device for root: that can misattribute root I/O to a
-				// different disk while usage remains tied to root mountpoint.
+				// Try to map root device to a diskIoCounters entry. First
+				// checks for an exact key match, then uses findIoDevice for
+				// normalized / prefix-based matching (e.g. nda0p2 → nda0),
+				// and finally falls back to the FILESYSTEM env var.
 				if _, ioMatch = diskIoCounters[key]; !ioMatch {
-					if matchedKey, match := findIoDevice(filesystem, diskIoCounters); match {
+					if matchedKey, match := findIoDevice(key, diskIoCounters); match {
 						key = matchedKey
 						ioMatch = true
-					} else {
+					} else if filesystem != "" {
+						if matchedKey, match := findIoDevice(filesystem, diskIoCounters); match {
+							key = matchedKey
+							ioMatch = true
+						}
+					}
+					if !ioMatch {
 						slog.Warn("Root I/O unmapped; set FILESYSTEM", "device", device, "mountpoint", mountpoint)
 					}
 				}
@@ -114,20 +122,28 @@ func (a *Agent) initializeDiskInfo() {
 	// Use FILESYSTEM env var to find root filesystem
 	if filesystem != "" {
 		for _, p := range partitions {
-			if strings.HasSuffix(p.Device, filesystem) || p.Mountpoint == filesystem {
+			if filesystemMatchesPartitionSetting(filesystem, p) {
 				addFsStat(p.Device, p.Mountpoint, true)
 				hasRoot = true
 				break
 			}
 		}
 		if !hasRoot {
-			slog.Warn("Partition details not found", "filesystem", filesystem)
+			// FILESYSTEM may name a physical disk absent from partitions (e.g.
+			// ZFS lists dataset paths like zroot/ROOT/default, not block devices).
+			// Try matching directly against diskIoCounters.
+			if ioKey, match := findIoDevice(filesystem, diskIoCounters); match {
+				a.fsStats[ioKey] = &system.FsStats{Root: true, Mountpoint: rootMountPoint}
+				hasRoot = true
+			} else {
+				slog.Warn("Partition details not found", "filesystem", filesystem)
+			}
 		}
 	}
 
 	// Add EXTRA_FILESYSTEMS env var values to fsStats
-	if extraFilesystems, exists := GetEnv("EXTRA_FILESYSTEMS"); exists {
-		for _, fsEntry := range strings.Split(extraFilesystems, ",") {
+	if extraFilesystems, exists := utils.GetEnv("EXTRA_FILESYSTEMS"); exists {
+		for fsEntry := range strings.SplitSeq(extraFilesystems, ",") {
 			// Parse custom name from format: device__customname
 			fs, customName := parseFilesystemEntry(fsEntry)
 
@@ -187,28 +203,180 @@ func (a *Agent) initializeDiskInfo() {
 		}
 	}
 
-	// If no root filesystem set, use fallback
+	// If no root filesystem set, try the most active I/O device as a last
+	// resort (e.g. ZFS where dataset names are unrelated to disk names).
 	if !hasRoot {
-		rootKey := filepath.Base(rootMountPoint)
-		if _, exists := a.fsStats[rootKey]; exists {
-			rootKey = "root"
+		rootKey := mostActiveIoDevice(diskIoCounters)
+		if rootKey != "" {
+			slog.Warn("Using most active device for root I/O; set FILESYSTEM to override", "device", rootKey)
+		} else {
+			rootKey = filepath.Base(rootMountPoint)
+			if _, exists := a.fsStats[rootKey]; exists {
+				rootKey = "root"
+			}
+			slog.Warn("Root I/O device not detected; set FILESYSTEM to override")
 		}
-		slog.Warn("Root device not detected; root I/O disabled", "mountpoint", rootMountPoint)
 		a.fsStats[rootKey] = &system.FsStats{Root: true, Mountpoint: rootMountPoint}
 	}
 
+	a.pruneDuplicateRootExtraFilesystems()
 	a.initializeDiskIoStats(diskIoCounters)
 }
 
-// Returns matching device from /proc/diskstats.
-// bool is true if a match was found.
-func findIoDevice(filesystem string, diskIoCounters map[string]disk.IOCountersStat) (string, bool) {
-	for _, d := range diskIoCounters {
-		if d.Name == filesystem || (d.Label != "" && d.Label == filesystem) {
-			return d.Name, true
+// Removes extra filesystems that mirror root usage (https://github.com/henrygd/beszel/issues/1428).
+func (a *Agent) pruneDuplicateRootExtraFilesystems() {
+	var rootMountpoint string
+	for _, stats := range a.fsStats {
+		if stats != nil && stats.Root {
+			rootMountpoint = stats.Mountpoint
+			break
 		}
 	}
-	return "", false
+	if rootMountpoint == "" {
+		return
+	}
+	rootUsage, err := disk.Usage(rootMountpoint)
+	if err != nil {
+		return
+	}
+	for name, stats := range a.fsStats {
+		if stats == nil || stats.Root {
+			continue
+		}
+		extraUsage, err := disk.Usage(stats.Mountpoint)
+		if err != nil {
+			continue
+		}
+		if hasSameDiskUsage(rootUsage, extraUsage) {
+			slog.Info("Ignoring duplicate FS", "name", name, "mount", stats.Mountpoint)
+			delete(a.fsStats, name)
+		}
+	}
+}
+
+// hasSameDiskUsage compares root/extra usage with a small byte tolerance.
+func hasSameDiskUsage(a, b *disk.UsageStat) bool {
+	if a == nil || b == nil || a.Total == 0 || b.Total == 0 {
+		return false
+	}
+	// Allow minor drift between sequential disk usage calls.
+	const toleranceBytes uint64 = 16 * 1024 * 1024
+	return withinUsageTolerance(a.Total, b.Total, toleranceBytes) &&
+		withinUsageTolerance(a.Used, b.Used, toleranceBytes)
+}
+
+// withinUsageTolerance reports whether two byte values differ by at most tolerance.
+func withinUsageTolerance(a, b, tolerance uint64) bool {
+	if a >= b {
+		return a-b <= tolerance
+	}
+	return b-a <= tolerance
+}
+
+type ioMatchCandidate struct {
+	name  string
+	bytes uint64
+	ops   uint64
+}
+
+// findIoDevice prefers exact device/label matches, then falls back to a
+// prefix-related candidate with the highest recent activity.
+func findIoDevice(filesystem string, diskIoCounters map[string]disk.IOCountersStat) (string, bool) {
+	filesystem = normalizeDeviceName(filesystem)
+	if filesystem == "" {
+		return "", false
+	}
+
+	candidates := []ioMatchCandidate{}
+
+	for _, d := range diskIoCounters {
+		if normalizeDeviceName(d.Name) == filesystem || (d.Label != "" && normalizeDeviceName(d.Label) == filesystem) {
+			return d.Name, true
+		}
+		if prefixRelated(normalizeDeviceName(d.Name), filesystem) ||
+			(d.Label != "" && prefixRelated(normalizeDeviceName(d.Label), filesystem)) {
+			candidates = append(candidates, ioMatchCandidate{
+				name:  d.Name,
+				bytes: d.ReadBytes + d.WriteBytes,
+				ops:   d.ReadCount + d.WriteCount,
+			})
+		}
+	}
+
+	if len(candidates) == 0 {
+		return "", false
+	}
+
+	best := candidates[0]
+	for _, c := range candidates[1:] {
+		if c.bytes > best.bytes ||
+			(c.bytes == best.bytes && c.ops > best.ops) ||
+			(c.bytes == best.bytes && c.ops == best.ops && c.name < best.name) {
+			best = c
+		}
+	}
+
+	slog.Info("Using disk I/O fallback", "requested", filesystem, "selected", best.name)
+	return best.name, true
+}
+
+// mostActiveIoDevice returns the device with the highest I/O activity,
+// or "" if diskIoCounters is empty.
+func mostActiveIoDevice(diskIoCounters map[string]disk.IOCountersStat) string {
+	var best ioMatchCandidate
+	for _, d := range diskIoCounters {
+		c := ioMatchCandidate{
+			name:  d.Name,
+			bytes: d.ReadBytes + d.WriteBytes,
+			ops:   d.ReadCount + d.WriteCount,
+		}
+		if best.name == "" || c.bytes > best.bytes ||
+			(c.bytes == best.bytes && c.ops > best.ops) ||
+			(c.bytes == best.bytes && c.ops == best.ops && c.name < best.name) {
+			best = c
+		}
+	}
+	return best.name
+}
+
+// prefixRelated reports whether either identifier is a prefix of the other.
+func prefixRelated(a, b string) bool {
+	if a == "" || b == "" || a == b {
+		return false
+	}
+	return strings.HasPrefix(a, b) || strings.HasPrefix(b, a)
+}
+
+// filesystemMatchesPartitionSetting checks whether a FILESYSTEM env var value
+// matches a partition by mountpoint, exact device name, or prefix relationship
+// (e.g. FILESYSTEM=ada0 matches partition /dev/ada0p2).
+func filesystemMatchesPartitionSetting(filesystem string, p disk.PartitionStat) bool {
+	filesystem = strings.TrimSpace(filesystem)
+	if filesystem == "" {
+		return false
+	}
+	if p.Mountpoint == filesystem {
+		return true
+	}
+
+	fsName := normalizeDeviceName(filesystem)
+	partName := normalizeDeviceName(p.Device)
+	if fsName == "" || partName == "" {
+		return false
+	}
+	if fsName == partName {
+		return true
+	}
+	return prefixRelated(partName, fsName)
+}
+
+// normalizeDeviceName canonicalizes device strings for comparisons.
+func normalizeDeviceName(value string) string {
+	name := filepath.Base(strings.TrimSpace(value))
+	if name == "." {
+		return ""
+	}
+	return name
 }
 
 // Sets start values for disk I/O stats.
@@ -245,12 +413,12 @@ func (a *Agent) updateDiskUsage(systemStats *system.Stats) {
 			continue
 		}
 		if d, err := disk.Usage(stats.Mountpoint); err == nil {
-			stats.DiskTotal = bytesToGigabytes(d.Total)
-			stats.DiskUsed = bytesToGigabytes(d.Used)
+			stats.DiskTotal = utils.BytesToGigabytes(d.Total)
+			stats.DiskUsed = utils.BytesToGigabytes(d.Used)
 			if stats.Root {
-				systemStats.DiskTotal = bytesToGigabytes(d.Total)
-				systemStats.DiskUsed = bytesToGigabytes(d.Used)
-				systemStats.DiskPct = twoDecimals(d.UsedPercent)
+				systemStats.DiskTotal = utils.BytesToGigabytes(d.Total)
+				systemStats.DiskUsed = utils.BytesToGigabytes(d.Used)
+				systemStats.DiskPct = utils.TwoDecimals(d.UsedPercent)
 			}
 		} else {
 			// reset stats if error (likely unmounted)
@@ -303,8 +471,8 @@ func (a *Agent) updateDiskIo(cacheTimeMs uint16, systemStats *system.Stats) {
 
 			diskIORead := (d.ReadBytes - prev.readBytes) * 1000 / msElapsed
 			diskIOWrite := (d.WriteBytes - prev.writeBytes) * 1000 / msElapsed
-			readMbPerSecond := bytesToMegabytes(float64(diskIORead))
-			writeMbPerSecond := bytesToMegabytes(float64(diskIOWrite))
+			readMbPerSecond := utils.BytesToMegabytes(float64(diskIORead))
+			writeMbPerSecond := utils.BytesToMegabytes(float64(diskIOWrite))
 
 			// validate values
 			if readMbPerSecond > 50_000 || writeMbPerSecond > 50_000 {

agent/disk_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 
@@ -117,7 +116,7 @@ func TestFindIoDevice(t *testing.T) {
 		assert.Equal(t, "sda", device)
 	})
 
-	t.Run("returns no fallback when not found", func(t *testing.T) {
+	t.Run("returns no match when not found", func(t *testing.T) {
 		ioCounters := map[string]disk.IOCountersStat{
 			"sda": {Name: "sda"},
 			"sdb": {Name: "sdb"},
@@ -127,6 +126,106 @@ func TestFindIoDevice(t *testing.T) {
 		assert.False(t, ok)
 		assert.Equal(t, "", device)
 	})
+
+	t.Run("uses uncertain unique prefix fallback", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"nvme0n1": {Name: "nvme0n1"},
+			"sda":     {Name: "sda"},
+		}
+
+		device, ok := findIoDevice("nvme0n1p2", ioCounters)
+		assert.True(t, ok)
+		assert.Equal(t, "nvme0n1", device)
+	})
+
+	t.Run("uses dominant activity when prefix matches are ambiguous", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"sda": {Name: "sda", ReadBytes: 5000, WriteBytes: 5000, ReadCount: 100, WriteCount: 100},
+			"sdb": {Name: "sdb", ReadBytes: 1000, WriteBytes: 1000, ReadCount: 50, WriteCount: 50},
+		}
+
+		device, ok := findIoDevice("sd", ioCounters)
+		assert.True(t, ok)
+		assert.Equal(t, "sda", device)
+	})
+
+	t.Run("uses highest activity when ambiguous without dominance", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"sda": {Name: "sda", ReadBytes: 3000, WriteBytes: 3000, ReadCount: 50, WriteCount: 50},
+			"sdb": {Name: "sdb", ReadBytes: 2500, WriteBytes: 2500, ReadCount: 40, WriteCount: 40},
+		}
+
+		device, ok := findIoDevice("sd", ioCounters)
+		assert.True(t, ok)
+		assert.Equal(t, "sda", device)
+	})
+
+	t.Run("matches /dev/-prefixed partition to parent disk", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"nda0": {Name: "nda0", ReadBytes: 1000, WriteBytes: 1000},
+		}
+
+		device, ok := findIoDevice("/dev/nda0p2", ioCounters)
+		assert.True(t, ok)
+		assert.Equal(t, "nda0", device)
+	})
+
+	t.Run("uses deterministic name tie-breaker", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"sdb": {Name: "sdb", ReadBytes: 2000, WriteBytes: 2000, ReadCount: 10, WriteCount: 10},
+			"sda": {Name: "sda", ReadBytes: 2000, WriteBytes: 2000, ReadCount: 10, WriteCount: 10},
+		}
+
+		device, ok := findIoDevice("sd", ioCounters)
+		assert.True(t, ok)
+		assert.Equal(t, "sda", device)
+	})
+}
+
+func TestFilesystemMatchesPartitionSetting(t *testing.T) {
+	p := disk.PartitionStat{Device: "/dev/ada0p2", Mountpoint: "/"}
+
+	t.Run("matches mountpoint setting", func(t *testing.T) {
+		assert.True(t, filesystemMatchesPartitionSetting("/", p))
+	})
+
+	t.Run("matches exact partition setting", func(t *testing.T) {
+		assert.True(t, filesystemMatchesPartitionSetting("ada0p2", p))
+		assert.True(t, filesystemMatchesPartitionSetting("/dev/ada0p2", p))
+	})
+
+	t.Run("matches prefix-style parent setting", func(t *testing.T) {
+		assert.True(t, filesystemMatchesPartitionSetting("ada0", p))
+		assert.True(t, filesystemMatchesPartitionSetting("/dev/ada0", p))
+	})
+
+	t.Run("does not match unrelated device", func(t *testing.T) {
+		assert.False(t, filesystemMatchesPartitionSetting("sda", p))
+		assert.False(t, filesystemMatchesPartitionSetting("nvme0n1", p))
+		assert.False(t, filesystemMatchesPartitionSetting("", p))
+	})
+}
+
+func TestMostActiveIoDevice(t *testing.T) {
+	t.Run("returns most active device", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"nda0": {Name: "nda0", ReadBytes: 5000, WriteBytes: 5000, ReadCount: 100, WriteCount: 100},
+			"nda1": {Name: "nda1", ReadBytes: 1000, WriteBytes: 1000, ReadCount: 50, WriteCount: 50},
+		}
+		assert.Equal(t, "nda0", mostActiveIoDevice(ioCounters))
+	})
+
+	t.Run("uses deterministic tie-breaker", func(t *testing.T) {
+		ioCounters := map[string]disk.IOCountersStat{
+			"sdb": {Name: "sdb", ReadBytes: 1000, WriteBytes: 1000, ReadCount: 10, WriteCount: 10},
+			"sda": {Name: "sda", ReadBytes: 1000, WriteBytes: 1000, ReadCount: 10, WriteCount: 10},
+		}
+		assert.Equal(t, "sda", mostActiveIoDevice(ioCounters))
+	})
+
+	t.Run("returns empty for empty map", func(t *testing.T) {
+		assert.Equal(t, "", mostActiveIoDevice(map[string]disk.IOCountersStat{}))
+	})
 }
 
 func TestIsDockerSpecialMountpoint(t *testing.T) {
@@ -373,3 +472,37 @@ func TestDiskUsageCaching(t *testing.T) {
 			"lastDiskUsageUpdate should be refreshed when cache expires")
 	})
 }
+
+func TestHasSameDiskUsage(t *testing.T) {
+	const toleranceBytes uint64 = 16 * 1024 * 1024
+
+	t.Run("returns true when totals and usage are equal", func(t *testing.T) {
+		a := &disk.UsageStat{Total: 100 * 1024 * 1024 * 1024, Used: 42 * 1024 * 1024 * 1024}
+		b := &disk.UsageStat{Total: 100 * 1024 * 1024 * 1024, Used: 42 * 1024 * 1024 * 1024}
+		assert.True(t, hasSameDiskUsage(a, b))
+	})
+
+	t.Run("returns true within tolerance", func(t *testing.T) {
+		a := &disk.UsageStat{Total: 100 * 1024 * 1024 * 1024, Used: 42 * 1024 * 1024 * 1024}
+		b := &disk.UsageStat{
+			Total: a.Total + toleranceBytes - 1,
+			Used:  a.Used - toleranceBytes + 1,
+		}
+		assert.True(t, hasSameDiskUsage(a, b))
+	})
+
+	t.Run("returns false when total exceeds tolerance", func(t *testing.T) {
+		a := &disk.UsageStat{Total: 100 * 1024 * 1024 * 1024, Used: 42 * 1024 * 1024 * 1024}
+		b := &disk.UsageStat{
+			Total: a.Total + toleranceBytes + 1,
+			Used:  a.Used,
+		}
+		assert.False(t, hasSameDiskUsage(a, b))
+	})
+
+	t.Run("returns false for nil or zero total", func(t *testing.T) {
+		assert.False(t, hasSameDiskUsage(nil, &disk.UsageStat{Total: 1, Used: 1}))
+		assert.False(t, hasSameDiskUsage(&disk.UsageStat{Total: 1, Used: 1}, nil))
+		assert.False(t, hasSameDiskUsage(&disk.UsageStat{Total: 0, Used: 0}, &disk.UsageStat{Total: 1, Used: 1}))
+	})
+}

agent/docker.go

@@ -1,6 +1,7 @@
 package agent
 
 import (
+	"bufio"
 	"bytes"
 	"context"
 	"encoding/binary"
@@ -20,6 +21,7 @@ import (
 	"time"
 
 	"github.com/henrygd/beszel/agent/deltatracker"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/container"
 
 	"github.com/blang/semver"
@@ -28,6 +30,7 @@ import (
 // ansiEscapePattern matches ANSI escape sequences (colors, cursor movement, etc.)
 // This includes CSI sequences like \x1b[...m and simple escapes like \x1b[K
 var ansiEscapePattern = regexp.MustCompile(`\x1b\[[0-9;]*[a-zA-Z]|\x1b\][^\x07]*\x07|\x1b[@-Z\\-_]`)
+var dockerContainerIDPattern = regexp.MustCompile(`^[a-fA-F0-9]{12,64}$`)
 
 const (
 	// Docker API timeout in milliseconds
@@ -72,6 +75,7 @@ type dockerManager struct {
 	// cacheTimeMs -> DeltaTracker for network bytes sent/received
 	networkSentTrackers map[uint16]*deltatracker.DeltaTracker[string, uint64]
 	networkRecvTrackers map[uint16]*deltatracker.DeltaTracker[string, uint64]
+	retrySleep          func(time.Duration)
 }
 
 // userAgentRoundTripper is a custom http.RoundTripper that adds a User-Agent header to all requests
@@ -333,12 +337,12 @@ func validateCpuPercentage(cpuPct float64, containerName string) error {
 
 // updateContainerStatsValues updates the final stats values
 func updateContainerStatsValues(stats *container.Stats, cpuPct float64, usedMemory uint64, sent_delta, recv_delta uint64, readTime time.Time) {
-	stats.Cpu = twoDecimals(cpuPct)
-	stats.Mem = bytesToMegabytes(float64(usedMemory))
+	stats.Cpu = utils.TwoDecimals(cpuPct)
+	stats.Mem = utils.BytesToMegabytes(float64(usedMemory))
 	stats.Bandwidth = [2]uint64{sent_delta, recv_delta}
 	// TODO(0.19+): stop populating NetworkSent/NetworkRecv (deprecated in 0.18.3)
-	stats.NetworkSent = bytesToMegabytes(float64(sent_delta))
-	stats.NetworkRecv = bytesToMegabytes(float64(recv_delta))
+	stats.NetworkSent = utils.BytesToMegabytes(float64(sent_delta))
+	stats.NetworkRecv = utils.BytesToMegabytes(float64(recv_delta))
 	stats.PrevReadTime = readTime
 }
 
@@ -484,7 +488,7 @@ func (dm *dockerManager) deleteContainerStatsSync(id string) {
 
 // Creates a new http client for Docker or Podman API
 func newDockerManager() *dockerManager {
-	dockerHost, exists := GetEnv("DOCKER_HOST")
+	dockerHost, exists := utils.GetEnv("DOCKER_HOST")
 	if exists {
 		// return nil if set to empty string
 		if dockerHost == "" {
@@ -520,7 +524,7 @@ func newDockerManager() *dockerManager {
 
 	// configurable timeout
 	timeout := time.Millisecond * time.Duration(dockerTimeoutMs)
-	if t, set := GetEnv("DOCKER_TIMEOUT"); set {
+	if t, set := utils.GetEnv("DOCKER_TIMEOUT"); set {
 		timeout, err = time.ParseDuration(t)
 		if err != nil {
 			slog.Error(err.Error())
@@ -537,7 +541,7 @@ func newDockerManager() *dockerManager {
 
 	// Read container exclusion patterns from environment variable
 	var excludeContainers []string
-	if excludeStr, set := GetEnv("EXCLUDE_CONTAINERS"); set && excludeStr != "" {
+	if excludeStr, set := utils.GetEnv("EXCLUDE_CONTAINERS"); set && excludeStr != "" {
 		parts := strings.SplitSeq(excludeStr, ",")
 		for part := range parts {
 			trimmed := strings.TrimSpace(part)
@@ -565,6 +569,7 @@ func newDockerManager() *dockerManager {
 		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		retrySleep:          time.Sleep,
 	}
 
 	// If using podman, return client
@@ -574,7 +579,7 @@ func newDockerManager() *dockerManager {
 		return manager
 	}
 
-	// this can take up to 5 seconds with retry, so run in goroutine
+	// run version check in goroutine to avoid blocking (server may not be ready and requires retries)
 	go manager.checkDockerVersion()
 
 	// give version check a chance to complete before returning
@@ -594,18 +599,18 @@ func (dm *dockerManager) checkDockerVersion() {
 	const versionMaxTries = 2
 	for i := 1; i <= versionMaxTries; i++ {
 		resp, err = dm.client.Get("http://localhost/version")
-		if err == nil {
+		if err == nil && resp.StatusCode == http.StatusOK {
 			break
 		}
 		if resp != nil {
 			resp.Body.Close()
 		}
 		if i < versionMaxTries {
-			slog.Debug("Failed to get Docker version; retrying", "attempt", i, "error", err)
-			time.Sleep(5 * time.Second)
+			slog.Debug("Failed to get Docker version; retrying", "attempt", i, "err", err, "response", resp)
+			dm.retrySleep(5 * time.Second)
 		}
 	}
-	if err != nil {
+	if err != nil || resp.StatusCode != http.StatusOK {
 		return
 	}
 	if err := dm.decode(resp, &versionInfo); err != nil {
@@ -647,9 +652,34 @@ func getDockerHost() string {
 	return scheme + socks[0]
 }
 
+func validateContainerID(containerID string) error {
+	if !dockerContainerIDPattern.MatchString(containerID) {
+		return fmt.Errorf("invalid container id")
+	}
+	return nil
+}
+
+func buildDockerContainerEndpoint(containerID, action string, query url.Values) (string, error) {
+	if err := validateContainerID(containerID); err != nil {
+		return "", err
+	}
+	u := &url.URL{
+		Scheme: "http",
+		Host:   "localhost",
+		Path:   fmt.Sprintf("/containers/%s/%s", url.PathEscape(containerID), action),
+	}
+	if len(query) > 0 {
+		u.RawQuery = query.Encode()
+	}
+	return u.String(), nil
+}
+
 // getContainerInfo fetches the inspection data for a container
 func (dm *dockerManager) getContainerInfo(ctx context.Context, containerID string) ([]byte, error) {
-	endpoint := fmt.Sprintf("http://localhost/containers/%s/json", containerID)
+	endpoint, err := buildDockerContainerEndpoint(containerID, "json", nil)
+	if err != nil {
+		return nil, err
+	}
 	req, err := http.NewRequestWithContext(ctx, http.MethodGet, endpoint, nil)
 	if err != nil {
 		return nil, err
@@ -680,7 +710,15 @@ func (dm *dockerManager) getContainerInfo(ctx context.Context, containerID strin
 
 // getLogs fetches the logs for a container
 func (dm *dockerManager) getLogs(ctx context.Context, containerID string) (string, error) {
-	endpoint := fmt.Sprintf("http://localhost/containers/%s/logs?stdout=1&stderr=1&tail=%d", containerID, dockerLogsTail)
+	query := url.Values{
+		"stdout": []string{"1"},
+		"stderr": []string{"1"},
+		"tail":   []string{fmt.Sprintf("%d", dockerLogsTail)},
+	}
+	endpoint, err := buildDockerContainerEndpoint(containerID, "logs", query)
+	if err != nil {
+		return "", err
+	}
 	req, err := http.NewRequestWithContext(ctx, http.MethodGet, endpoint, nil)
 	if err != nil {
 		return "", err
@@ -698,8 +736,17 @@ func (dm *dockerManager) getLogs(ctx context.Context, containerID string) (strin
 	}
 
 	var builder strings.Builder
-	multiplexed := resp.Header.Get("Content-Type") == "application/vnd.docker.multiplexed-stream"
-	if err := decodeDockerLogStream(resp.Body, &builder, multiplexed); err != nil {
+	contentType := resp.Header.Get("Content-Type")
+	multiplexed := strings.HasSuffix(contentType, "multiplexed-stream")
+	logReader := io.Reader(resp.Body)
+	if !multiplexed {
+		// Podman may return multiplexed logs without Content-Type. Sniff the first frame header
+		// with a small buffered reader only when the header check fails.
+		bufferedReader := bufio.NewReaderSize(resp.Body, 8)
+		multiplexed = detectDockerMultiplexedStream(bufferedReader)
+		logReader = bufferedReader
+	}
+	if err := decodeDockerLogStream(logReader, &builder, multiplexed); err != nil {
 		return "", err
 	}
 
@@ -711,6 +758,23 @@ func (dm *dockerManager) getLogs(ctx context.Context, containerID string) (strin
 	return logs, nil
 }
 
+func detectDockerMultiplexedStream(reader *bufio.Reader) bool {
+	const headerSize = 8
+	header, err := reader.Peek(headerSize)
+	if err != nil {
+		return false
+	}
+	if header[0] != 0x01 && header[0] != 0x02 {
+		return false
+	}
+	// Docker's stream framing header reserves bytes 1-3 as zero.
+	if header[1] != 0 || header[2] != 0 || header[3] != 0 {
+		return false
+	}
+	frameLen := binary.BigEndian.Uint32(header[4:])
+	return frameLen <= maxLogFrameSize
+}
+
 func decodeDockerLogStream(reader io.Reader, builder *strings.Builder, multiplexed bool) error {
 	if !multiplexed {
 		_, err := io.Copy(builder, io.LimitReader(reader, maxTotalLogSize))

agent/docker_test.go

@@ -1,17 +1,24 @@
 //go:build testing
-// +build testing
 
 package agent
 
 import (
 	"bytes"
+	"context"
 	"encoding/json"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"net/http/httptest"
 	"os"
 	"strings"
 	"testing"
 	"time"
 
 	"github.com/henrygd/beszel/agent/deltatracker"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/container"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -19,6 +26,37 @@ import (
 
 var defaultCacheTimeMs = uint16(60_000)
 
+type recordingRoundTripper struct {
+	statusCode  int
+	body        string
+	contentType string
+	called      bool
+	lastPath    string
+	lastQuery   map[string]string
+}
+
+func (rt *recordingRoundTripper) RoundTrip(req *http.Request) (*http.Response, error) {
+	rt.called = true
+	rt.lastPath = req.URL.EscapedPath()
+	rt.lastQuery = map[string]string{}
+	for key, values := range req.URL.Query() {
+		if len(values) > 0 {
+			rt.lastQuery[key] = values[0]
+		}
+	}
+	resp := &http.Response{
+		StatusCode: rt.statusCode,
+		Status:     "200 OK",
+		Header:     make(http.Header),
+		Body:       io.NopCloser(strings.NewReader(rt.body)),
+		Request:    req,
+	}
+	if rt.contentType != "" {
+		resp.Header.Set("Content-Type", rt.contentType)
+	}
+	return resp, nil
+}
+
 // cycleCpuDeltas cycles the CPU tracking data for a specific cache time interval
 func (dm *dockerManager) cycleCpuDeltas(cacheTimeMs uint16) {
 	// Clear the CPU tracking maps for this cache time interval
@@ -110,6 +148,72 @@ func TestCalculateMemoryUsage(t *testing.T) {
 	}
 }
 
+func TestBuildDockerContainerEndpoint(t *testing.T) {
+	t.Run("valid container ID builds escaped endpoint", func(t *testing.T) {
+		endpoint, err := buildDockerContainerEndpoint("0123456789ab", "json", nil)
+		require.NoError(t, err)
+		assert.Equal(t, "http://localhost/containers/0123456789ab/json", endpoint)
+	})
+
+	t.Run("invalid container ID is rejected", func(t *testing.T) {
+		_, err := buildDockerContainerEndpoint("../../version", "json", nil)
+		require.Error(t, err)
+		assert.Contains(t, err.Error(), "invalid container id")
+	})
+}
+
+func TestContainerDetailsRequestsValidateContainerID(t *testing.T) {
+	rt := &recordingRoundTripper{
+		statusCode: 200,
+		body:       `{"Config":{"Env":["SECRET=1"]}}`,
+	}
+	dm := &dockerManager{
+		client: &http.Client{Transport: rt},
+	}
+
+	_, err := dm.getContainerInfo(context.Background(), "../version")
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "invalid container id")
+	assert.False(t, rt.called, "request should be rejected before dispatching to Docker API")
+}
+
+func TestContainerDetailsRequestsUseExpectedDockerPaths(t *testing.T) {
+	t.Run("container info uses container json endpoint", func(t *testing.T) {
+		rt := &recordingRoundTripper{
+			statusCode: 200,
+			body:       `{"Config":{"Env":["SECRET=1"]},"Name":"demo"}`,
+		}
+		dm := &dockerManager{
+			client: &http.Client{Transport: rt},
+		}
+
+		body, err := dm.getContainerInfo(context.Background(), "0123456789ab")
+		require.NoError(t, err)
+		assert.True(t, rt.called)
+		assert.Equal(t, "/containers/0123456789ab/json", rt.lastPath)
+		assert.NotContains(t, string(body), "SECRET=1", "sensitive env vars should be removed")
+	})
+
+	t.Run("container logs uses expected endpoint and query params", func(t *testing.T) {
+		rt := &recordingRoundTripper{
+			statusCode: 200,
+			body:       "line1\nline2\n",
+		}
+		dm := &dockerManager{
+			client: &http.Client{Transport: rt},
+		}
+
+		logs, err := dm.getLogs(context.Background(), "abcdef123456")
+		require.NoError(t, err)
+		assert.True(t, rt.called)
+		assert.Equal(t, "/containers/abcdef123456/logs", rt.lastPath)
+		assert.Equal(t, "1", rt.lastQuery["stdout"])
+		assert.Equal(t, "1", rt.lastQuery["stderr"])
+		assert.Equal(t, "200", rt.lastQuery["tail"])
+		assert.Equal(t, "line1\nline2\n", logs)
+	})
+}
+
 func TestValidateCpuPercentage(t *testing.T) {
 	tests := []struct {
 		name          string
@@ -195,48 +299,6 @@ func TestUpdateContainerStatsValues(t *testing.T) {
 	assert.Equal(t, testTime, stats.PrevReadTime)
 }
 
-func TestTwoDecimals(t *testing.T) {
-	tests := []struct {
-		name     string
-		input    float64
-		expected float64
-	}{
-		{"round down", 1.234, 1.23},
-		{"round half up", 1.235, 1.24}, // math.Round rounds half up
-		{"no rounding needed", 1.23, 1.23},
-		{"negative number", -1.235, -1.24}, // math.Round rounds half up (more negative)
-		{"zero", 0.0, 0.0},
-		{"large number", 123.456, 123.46}, // rounds 5 up
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			result := twoDecimals(tt.input)
-			assert.Equal(t, tt.expected, result)
-		})
-	}
-}
-
-func TestBytesToMegabytes(t *testing.T) {
-	tests := []struct {
-		name     string
-		input    float64
-		expected float64
-	}{
-		{"1 MB", 1048576, 1.0},
-		{"512 KB", 524288, 0.5},
-		{"zero", 0, 0},
-		{"large value", 1073741824, 1024}, // 1 GB = 1024 MB
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			result := bytesToMegabytes(tt.input)
-			assert.Equal(t, tt.expected, result)
-		})
-	}
-}
-
 func TestInitializeCpuTracking(t *testing.T) {
 	dm := &dockerManager{
 		lastCpuContainer: make(map[uint16]map[string]uint64),
@@ -379,6 +441,117 @@ func TestDockerManagerCreation(t *testing.T) {
 	assert.NotNil(t, dm.networkRecvTrackers)
 }
 
+func TestCheckDockerVersion(t *testing.T) {
+	tests := []struct {
+		name      string
+		responses []struct {
+			statusCode int
+			body       string
+		}
+		expectedGood     bool
+		expectedRequests int
+	}{
+		{
+			name: "200 with good version on first try",
+			responses: []struct {
+				statusCode int
+				body       string
+			}{
+				{http.StatusOK, `{"Version":"25.0.1"}`},
+			},
+			expectedGood:     true,
+			expectedRequests: 1,
+		},
+		{
+			name: "200 with old version on first try",
+			responses: []struct {
+				statusCode int
+				body       string
+			}{
+				{http.StatusOK, `{"Version":"24.0.7"}`},
+			},
+			expectedGood:     false,
+			expectedRequests: 1,
+		},
+		{
+			name: "non-200 then 200 with good version",
+			responses: []struct {
+				statusCode int
+				body       string
+			}{
+				{http.StatusServiceUnavailable, `"not ready"`},
+				{http.StatusOK, `{"Version":"25.1.0"}`},
+			},
+			expectedGood:     true,
+			expectedRequests: 2,
+		},
+		{
+			name: "non-200 on all retries",
+			responses: []struct {
+				statusCode int
+				body       string
+			}{
+				{http.StatusInternalServerError, `"error"`},
+				{http.StatusUnauthorized, `"error"`},
+			},
+			expectedGood:     false,
+			expectedRequests: 2,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			requestCount := 0
+			server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				idx := requestCount
+				requestCount++
+				if idx >= len(tt.responses) {
+					idx = len(tt.responses) - 1
+				}
+				w.WriteHeader(tt.responses[idx].statusCode)
+				fmt.Fprint(w, tt.responses[idx].body)
+			}))
+			defer server.Close()
+
+			dm := &dockerManager{
+				client: &http.Client{
+					Transport: &http.Transport{
+						DialContext: func(_ context.Context, network, _ string) (net.Conn, error) {
+							return net.Dial(network, server.Listener.Addr().String())
+						},
+					},
+				},
+				retrySleep: func(time.Duration) {},
+			}
+
+			dm.checkDockerVersion()
+
+			assert.Equal(t, tt.expectedGood, dm.goodDockerVersion)
+			assert.Equal(t, tt.expectedRequests, requestCount)
+		})
+	}
+
+	t.Run("request error on all retries", func(t *testing.T) {
+		requestCount := 0
+		dm := &dockerManager{
+			client: &http.Client{
+				Transport: &http.Transport{
+					DialContext: func(_ context.Context, _, _ string) (net.Conn, error) {
+						requestCount++
+						return nil, errors.New("connection refused")
+					},
+				},
+			},
+			retrySleep: func(time.Duration) {},
+		}
+
+		dm.checkDockerVersion()
+
+		assert.False(t, dm.goodDockerVersion)
+		assert.Equal(t, 2, requestCount)
+	})
+}
+
 func TestCycleCpuDeltas(t *testing.T) {
 	dm := &dockerManager{
 		lastCpuContainer: map[uint16]map[string]uint64{
@@ -691,14 +864,50 @@ func TestContainerStatsEndToEndWithRealData(t *testing.T) {
 	updateContainerStatsValues(testStats, cpuPct, usedMemory, 1000000, 500000, testTime)
 
 	assert.Equal(t, cpuPct, testStats.Cpu)
-	assert.Equal(t, bytesToMegabytes(float64(usedMemory)), testStats.Mem)
+	assert.Equal(t, utils.BytesToMegabytes(float64(usedMemory)), testStats.Mem)
 	assert.Equal(t, [2]uint64{1000000, 500000}, testStats.Bandwidth)
 	// Deprecated fields still populated for backward compatibility with older hubs
-	assert.Equal(t, bytesToMegabytes(1000000), testStats.NetworkSent)
-	assert.Equal(t, bytesToMegabytes(500000), testStats.NetworkRecv)
+	assert.Equal(t, utils.BytesToMegabytes(1000000), testStats.NetworkSent)
+	assert.Equal(t, utils.BytesToMegabytes(500000), testStats.NetworkRecv)
 	assert.Equal(t, testTime, testStats.PrevReadTime)
 }
 
+func TestGetLogsDetectsMultiplexedWithoutContentType(t *testing.T) {
+	// Docker multiplexed frame: [stream][0,0,0][len(4 bytes BE)][payload]
+	frame := []byte{
+		0x01, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x05,
+		'H', 'e', 'l', 'l', 'o',
+	}
+	rt := &recordingRoundTripper{
+		statusCode: 200,
+		body:       string(frame),
+		// Intentionally omit content type to simulate Podman behavior.
+	}
+	dm := &dockerManager{
+		client: &http.Client{Transport: rt},
+	}
+
+	logs, err := dm.getLogs(context.Background(), "abcdef123456")
+	require.NoError(t, err)
+	assert.Equal(t, "Hello", logs)
+}
+
+func TestGetLogsDoesNotMisclassifyRawStreamAsMultiplexed(t *testing.T) {
+	// Starts with 0x01, but doesn't match Docker frame signature (reserved bytes aren't all zero).
+	raw := []byte{0x01, 0x02, 0x03, 0x04, 'r', 'a', 'w'}
+	rt := &recordingRoundTripper{
+		statusCode: 200,
+		body:       string(raw),
+	}
+	dm := &dockerManager{
+		client: &http.Client{Transport: rt},
+	}
+
+	logs, err := dm.getLogs(context.Background(), "abcdef123456")
+	require.NoError(t, err)
+	assert.Equal(t, raw, []byte(logs))
+}
+
 func TestEdgeCasesWithRealData(t *testing.T) {
 	// Test with minimal container stats
 	minimalStats := &container.ApiStats{
@@ -940,13 +1149,13 @@ func TestConstantsAndUtilityFunctions(t *testing.T) {
 	assert.Equal(t, 5*1024*1024, maxTotalLogSize)               // 5MB
 
 	// Test utility functions
-	assert.Equal(t, 1.5, twoDecimals(1.499))
-	assert.Equal(t, 1.5, twoDecimals(1.5))
-	assert.Equal(t, 1.5, twoDecimals(1.501))
+	assert.Equal(t, 1.5, utils.TwoDecimals(1.499))
+	assert.Equal(t, 1.5, utils.TwoDecimals(1.5))
+	assert.Equal(t, 1.5, utils.TwoDecimals(1.501))
 
-	assert.Equal(t, 1.0, bytesToMegabytes(1048576)) // 1 MB
-	assert.Equal(t, 0.5, bytesToMegabytes(524288))  // 512 KB
-	assert.Equal(t, 0.0, bytesToMegabytes(0))
+	assert.Equal(t, 1.0, utils.BytesToMegabytes(1048576)) // 1 MB
+	assert.Equal(t, 0.5, utils.BytesToMegabytes(524288))  // 512 KB
+	assert.Equal(t, 0.0, utils.BytesToMegabytes(0))
 }
 
 func TestDecodeDockerLogStream(t *testing.T) {

agent/emmc_linux.go

@@ -8,6 +8,7 @@ import (
 	"strconv"
 	"strings"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/smart"
 )
 
@@ -141,9 +142,9 @@ func readEmmcHealth(blockName string) (emmcHealth, bool) {
 	out.lifeA = lifeA
 	out.lifeB = lifeB
 
-	out.model = readStringFile(filepath.Join(deviceDir, "name"))
-	out.serial = readStringFile(filepath.Join(deviceDir, "serial"))
-	out.revision = readStringFile(filepath.Join(deviceDir, "prv"))
+	out.model = utils.ReadStringFile(filepath.Join(deviceDir, "name"))
+	out.serial = utils.ReadStringFile(filepath.Join(deviceDir, "serial"))
+	out.revision = utils.ReadStringFile(filepath.Join(deviceDir, "prv"))
 
 	if capBytes, ok := readBlockCapacityBytes(blockName); ok {
 		out.capacity = capBytes
@@ -153,7 +154,7 @@ func readEmmcHealth(blockName string) (emmcHealth, bool) {
 }
 
 func readLifeTime(deviceDir string) (uint8, uint8, bool) {
-	if content, ok := readStringFileOK(filepath.Join(deviceDir, "life_time")); ok {
+	if content, ok := utils.ReadStringFileOK(filepath.Join(deviceDir, "life_time")); ok {
 		a, b, ok := parseHexBytePair(content)
 		return a, b, ok
 	}
@@ -170,7 +171,7 @@ func readBlockCapacityBytes(blockName string) (uint64, bool) {
 	sizePath := filepath.Join(emmcSysfsRoot, "class", "block", blockName, "size")
 	lbsPath := filepath.Join(emmcSysfsRoot, "class", "block", blockName, "queue", "logical_block_size")
 
-	sizeStr, ok := readStringFileOK(sizePath)
+	sizeStr, ok := utils.ReadStringFileOK(sizePath)
 	if !ok {
 		return 0, false
 	}
@@ -179,7 +180,7 @@ func readBlockCapacityBytes(blockName string) (uint64, bool) {
 		return 0, false
 	}
 
-	lbsStr, ok := readStringFileOK(lbsPath)
+	lbsStr, ok := utils.ReadStringFileOK(lbsPath)
 	logicalBlockSize := uint64(512)
 	if ok {
 		if parsed, err := strconv.ParseUint(lbsStr, 10, 64); err == nil && parsed > 0 {
@@ -191,7 +192,7 @@ func readBlockCapacityBytes(blockName string) (uint64, bool) {
 }
 
 func readHexByteFile(path string) (uint8, bool) {
-	content, ok := readStringFileOK(path)
+	content, ok := utils.ReadStringFileOK(path)
 	if !ok {
 		return 0, false
 	}
@@ -199,19 +200,6 @@ 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/fingerprint_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/gpu.go

@@ -9,21 +9,25 @@ import (
 	"maps"
 	"os/exec"
 	"regexp"
+	"runtime"
 	"strconv"
 	"strings"
 	"sync"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
 const (
 	// Commands
-	nvidiaSmiCmd  string = "nvidia-smi"
-	rocmSmiCmd    string = "rocm-smi"
-	tegraStatsCmd string = "tegrastats"
-	nvtopCmd      string = "nvtop"
-	noGPUFoundMsg string = "no GPU found - see https://beszel.dev/guide/gpu"
+	nvidiaSmiCmd    string = "nvidia-smi"
+	rocmSmiCmd      string = "rocm-smi"
+	tegraStatsCmd   string = "tegrastats"
+	nvtopCmd        string = "nvtop"
+	powermetricsCmd string = "powermetrics"
+	macmonCmd       string = "macmon"
+	noGPUFoundMsg   string = "no GPU found - see https://beszel.dev/guide/gpu"
 
 	// Command retry and timeout constants
 	retryWaitTime     time.Duration = 5 * time.Second
@@ -82,15 +86,18 @@ var errNoValidData = fmt.Errorf("no valid GPU data found") // Error for missing
 type collectorSource string
 
 const (
-	collectorSourceNVTop       collectorSource = collectorSource(nvtopCmd)
-	collectorSourceNVML        collectorSource = "nvml"
-	collectorSourceNvidiaSMI   collectorSource = collectorSource(nvidiaSmiCmd)
-	collectorSourceIntelGpuTop collectorSource = collectorSource(intelGpuStatsCmd)
-	collectorSourceAmdSysfs    collectorSource = "amd_sysfs"
-	collectorSourceRocmSMI     collectorSource = collectorSource(rocmSmiCmd)
-	collectorGroupNvidia       string          = "nvidia"
-	collectorGroupIntel        string          = "intel"
-	collectorGroupAmd          string          = "amd"
+	collectorSourceNVTop        collectorSource = collectorSource(nvtopCmd)
+	collectorSourceNVML         collectorSource = "nvml"
+	collectorSourceNvidiaSMI    collectorSource = collectorSource(nvidiaSmiCmd)
+	collectorSourceIntelGpuTop  collectorSource = collectorSource(intelGpuStatsCmd)
+	collectorSourceAmdSysfs     collectorSource = "amd_sysfs"
+	collectorSourceRocmSMI      collectorSource = collectorSource(rocmSmiCmd)
+	collectorSourceMacmon       collectorSource = collectorSource(macmonCmd)
+	collectorSourcePowermetrics collectorSource = collectorSource(powermetricsCmd)
+	collectorGroupNvidia        string          = "nvidia"
+	collectorGroupIntel         string          = "intel"
+	collectorGroupAmd           string          = "amd"
+	collectorGroupApple         string          = "apple"
 )
 
 func isValidCollectorSource(source collectorSource) bool {
@@ -100,7 +107,9 @@ func isValidCollectorSource(source collectorSource) bool {
 		collectorSourceNvidiaSMI,
 		collectorSourceIntelGpuTop,
 		collectorSourceAmdSysfs,
-		collectorSourceRocmSMI:
+		collectorSourceRocmSMI,
+		collectorSourceMacmon,
+		collectorSourcePowermetrics:
 		return true
 	}
 	return false
@@ -108,12 +117,14 @@ func isValidCollectorSource(source collectorSource) bool {
 
 // gpuCapabilities describes detected GPU tooling and sysfs support on the host.
 type gpuCapabilities struct {
-	hasNvidiaSmi   bool
-	hasRocmSmi     bool
-	hasAmdSysfs    bool
-	hasTegrastats  bool
-	hasIntelGpuTop bool
-	hasNvtop       bool
+	hasNvidiaSmi    bool
+	hasRocmSmi      bool
+	hasAmdSysfs     bool
+	hasTegrastats   bool
+	hasIntelGpuTop  bool
+	hasNvtop        bool
+	hasMacmon       bool
+	hasPowermetrics bool
 }
 
 type collectorDefinition struct {
@@ -281,8 +292,8 @@ func (gm *GPUManager) parseAmdData(output []byte) bool {
 		}
 		gpu := gm.GpuDataMap[id]
 		gpu.Temperature, _ = strconv.ParseFloat(v.Temperature, 64)
-		gpu.MemoryUsed = bytesToMegabytes(memoryUsage)
-		gpu.MemoryTotal = bytesToMegabytes(totalMemory)
+		gpu.MemoryUsed = utils.BytesToMegabytes(memoryUsage)
+		gpu.MemoryTotal = utils.BytesToMegabytes(totalMemory)
 		gpu.Usage += usage
 		gpu.Power += power
 		gpu.Count++
@@ -356,16 +367,16 @@ func (gm *GPUManager) calculateGPUAverage(id string, gpu *system.GPUData, cacheK
 	gpuAvg := *gpu
 	deltaUsage, deltaPower, deltaPowerPkg := gm.calculateDeltas(gpu, lastSnapshot)
 
-	gpuAvg.Power = twoDecimals(deltaPower / float64(deltaCount))
+	gpuAvg.Power = utils.TwoDecimals(deltaPower / float64(deltaCount))
 
 	if gpu.Engines != nil {
 		// make fresh map for averaged engine metrics to avoid mutating
 		// the accumulator map stored in gm.GpuDataMap
 		gpuAvg.Engines = make(map[string]float64, len(gpu.Engines))
 		gpuAvg.Usage = gm.calculateIntelGPUUsage(&gpuAvg, gpu, lastSnapshot, deltaCount)
-		gpuAvg.PowerPkg = twoDecimals(deltaPowerPkg / float64(deltaCount))
+		gpuAvg.PowerPkg = utils.TwoDecimals(deltaPowerPkg / float64(deltaCount))
 	} else {
-		gpuAvg.Usage = twoDecimals(deltaUsage / float64(deltaCount))
+		gpuAvg.Usage = utils.TwoDecimals(deltaUsage / float64(deltaCount))
 	}
 
 	gm.lastAvgData[id] = gpuAvg
@@ -400,17 +411,17 @@ func (gm *GPUManager) calculateIntelGPUUsage(gpuAvg, gpu *system.GPUData, lastSn
 		} else {
 			deltaEngine = engine
 		}
-		gpuAvg.Engines[name] = twoDecimals(deltaEngine / float64(deltaCount))
+		gpuAvg.Engines[name] = utils.TwoDecimals(deltaEngine / float64(deltaCount))
 		maxEngineUsage = max(maxEngineUsage, deltaEngine/float64(deltaCount))
 	}
-	return twoDecimals(maxEngineUsage)
+	return utils.TwoDecimals(maxEngineUsage)
 }
 
 // updateInstantaneousValues updates values that should reflect current state, not averages
 func (gm *GPUManager) updateInstantaneousValues(gpuAvg *system.GPUData, gpu *system.GPUData) {
-	gpuAvg.Temperature = twoDecimals(gpu.Temperature)
-	gpuAvg.MemoryUsed = twoDecimals(gpu.MemoryUsed)
-	gpuAvg.MemoryTotal = twoDecimals(gpu.MemoryTotal)
+	gpuAvg.Temperature = utils.TwoDecimals(gpu.Temperature)
+	gpuAvg.MemoryUsed = utils.TwoDecimals(gpu.MemoryUsed)
+	gpuAvg.MemoryTotal = utils.TwoDecimals(gpu.MemoryTotal)
 }
 
 // storeSnapshot saves the current GPU state for this cache key
@@ -449,11 +460,19 @@ func (gm *GPUManager) discoverGpuCapabilities() gpuCapabilities {
 	if _, err := exec.LookPath(nvtopCmd); err == nil {
 		caps.hasNvtop = true
 	}
+	if runtime.GOOS == "darwin" {
+		if _, err := exec.LookPath(macmonCmd); err == nil {
+			caps.hasMacmon = true
+		}
+		if _, err := exec.LookPath(powermetricsCmd); err == nil {
+			caps.hasPowermetrics = true
+		}
+	}
 	return caps
 }
 
 func hasAnyGpuCollector(caps gpuCapabilities) bool {
-	return caps.hasNvidiaSmi || caps.hasRocmSmi || caps.hasAmdSysfs || caps.hasTegrastats || caps.hasIntelGpuTop || caps.hasNvtop
+	return caps.hasNvidiaSmi || caps.hasRocmSmi || caps.hasAmdSysfs || caps.hasTegrastats || caps.hasIntelGpuTop || caps.hasNvtop || caps.hasMacmon || caps.hasPowermetrics
 }
 
 func (gm *GPUManager) startIntelCollector() {
@@ -567,6 +586,22 @@ func (gm *GPUManager) collectorDefinitions(caps gpuCapabilities) map[collectorSo
 				return true
 			},
 		},
+		collectorSourceMacmon: {
+			group:     collectorGroupApple,
+			available: caps.hasMacmon,
+			start: func(_ func()) bool {
+				gm.startMacmonCollector()
+				return true
+			},
+		},
+		collectorSourcePowermetrics: {
+			group:     collectorGroupApple,
+			available: caps.hasPowermetrics,
+			start: func(_ func()) bool {
+				gm.startPowermetricsCollector()
+				return true
+			},
+		},
 	}
 }
 
@@ -653,7 +688,7 @@ func (gm *GPUManager) resolveLegacyCollectorPriority(caps gpuCapabilities) []col
 	priorities := make([]collectorSource, 0, 4)
 
 	if caps.hasNvidiaSmi && !caps.hasTegrastats {
-		if nvml, _ := GetEnv("NVML"); nvml == "true" {
+		if nvml, _ := utils.GetEnv("NVML"); nvml == "true" {
 			priorities = append(priorities, collectorSourceNVML, collectorSourceNvidiaSMI)
 		} else {
 			priorities = append(priorities, collectorSourceNvidiaSMI)
@@ -661,7 +696,7 @@ func (gm *GPUManager) resolveLegacyCollectorPriority(caps gpuCapabilities) []col
 	}
 
 	if caps.hasRocmSmi {
-		if val, _ := GetEnv("AMD_SYSFS"); val == "true" {
+		if val, _ := utils.GetEnv("AMD_SYSFS"); val == "true" {
 			priorities = append(priorities, collectorSourceAmdSysfs)
 		} else {
 			priorities = append(priorities, collectorSourceRocmSMI)
@@ -674,7 +709,18 @@ func (gm *GPUManager) resolveLegacyCollectorPriority(caps gpuCapabilities) []col
 		priorities = append(priorities, collectorSourceIntelGpuTop)
 	}
 
-	// Keep nvtop as a legacy last resort only when no vendor collector exists.
+	// Apple collectors are currently opt-in only for testing.
+	// Enable them with GPU_COLLECTOR=macmon or GPU_COLLECTOR=powermetrics.
+	// TODO: uncomment below when Apple collectors are confirmed to be working.
+	//
+	// Prefer macmon on macOS (no sudo). Fall back to powermetrics if present.
+	// if caps.hasMacmon {
+	// 	priorities = append(priorities, collectorSourceMacmon)
+	// } else if caps.hasPowermetrics {
+	// 	priorities = append(priorities, collectorSourcePowermetrics)
+	// }
+
+	// Keep nvtop as a last resort only when no vendor collector exists.
 	if len(priorities) == 0 && caps.hasNvtop {
 		priorities = append(priorities, collectorSourceNVTop)
 	}
@@ -683,7 +729,7 @@ func (gm *GPUManager) resolveLegacyCollectorPriority(caps gpuCapabilities) []col
 
 // NewGPUManager creates and initializes a new GPUManager
 func NewGPUManager() (*GPUManager, error) {
-	if skipGPU, _ := GetEnv("SKIP_GPU"); skipGPU == "true" {
+	if skipGPU, _ := utils.GetEnv("SKIP_GPU"); skipGPU == "true" {
 		return nil, nil
 	}
 	var gm GPUManager
@@ -700,7 +746,7 @@ func NewGPUManager() (*GPUManager, error) {
 	}
 
 	// if GPU_COLLECTOR is set, start user-defined collectors.
-	if collectorConfig, ok := GetEnv("GPU_COLLECTOR"); ok && strings.TrimSpace(collectorConfig) != "" {
+	if collectorConfig, ok := utils.GetEnv("GPU_COLLECTOR"); ok && strings.TrimSpace(collectorConfig) != "" {
 		priorities := parseCollectorPriority(collectorConfig)
 		if gm.startCollectorsByPriority(priorities, caps) == 0 {
 			return nil, fmt.Errorf("no configured GPU collectors are available")

agent/gpu_amd_linux.go

@@ -13,6 +13,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
@@ -32,8 +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
 		}
 	}
@@ -87,12 +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.
@@ -103,10 +103,8 @@ func (gm *GPUManager) updateAmdGpuData(cardPath string) bool {
 
 	// Read all sysfs values first (no lock needed - these can be slow)
 	usage, usageErr := readSysfsFloat(filepath.Join(devicePath, "gpu_busy_percent"))
-	vramUsed, memUsedErr := readSysfsFloat(filepath.Join(devicePath, "mem_info_vram_used"))
-	vramTotal, _ := readSysfsFloat(filepath.Join(devicePath, "mem_info_vram_total"))
-	memUsed := vramUsed
-	memTotal := vramTotal
+	memUsed, memUsedErr := readSysfsFloat(filepath.Join(devicePath, "mem_info_vram_used"))
+	memTotal, _ := readSysfsFloat(filepath.Join(devicePath, "mem_info_vram_total"))
 	// if gtt is present, add it to the memory used and total (https://github.com/henrygd/beszel/issues/1569#issuecomment-3837640484)
 	if gttUsed, err := readSysfsFloat(filepath.Join(devicePath, "mem_info_gtt_used")); err == nil && gttUsed > 0 {
 		if gttTotal, err := readSysfsFloat(filepath.Join(devicePath, "mem_info_gtt_total")); err == nil {
@@ -146,8 +144,8 @@ func (gm *GPUManager) updateAmdGpuData(cardPath string) bool {
 	if usageErr == nil {
 		gpu.Usage += usage
 	}
-	gpu.MemoryUsed = bytesToMegabytes(memUsed)
-	gpu.MemoryTotal = bytesToMegabytes(memTotal)
+	gpu.MemoryUsed = utils.BytesToMegabytes(memUsed)
+	gpu.MemoryTotal = utils.BytesToMegabytes(memTotal)
 	gpu.Temperature = temp
 	gpu.Power += power
 	gpu.Count++
@@ -156,11 +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.
@@ -243,7 +241,10 @@ func getCachedAmdgpuName(deviceID, revisionID string) (name string, found bool,
 
 // normalizeAmdgpuName trims standard suffixes from AMDGPU product names.
 func normalizeAmdgpuName(name string) string {
-	return strings.TrimSuffix(strings.TrimSpace(name), " Graphics")
+	for _, suffix := range []string{" Graphics", " Series"} {
+		name = strings.TrimSuffix(name, suffix)
+	}
+	return name
 }
 
 // cacheAmdgpuName stores a resolved AMDGPU name in the lookup cache.
@@ -272,16 +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/gpu_amd_linux_test.go

@@ -7,6 +7,7 @@ import (
 	"path/filepath"
 	"testing"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -128,14 +129,14 @@ func TestUpdateAmdGpuDataWithFakeSysfs(t *testing.T) {
 		{
 			name:            "sums vram and gtt when gtt is present",
 			writeGTT:        true,
-			wantMemoryUsed:  bytesToMegabytes(1073741824 + 536870912),
-			wantMemoryTotal: bytesToMegabytes(2147483648 + 4294967296),
+			wantMemoryUsed:  utils.BytesToMegabytes(1073741824 + 536870912),
+			wantMemoryTotal: utils.BytesToMegabytes(2147483648 + 4294967296),
 		},
 		{
 			name:            "falls back to vram when gtt is missing",
 			writeGTT:        false,
-			wantMemoryUsed:  bytesToMegabytes(1073741824),
-			wantMemoryTotal: bytesToMegabytes(2147483648),
+			wantMemoryUsed:  utils.BytesToMegabytes(1073741824),
+			wantMemoryTotal: utils.BytesToMegabytes(2147483648),
 		},
 	}
 

agent/gpu_darwin.go

@@ -0,0 +1,252 @@
+//go:build darwin
+
+package agent
+
+import (
+	"bufio"
+	"bytes"
+	"encoding/json"
+	"io"
+	"log/slog"
+	"os/exec"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/henrygd/beszel/internal/entities/system"
+)
+
+const (
+	// powermetricsSampleIntervalMs is the sampling interval passed to powermetrics (-i).
+	powermetricsSampleIntervalMs = 500
+	// powermetricsPollInterval is how often we run powermetrics to collect a new sample.
+	powermetricsPollInterval = 2 * time.Second
+	// macmonIntervalMs is the sampling interval passed to macmon pipe (-i), in milliseconds.
+	macmonIntervalMs = 2500
+)
+
+const appleGPUID = "0"
+
+// startPowermetricsCollector runs powermetrics --samplers gpu_power in a loop and updates
+// GPU usage and power. Requires root (sudo) on macOS. A single logical GPU is reported as id "0".
+func (gm *GPUManager) startPowermetricsCollector() {
+	// Ensure single GPU entry for Apple GPU
+	if _, ok := gm.GpuDataMap[appleGPUID]; !ok {
+		gm.GpuDataMap[appleGPUID] = &system.GPUData{Name: "Apple GPU"}
+	}
+
+	go func() {
+		failures := 0
+		for {
+			if err := gm.collectPowermetrics(); err != nil {
+				failures++
+				if failures > maxFailureRetries {
+					slog.Warn("powermetrics GPU collector failed repeatedly, stopping", "err", err)
+					break
+				}
+				slog.Warn("Error collecting macOS GPU data via powermetrics (may require sudo)", "err", err)
+				time.Sleep(retryWaitTime)
+				continue
+			}
+			failures = 0
+			time.Sleep(powermetricsPollInterval)
+		}
+	}()
+}
+
+// collectPowermetrics runs powermetrics once and parses GPU usage and power from its output.
+func (gm *GPUManager) collectPowermetrics() error {
+	interval := strconv.Itoa(powermetricsSampleIntervalMs)
+	cmd := exec.Command(powermetricsCmd, "--samplers", "gpu_power", "-i", interval, "-n", "1")
+	cmd.Stderr = nil
+	out, err := cmd.Output()
+	if err != nil {
+		return err
+	}
+	if !gm.parsePowermetricsData(out) {
+		return errNoValidData
+	}
+	return nil
+}
+
+// parsePowermetricsData parses powermetrics gpu_power output and updates GpuDataMap["0"].
+// Example output:
+//
+//	**** GPU usage ****
+//	GPU HW active frequency: 444 MHz
+//	GPU HW active residency:   0.97% (444 MHz: .97% ...
+//	GPU idle residency:  99.03%
+//	GPU Power: 4 mW
+func (gm *GPUManager) parsePowermetricsData(output []byte) bool {
+	var idleResidency, powerMW float64
+	var gotIdle, gotPower bool
+
+	scanner := bufio.NewScanner(bytes.NewReader(output))
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if strings.HasPrefix(line, "GPU idle residency:") {
+			// "GPU idle residency:  99.03%"
+			fields := strings.Fields(strings.TrimPrefix(line, "GPU idle residency:"))
+			if len(fields) >= 1 {
+				pct := strings.TrimSuffix(fields[0], "%")
+				if v, err := strconv.ParseFloat(pct, 64); err == nil {
+					idleResidency = v
+					gotIdle = true
+				}
+			}
+		} else if strings.HasPrefix(line, "GPU Power:") {
+			// "GPU Power: 4 mW"
+			fields := strings.Fields(strings.TrimPrefix(line, "GPU Power:"))
+			if len(fields) >= 1 {
+				if v, err := strconv.ParseFloat(fields[0], 64); err == nil {
+					powerMW = v
+					gotPower = true
+				}
+			}
+		}
+	}
+	if err := scanner.Err(); err != nil {
+		return false
+	}
+	if !gotIdle && !gotPower {
+		return false
+	}
+
+	gm.Lock()
+	defer gm.Unlock()
+
+	if _, ok := gm.GpuDataMap[appleGPUID]; !ok {
+		gm.GpuDataMap[appleGPUID] = &system.GPUData{Name: "Apple GPU"}
+	}
+	gpu := gm.GpuDataMap[appleGPUID]
+
+	if gotIdle {
+		// Usage = 100 - idle residency (e.g. 100 - 99.03 = 0.97%)
+		gpu.Usage += 100 - idleResidency
+	}
+	if gotPower {
+		// mW -> W
+		gpu.Power += powerMW / milliwattsInAWatt
+	}
+	gpu.Count++
+	return true
+}
+
+// startMacmonCollector runs `macmon pipe` in a loop and parses one JSON object per line.
+// This collector does not require sudo. A single logical GPU is reported as id "0".
+func (gm *GPUManager) startMacmonCollector() {
+	if _, ok := gm.GpuDataMap[appleGPUID]; !ok {
+		gm.GpuDataMap[appleGPUID] = &system.GPUData{Name: "Apple GPU"}
+	}
+
+	go func() {
+		failures := 0
+		for {
+			if err := gm.collectMacmonPipe(); err != nil {
+				failures++
+				if failures > maxFailureRetries {
+					slog.Warn("macmon GPU collector failed repeatedly, stopping", "err", err)
+					break
+				}
+				slog.Warn("Error collecting macOS GPU data via macmon", "err", err)
+				time.Sleep(retryWaitTime)
+				continue
+			}
+			failures = 0
+			// `macmon pipe` is long-running; if it returns, wait a bit before restarting.
+			time.Sleep(retryWaitTime)
+		}
+	}()
+}
+
+type macmonTemp struct {
+	GPUTempAvg float64 `json:"gpu_temp_avg"`
+}
+
+type macmonSample struct {
+	GPUPower    float64    `json:"gpu_power"`     // watts (macmon reports fractional values)
+	GPURAMPower float64    `json:"gpu_ram_power"` // watts
+	GPUUsage    []float64  `json:"gpu_usage"`     // [freq_mhz, usage] where usage is typically 0..1
+	Temp        macmonTemp `json:"temp"`
+}
+
+func (gm *GPUManager) collectMacmonPipe() (err error) {
+	cmd := exec.Command(macmonCmd, "pipe", "-i", strconv.Itoa(macmonIntervalMs))
+	// Avoid blocking if macmon writes to stderr.
+	cmd.Stderr = io.Discard
+	stdout, err := cmd.StdoutPipe()
+	if err != nil {
+		return err
+	}
+	if err := cmd.Start(); err != nil {
+		return err
+	}
+
+	// Ensure we always reap the child to avoid zombies on any return path and
+	// propagate a non-zero exit code if no other error was set.
+	defer func() {
+		_ = stdout.Close()
+		if cmd.ProcessState == nil || !cmd.ProcessState.Exited() {
+			_ = cmd.Process.Kill()
+		}
+		if waitErr := cmd.Wait(); err == nil && waitErr != nil {
+			err = waitErr
+		}
+	}()
+
+	scanner := bufio.NewScanner(stdout)
+	var hadSample bool
+	for scanner.Scan() {
+		line := bytes.TrimSpace(scanner.Bytes())
+		if len(line) == 0 {
+			continue
+		}
+		if gm.parseMacmonLine(line) {
+			hadSample = true
+		}
+	}
+	if scanErr := scanner.Err(); scanErr != nil {
+		return scanErr
+	}
+	if !hadSample {
+		return errNoValidData
+	}
+	return nil
+}
+
+// parseMacmonLine parses a single macmon JSON line and updates Apple GPU metrics.
+func (gm *GPUManager) parseMacmonLine(line []byte) bool {
+	var sample macmonSample
+	if err := json.Unmarshal(line, &sample); err != nil {
+		return false
+	}
+
+	usage := 0.0
+	if len(sample.GPUUsage) >= 2 {
+		usage = sample.GPUUsage[1]
+		// Heuristic: macmon typically reports 0..1; convert to percentage.
+		if usage <= 1.0 {
+			usage *= 100
+		}
+	}
+
+	// Consider the line valid if it contains at least one GPU metric.
+	if usage == 0 && sample.GPUPower == 0 && sample.Temp.GPUTempAvg == 0 {
+		return false
+	}
+
+	gm.Lock()
+	defer gm.Unlock()
+
+	gpu, ok := gm.GpuDataMap[appleGPUID]
+	if !ok {
+		gpu = &system.GPUData{Name: "Apple GPU"}
+		gm.GpuDataMap[appleGPUID] = gpu
+	}
+	gpu.Temperature = sample.Temp.GPUTempAvg
+	gpu.Usage += usage
+	// macmon reports power in watts; include VRAM power if present.
+	gpu.Power += sample.GPUPower + sample.GPURAMPower
+	gpu.Count++
+	return true
+}

agent/gpu_darwin_test.go

@@ -0,0 +1,81 @@
+//go:build darwin
+
+package agent
+
+import (
+	"testing"
+
+	"github.com/henrygd/beszel/internal/entities/system"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestParsePowermetricsData(t *testing.T) {
+	input := `
+Machine model: Mac14,10
+OS version: 25D125
+
+*** Sampled system activity (Sat Feb 14 00:42:06 2026 -0500) (503.05ms elapsed) ***
+
+**** GPU usage ****
+
+GPU HW active frequency: 444 MHz
+GPU HW active residency:   0.97% (444 MHz: .97% 612 MHz:   0% 808 MHz:   0% 968 MHz:   0% 1110 MHz:   0% 1236 MHz:   0% 1338 MHz:   0% 1398 MHz:   0%)
+GPU SW requested state: (P1 : 100% P2 :   0% P3 :   0% P4 :   0% P5 :   0% P6 :   0% P7 :   0% P8 :   0%)
+GPU idle residency:  99.03%
+GPU Power: 4 mW
+`
+	gm := &GPUManager{
+		GpuDataMap: make(map[string]*system.GPUData),
+	}
+	valid := gm.parsePowermetricsData([]byte(input))
+	require.True(t, valid)
+
+	g0, ok := gm.GpuDataMap["0"]
+	require.True(t, ok)
+	assert.Equal(t, "Apple GPU", g0.Name)
+	// Usage = 100 - 99.03 = 0.97
+	assert.InDelta(t, 0.97, g0.Usage, 0.01)
+	// 4 mW -> 0.004 W
+	assert.InDelta(t, 0.004, g0.Power, 0.0001)
+	assert.Equal(t, 1.0, g0.Count)
+}
+
+func TestParsePowermetricsDataPartial(t *testing.T) {
+	// Only power line (e.g. older macOS or different sampler output)
+	input := `
+**** GPU usage ****
+GPU Power: 120 mW
+`
+	gm := &GPUManager{
+		GpuDataMap: make(map[string]*system.GPUData),
+	}
+	valid := gm.parsePowermetricsData([]byte(input))
+	require.True(t, valid)
+
+	g0, ok := gm.GpuDataMap["0"]
+	require.True(t, ok)
+	assert.Equal(t, "Apple GPU", g0.Name)
+	assert.InDelta(t, 0.12, g0.Power, 0.001)
+	assert.Equal(t, 1.0, g0.Count)
+}
+
+func TestParseMacmonLine(t *testing.T) {
+	input := `{"all_power":0.6468324661254883,"ane_power":0.0,"cpu_power":0.6359732151031494,"ecpu_usage":[2061,0.1726151406764984],"gpu_power":0.010859241709113121,"gpu_ram_power":0.000965250947047025,"gpu_usage":[503,0.013633215799927711],"memory":{"ram_total":17179869184,"ram_usage":12322914304,"swap_total":0,"swap_usage":0},"pcpu_usage":[1248,0.11792058497667313],"ram_power":0.14885640144348145,"sys_power":10.4955415725708,"temp":{"cpu_temp_avg":23.041261672973633,"gpu_temp_avg":29.44516944885254},"timestamp":"2026-02-17T19:34:27.942556+00:00"}`
+
+	gm := &GPUManager{
+		GpuDataMap: make(map[string]*system.GPUData),
+	}
+	valid := gm.parseMacmonLine([]byte(input))
+	require.True(t, valid)
+
+	g0, ok := gm.GpuDataMap["0"]
+	require.True(t, ok)
+	assert.Equal(t, "Apple GPU", g0.Name)
+	// macmon reports usage fraction 0..1; expect percent conversion.
+	assert.InDelta(t, 1.3633, g0.Usage, 0.05)
+	// power includes gpu_power + gpu_ram_power
+	assert.InDelta(t, 0.011824, g0.Power, 0.0005)
+	assert.InDelta(t, 29.445, g0.Temperature, 0.01)
+	assert.Equal(t, 1.0, g0.Count)
+}

agent/gpu_darwin_unsupported.go

@@ -0,0 +1,9 @@
+//go:build !darwin
+
+package agent
+
+// startPowermetricsCollector is a no-op on non-darwin platforms; the real implementation is in gpu_darwin.go.
+func (gm *GPUManager) startPowermetricsCollector() {}
+
+// startMacmonCollector is a no-op on non-darwin platforms; the real implementation is in gpu_darwin.go.
+func (gm *GPUManager) startMacmonCollector() {}

agent/gpu_intel.go

@@ -7,6 +7,7 @@ import (
 	"strconv"
 	"strings"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
@@ -52,7 +53,7 @@ func (gm *GPUManager) updateIntelFromStats(sample *intelGpuStats) bool {
 func (gm *GPUManager) collectIntelStats() (err error) {
 	// Build command arguments, optionally selecting a device via -d
 	args := []string{"-s", intelGpuStatsInterval, "-l"}
-	if dev, ok := GetEnv("INTEL_GPU_DEVICE"); ok && dev != "" {
+	if dev, ok := utils.GetEnv("INTEL_GPU_DEVICE"); ok && dev != "" {
 		args = append(args, "-d", dev)
 	}
 	cmd := exec.Command(intelGpuStatsCmd, args...)

agent/gpu_nvtop.go

@@ -9,6 +9,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
@@ -80,10 +81,10 @@ func (gm *GPUManager) updateNvtopSnapshots(snapshots []nvtopSnapshot) bool {
 			gpu.Temperature = parseNvtopNumber(*sample.Temp)
 		}
 		if sample.MemUsed != nil {
-			gpu.MemoryUsed = bytesToMegabytes(parseNvtopNumber(*sample.MemUsed))
+			gpu.MemoryUsed = utils.BytesToMegabytes(parseNvtopNumber(*sample.MemUsed))
 		}
 		if sample.MemTotal != nil {
-			gpu.MemoryTotal = bytesToMegabytes(parseNvtopNumber(*sample.MemTotal))
+			gpu.MemoryTotal = utils.BytesToMegabytes(parseNvtopNumber(*sample.MemTotal))
 		}
 		if sample.GpuUtil != nil {
 			gpu.Usage += parseNvtopNumber(*sample.GpuUtil)

agent/gpu_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 
@@ -11,6 +10,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 
 	"github.com/stretchr/testify/assert"
@@ -266,8 +266,8 @@ func TestParseNvtopData(t *testing.T) {
 	assert.Equal(t, 48.0, g0.Temperature)
 	assert.Equal(t, 5.0, g0.Usage)
 	assert.Equal(t, 13.0, g0.Power)
-	assert.Equal(t, bytesToMegabytes(349372416), g0.MemoryUsed)
-	assert.Equal(t, bytesToMegabytes(4294967296), g0.MemoryTotal)
+	assert.Equal(t, utils.BytesToMegabytes(349372416), g0.MemoryUsed)
+	assert.Equal(t, utils.BytesToMegabytes(4294967296), g0.MemoryTotal)
 	assert.Equal(t, 1.0, g0.Count)
 
 	g1, ok := gm.GpuDataMap["n1"]
@@ -276,8 +276,8 @@ func TestParseNvtopData(t *testing.T) {
 	assert.Equal(t, 48.0, g1.Temperature)
 	assert.Equal(t, 12.0, g1.Usage)
 	assert.Equal(t, 9.0, g1.Power)
-	assert.Equal(t, bytesToMegabytes(1213784064), g1.MemoryUsed)
-	assert.Equal(t, bytesToMegabytes(16929173504), g1.MemoryTotal)
+	assert.Equal(t, utils.BytesToMegabytes(1213784064), g1.MemoryUsed)
+	assert.Equal(t, utils.BytesToMegabytes(16929173504), g1.MemoryTotal)
 	assert.Equal(t, 1.0, g1.Count)
 }
 

agent/handlers_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/health/health_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package health
 
@@ -37,7 +36,6 @@ func TestHealth(t *testing.T) {
 	})
 
 	// This test uses synctest to simulate time passing.
-	// NOTE: This test requires GOEXPERIMENT=synctest to run.
 	t.Run("check with simulated time", func(t *testing.T) {
 		synctest.Test(t, func(t *testing.T) {
 			// Update the file to set the initial timestamp.

agent/mdraid_linux.go

@@ -0,0 +1,233 @@
+//go:build linux
+
+package agent
+
+import (
+	"fmt"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"github.com/henrygd/beszel/agent/utils"
+	"github.com/henrygd/beszel/internal/entities/smart"
+)
+
+// mdraidSysfsRoot is a test hook; production value is "/sys".
+var mdraidSysfsRoot = "/sys"
+
+type mdraidHealth struct {
+	level         string
+	arrayState    string
+	degraded      uint64
+	raidDisks     uint64
+	syncAction    string
+	syncCompleted string
+	syncSpeed     string
+	mismatchCnt   uint64
+	capacity      uint64
+}
+
+// scanMdraidDevices discovers Linux md arrays exposed in sysfs.
+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 !utils.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
+}
+
+// collectMdraidHealth reads mdraid health and stores it in SmartDataMap.
+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.Capacity = health.capacity
+	data.SmartStatus = status
+	data.DiskName = filepath.Join("/dev", base)
+	data.DiskType = "mdraid"
+	data.Attributes = attrs
+
+	return true, nil
+}
+
+// readMdraidHealth reads md array health fields from sysfs.
+func readMdraidHealth(blockName string) (mdraidHealth, bool) {
+	var out mdraidHealth
+
+	if !isMdraidBlockName(blockName) {
+		return out, false
+	}
+
+	mdDir := filepath.Join(mdraidSysfsRoot, "block", blockName, "md")
+	arrayState, okState := utils.ReadStringFileOK(filepath.Join(mdDir, "array_state"))
+	if !okState {
+		return out, false
+	}
+
+	out.arrayState = arrayState
+	out.level = utils.ReadStringFile(filepath.Join(mdDir, "level"))
+	out.syncAction = utils.ReadStringFile(filepath.Join(mdDir, "sync_action"))
+	out.syncCompleted = utils.ReadStringFile(filepath.Join(mdDir, "sync_completed"))
+	out.syncSpeed = utils.ReadStringFile(filepath.Join(mdDir, "sync_speed"))
+
+	if val, ok := utils.ReadUintFile(filepath.Join(mdDir, "raid_disks")); ok {
+		out.raidDisks = val
+	}
+	if val, ok := utils.ReadUintFile(filepath.Join(mdDir, "degraded")); ok {
+		out.degraded = val
+	}
+	if val, ok := utils.ReadUintFile(filepath.Join(mdDir, "mismatch_cnt")); ok {
+		out.mismatchCnt = val
+	}
+
+	if capBytes, ok := readMdraidBlockCapacityBytes(blockName, mdraidSysfsRoot); ok {
+		out.capacity = capBytes
+	}
+
+	return out, true
+}
+
+// mdraidSmartStatus maps md state/sync signals to a SMART-like status.
+func mdraidSmartStatus(health mdraidHealth) string {
+	state := strings.ToLower(strings.TrimSpace(health.arrayState))
+	switch state {
+	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 syncAction {
+	case "check", "repair":
+		return "WARNING"
+	}
+	switch state {
+	case "clean", "active", "active-idle", "write-pending", "read-auto", "readonly":
+		return "PASSED"
+	}
+	return "UNKNOWN"
+}
+
+// isMdraidBlockName matches /dev/mdN-style block device names.
+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
+}
+
+// readMdraidBlockCapacityBytes converts block size metadata into bytes.
+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 := utils.ReadStringFileOK(sizePath)
+	if !ok {
+		return 0, false
+	}
+	sectors, err := strconv.ParseUint(sizeStr, 10, 64)
+	if err != nil || sectors == 0 {
+		return 0, false
+	}
+
+	logicalBlockSize := uint64(512)
+	if lbsStr, ok := utils.ReadStringFileOK(lbsPath); ok {
+		if parsed, err := strconv.ParseUint(lbsStr, 10, 64); err == nil && parsed > 0 {
+			logicalBlockSize = parsed
+		}
+	}
+
+	return sectors * logicalBlockSize, true
+}

agent/mdraid_linux_test.go

@@ -0,0 +1,103 @@
+//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, 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)
+	}
+	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/network.go

@@ -8,6 +8,7 @@ import (
 	"time"
 
 	"github.com/henrygd/beszel/agent/deltatracker"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 	psutilNet "github.com/shirou/gopsutil/v4/net"
 )
@@ -94,7 +95,7 @@ func (a *Agent) initializeNetIoStats() {
 	a.netInterfaces = make(map[string]struct{}, 0)
 
 	// parse NICS env var for whitelist / blacklist
-	nicsEnvVal, nicsEnvExists := GetEnv("NICS")
+	nicsEnvVal, nicsEnvExists := utils.GetEnv("NICS")
 	var nicCfg *NicConfig
 	if nicsEnvExists {
 		nicCfg = newNicConfig(nicsEnvVal)
@@ -103,10 +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)
@@ -215,10 +213,8 @@ func (a *Agent) applyNetworkTotals(
 	totalBytesSent, totalBytesRecv uint64,
 	bytesSentPerSecond, bytesRecvPerSecond uint64,
 ) {
-	networkSentPs := bytesToMegabytes(float64(bytesSentPerSecond))
-	networkRecvPs := bytesToMegabytes(float64(bytesRecvPerSecond))
-	if networkSentPs > 10_000 || networkRecvPs > 10_000 {
-		slog.Warn("Invalid net stats. Resetting.", "sent", networkSentPs, "recv", networkRecvPs)
+	if bytesSentPerSecond > 10_000_000_000 || bytesRecvPerSecond > 10_000_000_000 {
+		slog.Warn("Invalid net stats. Resetting.", "sent", bytesSentPerSecond, "recv", bytesRecvPerSecond)
 		for _, v := range netIO {
 			if _, exists := a.netInterfaces[v.Name]; !exists {
 				continue
@@ -228,21 +224,29 @@ func (a *Agent) applyNetworkTotals(
 		a.initializeNetIoStats()
 		delete(a.netIoStats, cacheTimeMs)
 		delete(a.netInterfaceDeltaTrackers, cacheTimeMs)
-		systemStats.NetworkSent = 0
-		systemStats.NetworkRecv = 0
 		systemStats.Bandwidth[0], systemStats.Bandwidth[1] = 0, 0
 		return
 	}
 
-	systemStats.NetworkSent = networkSentPs
-	systemStats.NetworkRecv = networkRecvPs
 	systemStats.Bandwidth[0], systemStats.Bandwidth[1] = bytesSentPerSecond, bytesRecvPerSecond
 	nis.BytesSent = totalBytesSent
 	nis.BytesRecv = totalBytesRecv
 	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
@@ -383,8 +416,6 @@ func TestApplyNetworkTotals(t *testing.T) {
 		totalBytesSent        uint64
 		totalBytesRecv        uint64
 		expectReset           bool
-		expectedNetworkSent   float64
-		expectedNetworkRecv   float64
 		expectedBandwidthSent uint64
 		expectedBandwidthRecv uint64
 	}{
@@ -395,8 +426,6 @@ func TestApplyNetworkTotals(t *testing.T) {
 			totalBytesSent:        10000000,
 			totalBytesRecv:        20000000,
 			expectReset:           false,
-			expectedNetworkSent:   0.95, // ~1 MB/s rounded to 2 decimals
-			expectedNetworkRecv:   1.91, // ~2 MB/s rounded to 2 decimals
 			expectedBandwidthSent: 1000000,
 			expectedBandwidthRecv: 2000000,
 		},
@@ -424,18 +453,6 @@ func TestApplyNetworkTotals(t *testing.T) {
 			totalBytesRecv:     20000000,
 			expectReset:        true,
 		},
-		{
-			name:                  "Valid network stats - at threshold boundary",
-			bytesSentPerSecond:    10485750000, // ~9999.99 MB/s (rounds to 9999.99)
-			bytesRecvPerSecond:    10485750000, // ~9999.99 MB/s (rounds to 9999.99)
-			totalBytesSent:        10000000,
-			totalBytesRecv:        20000000,
-			expectReset:           false,
-			expectedNetworkSent:   9999.99,
-			expectedNetworkRecv:   9999.99,
-			expectedBandwidthSent: 10485750000,
-			expectedBandwidthRecv: 10485750000,
-		},
 		{
 			name:                  "Zero values",
 			bytesSentPerSecond:    0,
@@ -443,8 +460,6 @@ func TestApplyNetworkTotals(t *testing.T) {
 			totalBytesSent:        0,
 			totalBytesRecv:        0,
 			expectReset:           false,
-			expectedNetworkSent:   0.0,
-			expectedNetworkRecv:   0.0,
 			expectedBandwidthSent: 0,
 			expectedBandwidthRecv: 0,
 		},
@@ -481,14 +496,10 @@ func TestApplyNetworkTotals(t *testing.T) {
 				// Should have reset network tracking state - maps cleared and stats zeroed
 				assert.NotContains(t, a.netIoStats, cacheTimeMs, "cache entry should be cleared after reset")
 				assert.NotContains(t, a.netInterfaceDeltaTrackers, cacheTimeMs, "tracker should be cleared on reset")
-				assert.Zero(t, systemStats.NetworkSent)
-				assert.Zero(t, systemStats.NetworkRecv)
 				assert.Zero(t, systemStats.Bandwidth[0])
 				assert.Zero(t, systemStats.Bandwidth[1])
 			} else {
 				// Should have applied stats
-				assert.Equal(t, tt.expectedNetworkSent, systemStats.NetworkSent)
-				assert.Equal(t, tt.expectedNetworkRecv, systemStats.NetworkRecv)
 				assert.Equal(t, tt.expectedBandwidthSent, systemStats.Bandwidth[0])
 				assert.Equal(t, tt.expectedBandwidthRecv, systemStats.Bandwidth[1])
 

agent/sensors.go

@@ -10,6 +10,7 @@ import (
 	"strings"
 	"unicode/utf8"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 
 	"github.com/shirou/gopsutil/v4/common"
@@ -26,9 +27,9 @@ type SensorConfig struct {
 }
 
 func (a *Agent) newSensorConfig() *SensorConfig {
-	primarySensor, _ := GetEnv("PRIMARY_SENSOR")
-	sysSensors, _ := GetEnv("SYS_SENSORS")
-	sensorsEnvVal, sensorsSet := GetEnv("SENSORS")
+	primarySensor, _ := utils.GetEnv("PRIMARY_SENSOR")
+	sysSensors, _ := utils.GetEnv("SYS_SENSORS")
+	sensorsEnvVal, sensorsSet := utils.GetEnv("SENSORS")
 	skipCollection := sensorsSet && sensorsEnvVal == ""
 
 	return a.newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal, skipCollection)
@@ -135,7 +136,7 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 		case sensorName:
 			a.systemInfo.DashboardTemp = sensor.Temperature
 		}
-		systemStats.Temperatures[sensorName] = twoDecimals(sensor.Temperature)
+		systemStats.Temperatures[sensorName] = utils.TwoDecimals(sensor.Temperature)
 	}
 }
 

agent/sensors_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/server.go

@@ -12,6 +12,7 @@ import (
 	"time"
 
 	"github.com/henrygd/beszel"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/common"
 	"github.com/henrygd/beszel/internal/entities/system"
 
@@ -36,7 +37,7 @@ var hubVersions map[string]semver.Version
 // and begins listening for connections. Returns an error if the server
 // is already running or if there's an issue starting the server.
 func (a *Agent) StartServer(opts ServerOptions) error {
-	if disableSSH, _ := GetEnv("DISABLE_SSH"); disableSSH == "true" {
+	if disableSSH, _ := utils.GetEnv("DISABLE_SSH"); disableSSH == "true" {
 		return errors.New("SSH disabled")
 	}
 	if a.server != nil {
@@ -238,11 +239,11 @@ func ParseKeys(input string) ([]gossh.PublicKey, error) {
 // and finally defaults to ":45876".
 func GetAddress(addr string) string {
 	if addr == "" {
-		addr, _ = GetEnv("LISTEN")
+		addr, _ = utils.GetEnv("LISTEN")
 	}
 	if addr == "" {
 		// Legacy PORT environment variable support
-		addr, _ = GetEnv("PORT")
+		addr, _ = utils.GetEnv("PORT")
 	}
 	if addr == "" {
 		return ":45876"
@@ -258,7 +259,7 @@ func GetAddress(addr string) string {
 // It checks the NETWORK environment variable first, then infers from
 // the address format: addresses starting with "/" are "unix", others are "tcp".
 func GetNetwork(addr string) string {
-	if network, ok := GetEnv("NETWORK"); ok && network != "" {
+	if network, ok := utils.GetEnv("NETWORK"); ok && network != "" {
 		return network
 	}
 	if strings.HasPrefix(addr, "/") {

agent/server_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 

agent/smart.go

@@ -18,6 +18,7 @@ import (
 	"sync"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/smart"
 )
 
@@ -156,7 +157,7 @@ func (sm *SmartManager) ScanDevices(force bool) error {
 	currentDevices := sm.devicesSnapshot()
 
 	var configuredDevices []*DeviceInfo
-	if configuredRaw, ok := GetEnv("SMART_DEVICES"); ok {
+	if configuredRaw, ok := utils.GetEnv("SMART_DEVICES"); ok {
 		slog.Info("SMART_DEVICES", "value", configuredRaw)
 		config := strings.TrimSpace(configuredRaw)
 		if config == "" {
@@ -199,6 +200,13 @@ 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)
@@ -215,7 +223,7 @@ func (sm *SmartManager) ScanDevices(force bool) error {
 }
 
 func (sm *SmartManager) parseConfiguredDevices(config string) ([]*DeviceInfo, error) {
-	splitChar := os.Getenv("SMART_DEVICES_SEPARATOR")
+	splitChar, _ := utils.GetEnv("SMART_DEVICES_SEPARATOR")
 	if splitChar == "" {
 		splitChar = ","
 	}
@@ -253,7 +261,7 @@ func (sm *SmartManager) parseConfiguredDevices(config string) ([]*DeviceInfo, er
 }
 
 func (sm *SmartManager) refreshExcludedDevices() {
-	rawValue, _ := GetEnv("EXCLUDE_SMART")
+	rawValue, _ := utils.GetEnv("EXCLUDE_SMART")
 	sm.excludedDevices = make(map[string]struct{})
 
 	for entry := range strings.SplitSeq(rawValue, ",") {
@@ -450,6 +458,12 @@ 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 {
@@ -476,7 +490,7 @@ func (sm *SmartManager) CollectSmart(deviceInfo *DeviceInfo) error {
 	output, err := cmd.CombinedOutput()
 
 	// Check if device is in standby (exit status 2)
-	if exitErr, ok := err.(*exec.ExitError); ok && exitErr.ExitCode() == 2 {
+	if exitErr, ok := errors.AsType[*exec.ExitError](err); ok && exitErr.ExitCode() == 2 {
 		if hasExistingData {
 			// Device is in standby and we have cached data, keep using cache
 			return nil
@@ -857,15 +871,18 @@ func (sm *SmartManager) parseSmartForSata(output []byte) (bool, int) {
 	smartData.FirmwareVersion = data.FirmwareVersion
 	smartData.Capacity = data.UserCapacity.Bytes
 	smartData.Temperature = data.Temperature.Current
-	if smartData.Temperature == 0 {
-		if temp, ok := temperatureFromAtaDeviceStatistics(data.AtaDeviceStatistics); ok {
-			smartData.Temperature = temp
-		}
-	}
 	smartData.SmartStatus = getSmartStatus(smartData.Temperature, data.SmartStatus.Passed)
 	smartData.DiskName = data.Device.Name
 	smartData.DiskType = data.Device.Type
 
+	// get values from ata_device_statistics if necessary
+	var ataDeviceStats smart.AtaDeviceStatistics
+	if smartData.Temperature == 0 {
+		if temp := findAtaDeviceStatisticsValue(&data, &ataDeviceStats, 5, "Current Temperature", 0, 255); temp != nil {
+			smartData.Temperature = uint8(*temp)
+		}
+	}
+
 	// update SmartAttributes
 	smartData.Attributes = make([]*smart.SmartAttribute, 0, len(data.AtaSmartAttributes.Table))
 	for _, attr := range data.AtaSmartAttributes.Table {
@@ -900,23 +917,20 @@ func getSmartStatus(temperature uint8, passed bool) string {
 	}
 }
 
-func temperatureFromAtaDeviceStatistics(stats smart.AtaDeviceStatistics) (uint8, bool) {
-	entry := findAtaDeviceStatisticsEntry(stats, 5, "Current Temperature")
-	if entry == nil || entry.Value == nil {
-		return 0, false
-	}
-	if *entry.Value > 255 {
-		return 0, false
-	}
-	return uint8(*entry.Value), true
-}
-
 // findAtaDeviceStatisticsEntry centralizes ATA devstat lookups so additional
 // metrics can be pulled from the same structure in the future.
-func findAtaDeviceStatisticsEntry(stats smart.AtaDeviceStatistics, pageNumber uint8, entryName string) *smart.AtaDeviceStatisticsEntry {
-	for pageIdx := range stats.Pages {
-		page := &stats.Pages[pageIdx]
-		if page.Number != pageNumber {
+func findAtaDeviceStatisticsValue(data *smart.SmartInfoForSata, ataDeviceStats *smart.AtaDeviceStatistics, entryNumber uint8, entryName string, minValue, maxValue int64) *int64 {
+	if len(ataDeviceStats.Pages) == 0 {
+		if len(data.AtaDeviceStatistics) == 0 {
+			return nil
+		}
+		if err := json.Unmarshal(data.AtaDeviceStatistics, ataDeviceStats); err != nil {
+			return nil
+		}
+	}
+	for pageIdx := range ataDeviceStats.Pages {
+		page := &ataDeviceStats.Pages[pageIdx]
+		if page.Number != entryNumber {
 			continue
 		}
 		for entryIdx := range page.Table {
@@ -924,7 +938,10 @@ func findAtaDeviceStatisticsEntry(stats smart.AtaDeviceStatistics, pageNumber ui
 			if !strings.EqualFold(entry.Name, entryName) {
 				continue
 			}
-			return entry
+			if entry.Value == nil || *entry.Value < minValue || *entry.Value > maxValue {
+				return nil
+			}
+			return entry.Value
 		}
 	}
 	return nil
@@ -1146,9 +1163,11 @@ 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
 	}

agent/smart_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package agent
 
@@ -122,6 +121,78 @@ func TestParseSmartForSataDeviceStatisticsTemperature(t *testing.T) {
 	assert.Equal(t, uint8(22), deviceData.Temperature)
 }
 
+func TestParseSmartForSataAtaDeviceStatistics(t *testing.T) {
+	// tests that ata_device_statistics values are parsed correctly
+	jsonPayload := []byte(`{
+		"smartctl": {"exit_status": 0},
+		"device": {"name": "/dev/sdb", "type": "sat"},
+		"model_name": "SanDisk SSD U110 16GB",
+		"serial_number": "lksjfh23lhj",
+		"firmware_version": "U21B001",
+		"user_capacity": {"bytes": 16013942784},
+		"smart_status": {"passed": true},
+		"ata_smart_attributes": {"table": []},
+		"ata_device_statistics": {
+			"pages": [
+				{
+					"number": 5,
+					"name": "Temperature Statistics",
+					"table": [
+						{"name": "Current Temperature", "value": 43, "flags": {"valid": true}},
+						{"name": "Specified Minimum Operating Temperature", "value": -20, "flags": {"valid": true}}
+					]
+				}
+			]
+		}
+	}`)
+
+	sm := &SmartManager{SmartDataMap: make(map[string]*smart.SmartData)}
+	hasData, exitStatus := sm.parseSmartForSata(jsonPayload)
+	require.True(t, hasData)
+	assert.Equal(t, 0, exitStatus)
+
+	deviceData, ok := sm.SmartDataMap["lksjfh23lhj"]
+	require.True(t, ok, "expected smart data entry for serial lksjfh23lhj")
+	assert.Equal(t, uint8(43), deviceData.Temperature)
+}
+
+func TestParseSmartForSataNegativeDeviceStatistics(t *testing.T) {
+	// Tests that negative values in ata_device_statistics (e.g. min operating temp)
+	// do not cause the entire SAT parser to fail.
+	jsonPayload := []byte(`{
+		"smartctl": {"exit_status": 0},
+		"device": {"name": "/dev/sdb", "type": "sat"},
+		"model_name": "SanDisk SSD U110 16GB",
+		"serial_number": "NEGATIVE123",
+		"firmware_version": "U21B001",
+		"user_capacity": {"bytes": 16013942784},
+		"smart_status": {"passed": true},
+		"temperature": {"current": 38},
+		"ata_smart_attributes": {"table": []},
+		"ata_device_statistics": {
+			"pages": [
+				{
+					"number": 5,
+					"name": "Temperature Statistics",
+					"table": [
+						{"name": "Current Temperature", "value": 38, "flags": {"valid": true}},
+						{"name": "Specified Minimum Operating Temperature", "value": -20, "flags": {"valid": true}}
+					]
+				}
+			]
+		}
+	}`)
+
+	sm := &SmartManager{SmartDataMap: make(map[string]*smart.SmartData)}
+	hasData, exitStatus := sm.parseSmartForSata(jsonPayload)
+	require.True(t, hasData)
+	assert.Equal(t, 0, exitStatus)
+
+	deviceData, ok := sm.SmartDataMap["NEGATIVE123"]
+	require.True(t, ok, "expected smart data entry for serial NEGATIVE123")
+	assert.Equal(t, uint8(38), deviceData.Temperature)
+}
+
 func TestParseSmartForSataParentheticalRawValue(t *testing.T) {
 	jsonPayload := []byte(`{
 		"smartctl": {"exit_status": 0},
@@ -728,6 +799,182 @@ func TestIsVirtualDeviceScsi(t *testing.T) {
 	}
 }
 
+func TestFindAtaDeviceStatisticsValue(t *testing.T) {
+	val42 := int64(42)
+	val100 := int64(100)
+	valMinus20 := int64(-20)
+
+	tests := []struct {
+		name           string
+		data           smart.SmartInfoForSata
+		ataDeviceStats smart.AtaDeviceStatistics
+		entryNumber    uint8
+		entryName      string
+		minValue       int64
+		maxValue       int64
+		expectedValue  *int64
+	}{
+		{
+			name: "value in ataDeviceStats",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Current Temperature", Value: &val42},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: &val42,
+		},
+		{
+			name: "value unmarshaled from data",
+			data: smart.SmartInfoForSata{
+				AtaDeviceStatistics: []byte(`{"pages":[{"number":5,"table":[{"name":"Current Temperature","value":100}]}]}`),
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      255,
+			expectedValue: &val100,
+		},
+		{
+			name: "value out of range (too high)",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Current Temperature", Value: &val100},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      50,
+			expectedValue: nil,
+		},
+		{
+			name: "value out of range (too low)",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Min Temp", Value: &valMinus20},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Min Temp",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: nil,
+		},
+		{
+			name:          "no statistics available",
+			data:          smart.SmartInfoForSata{},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      255,
+			expectedValue: nil,
+		},
+		{
+			name: "wrong page number",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 1,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Current Temperature", Value: &val42},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: nil,
+		},
+		{
+			name: "wrong entry name",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Other Stat", Value: &val42},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: nil,
+		},
+		{
+			name: "case insensitive name match",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "CURRENT TEMPERATURE", Value: &val42},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: &val42,
+		},
+		{
+			name: "entry value is nil",
+			ataDeviceStats: smart.AtaDeviceStatistics{
+				Pages: []smart.AtaDeviceStatisticsPage{
+					{
+						Number: 5,
+						Table: []smart.AtaDeviceStatisticsEntry{
+							{Name: "Current Temperature", Value: nil},
+						},
+					},
+				},
+			},
+			entryNumber:   5,
+			entryName:     "Current Temperature",
+			minValue:      0,
+			maxValue:      100,
+			expectedValue: nil,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := findAtaDeviceStatisticsValue(&tt.data, &tt.ataDeviceStats, tt.entryNumber, tt.entryName, tt.minValue, tt.maxValue)
+			if tt.expectedValue == nil {
+				assert.Nil(t, result)
+			} else {
+				require.NotNil(t, result)
+				assert.Equal(t, *tt.expectedValue, *result)
+			}
+		})
+	}
+}
+
 func TestRefreshExcludedDevices(t *testing.T) {
 	tests := []struct {
 		name         string

agent/system.go

@@ -7,12 +7,13 @@ import (
 	"log/slog"
 	"os"
 	"runtime"
-	"strconv"
 	"strings"
 	"time"
 
 	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/agent/battery"
+	"github.com/henrygd/beszel/agent/utils"
+	"github.com/henrygd/beszel/agent/zfs"
 	"github.com/henrygd/beszel/internal/entities/container"
 	"github.com/henrygd/beszel/internal/entities/system"
 
@@ -107,7 +108,7 @@ func (a *Agent) refreshSystemDetails() {
 	}
 
 	// zfs
-	if _, err := getARCSize(); err != nil {
+	if _, err := zfs.ARCSize(); err != nil {
 		slog.Debug("Not monitoring ZFS ARC", "err", err)
 	} else {
 		a.zfs = true
@@ -127,13 +128,13 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	// cpu metrics
 	cpuMetrics, err := getCpuMetrics(cacheTimeMs)
 	if err == nil {
-		systemStats.Cpu = twoDecimals(cpuMetrics.Total)
+		systemStats.Cpu = utils.TwoDecimals(cpuMetrics.Total)
 		systemStats.CpuBreakdown = []float64{
-			twoDecimals(cpuMetrics.User),
-			twoDecimals(cpuMetrics.System),
-			twoDecimals(cpuMetrics.Iowait),
-			twoDecimals(cpuMetrics.Steal),
-			twoDecimals(cpuMetrics.Idle),
+			utils.TwoDecimals(cpuMetrics.User),
+			utils.TwoDecimals(cpuMetrics.System),
+			utils.TwoDecimals(cpuMetrics.Iowait),
+			utils.TwoDecimals(cpuMetrics.Steal),
+			utils.TwoDecimals(cpuMetrics.Idle),
 		}
 	} else {
 		slog.Error("Error getting cpu metrics", "err", err)
@@ -157,8 +158,8 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	// memory
 	if v, err := mem.VirtualMemory(); err == nil {
 		// swap
-		systemStats.Swap = bytesToGigabytes(v.SwapTotal)
-		systemStats.SwapUsed = bytesToGigabytes(v.SwapTotal - v.SwapFree - v.SwapCached)
+		systemStats.Swap = utils.BytesToGigabytes(v.SwapTotal)
+		systemStats.SwapUsed = utils.BytesToGigabytes(v.SwapTotal - v.SwapFree - v.SwapCached)
 		// cache + buffers value for default mem calculation
 		// note: gopsutil automatically adds SReclaimable to v.Cached
 		cacheBuff := v.Cached + v.Buffers - v.Shared
@@ -178,16 +179,16 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 		// }
 		// subtract ZFS ARC size from used memory and add as its own category
 		if a.zfs {
-			if arcSize, _ := getARCSize(); arcSize > 0 && arcSize < v.Used {
+			if arcSize, _ := zfs.ARCSize(); arcSize > 0 && arcSize < v.Used {
 				v.Used = v.Used - arcSize
 				v.UsedPercent = float64(v.Used) / float64(v.Total) * 100.0
-				systemStats.MemZfsArc = bytesToGigabytes(arcSize)
+				systemStats.MemZfsArc = utils.BytesToGigabytes(arcSize)
 			}
 		}
-		systemStats.Mem = bytesToGigabytes(v.Total)
-		systemStats.MemBuffCache = bytesToGigabytes(cacheBuff)
-		systemStats.MemUsed = bytesToGigabytes(v.Used)
-		systemStats.MemPct = twoDecimals(v.UsedPercent)
+		systemStats.Mem = utils.BytesToGigabytes(v.Total)
+		systemStats.MemBuffCache = utils.BytesToGigabytes(cacheBuff)
+		systemStats.MemUsed = utils.BytesToGigabytes(v.Used)
+		systemStats.MemPct = utils.TwoDecimals(v.UsedPercent)
 	}
 
 	// disk usage
@@ -250,32 +251,6 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	return systemStats
 }
 
-// Returns the size of the ZFS ARC memory cache in bytes
-func getARCSize() (uint64, error) {
-	file, err := os.Open("/proc/spl/kstat/zfs/arcstats")
-	if err != nil {
-		return 0, err
-	}
-	defer file.Close()
-
-	// Scan the lines
-	scanner := bufio.NewScanner(file)
-	for scanner.Scan() {
-		line := scanner.Text()
-		if strings.HasPrefix(line, "size") {
-			// Example line: size 4 15032385536
-			fields := strings.Fields(line)
-			if len(fields) < 3 {
-				return 0, err
-			}
-			// Return the size as uint64
-			return strconv.ParseUint(fields[2], 10, 64)
-		}
-	}
-
-	return 0, fmt.Errorf("failed to parse size field")
-}
-
 // getOsPrettyName attempts to get the pretty OS name from /etc/os-release on Linux systems
 func getOsPrettyName() (string, error) {
 	file, err := os.Open("/etc/os-release")

agent/systemd.go

@@ -15,6 +15,7 @@ import (
 	"time"
 
 	"github.com/coreos/go-systemd/v22/dbus"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/systemd"
 )
 
@@ -49,7 +50,7 @@ func isSystemdAvailable() bool {
 
 // newSystemdManager creates a new systemdManager.
 func newSystemdManager() (*systemdManager, error) {
-	if skipSystemd, _ := GetEnv("SKIP_SYSTEMD"); skipSystemd == "true" {
+	if skipSystemd, _ := utils.GetEnv("SKIP_SYSTEMD"); skipSystemd == "true" {
 		return nil, nil
 	}
 
@@ -294,7 +295,7 @@ func unescapeServiceName(name string) string {
 // otherwise defaults to "*service".
 func getServicePatterns() []string {
 	patterns := []string{}
-	if envPatterns, _ := GetEnv("SERVICE_PATTERNS"); envPatterns != "" {
+	if envPatterns, _ := utils.GetEnv("SERVICE_PATTERNS"); envPatterns != "" {
 		for pattern := range strings.SplitSeq(envPatterns, ",") {
 			pattern = strings.TrimSpace(pattern)
 			if pattern == "" {

agent/utils.go

@@ -1,15 +0,0 @@
-package agent
-
-import "math"
-
-func bytesToMegabytes(b float64) float64 {
-	return twoDecimals(b / 1048576)
-}
-
-func bytesToGigabytes(b uint64) float64 {
-	return twoDecimals(float64(b) / 1073741824)
-}
-
-func twoDecimals(value float64) float64 {
-	return math.Round(value*100) / 100
-}

agent/utils/utils.go

@@ -0,0 +1,88 @@
+package utils
+
+import (
+	"io"
+	"math"
+	"os"
+	"strconv"
+	"strings"
+)
+
+// GetEnv retrieves an environment variable with a "BESZEL_AGENT_" prefix, or falls back to the unprefixed key.
+func GetEnv(key string) (value string, exists bool) {
+	if value, exists = os.LookupEnv("BESZEL_AGENT_" + key); exists {
+		return value, exists
+	}
+	return os.LookupEnv(key)
+}
+
+// BytesToMegabytes converts bytes to megabytes and rounds to two decimal places.
+func BytesToMegabytes(b float64) float64 {
+	return TwoDecimals(b / 1048576)
+}
+
+// BytesToGigabytes converts bytes to gigabytes and rounds to two decimal places.
+func BytesToGigabytes(b uint64) float64 {
+	return TwoDecimals(float64(b) / 1073741824)
+}
+
+// TwoDecimals rounds a float64 value to two decimal places.
+func TwoDecimals(value float64) float64 {
+	return math.Round(value*100) / 100
+}
+
+// func RoundFloat(val float64, precision uint) float64 {
+//     ratio := math.Pow(10, float64(precision))
+//     return math.Round(val*ratio) / ratio
+// }
+
+// ReadStringFile returns trimmed file contents or empty string on error.
+func ReadStringFile(path string) string {
+	content, _ := ReadStringFileOK(path)
+	return content
+}
+
+// ReadStringFileOK returns trimmed file contents and read success.
+func ReadStringFileOK(path string) (string, bool) {
+	b, err := os.ReadFile(path)
+	if err != nil {
+		return "", false
+	}
+	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)
+	return err == nil
+}
+
+// ReadUintFile parses a decimal uint64 value from a file.
+func ReadUintFile(path string) (uint64, bool) {
+	raw, ok := ReadStringFileOK(path)
+	if !ok {
+		return 0, false
+	}
+	parsed, err := strconv.ParseUint(raw, 10, 64)
+	if err != nil {
+		return 0, false
+	}
+	return parsed, true
+}

agent/utils/utils_test.go

@@ -0,0 +1,165 @@
+package utils
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestTwoDecimals(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    float64
+		expected float64
+	}{
+		{"round down", 1.234, 1.23},
+		{"round half up", 1.235, 1.24}, // math.Round rounds half up
+		{"no rounding needed", 1.23, 1.23},
+		{"negative number", -1.235, -1.24}, // math.Round rounds half up (more negative)
+		{"zero", 0.0, 0.0},
+		{"large number", 123.456, 123.46}, // rounds 5 up
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := TwoDecimals(tt.input)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestBytesToMegabytes(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    float64
+		expected float64
+	}{
+		{"1 MB", 1048576, 1.0},
+		{"512 KB", 524288, 0.5},
+		{"zero", 0, 0},
+		{"large value", 1073741824, 1024}, // 1 GB = 1024 MB
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := BytesToMegabytes(tt.input)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestBytesToGigabytes(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    uint64
+		expected float64
+	}{
+		{"1 GB", 1073741824, 1.0},
+		{"512 MB", 536870912, 0.5},
+		{"0 GB", 0, 0},
+		{"2 GB", 2147483648, 2.0},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := BytesToGigabytes(tt.input)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}
+
+func TestFileFunctions(t *testing.T) {
+	tmpDir := t.TempDir()
+	testFilePath := filepath.Join(tmpDir, "test.txt")
+	testContent := "hello world"
+
+	// Test FileExists (false)
+	assert.False(t, FileExists(testFilePath))
+
+	// Test ReadStringFileOK (false)
+	content, ok := ReadStringFileOK(testFilePath)
+	assert.False(t, ok)
+	assert.Empty(t, content)
+
+	// Test ReadStringFile (empty)
+	assert.Empty(t, ReadStringFile(testFilePath))
+
+	// Write file
+	err := os.WriteFile(testFilePath, []byte(testContent+"\n "), 0644)
+	assert.NoError(t, err)
+
+	// Test FileExists (true)
+	assert.True(t, FileExists(testFilePath))
+
+	// Test ReadStringFileOK (true)
+	content, ok = ReadStringFileOK(testFilePath)
+	assert.True(t, ok)
+	assert.Equal(t, testContent, content)
+
+	// Test ReadStringFile (content)
+	assert.Equal(t, testContent, ReadStringFile(testFilePath))
+}
+
+func TestReadUintFile(t *testing.T) {
+	tmpDir := t.TempDir()
+
+	t.Run("valid uint", func(t *testing.T) {
+		path := filepath.Join(tmpDir, "uint.txt")
+		os.WriteFile(path, []byte(" 12345\n"), 0644)
+		val, ok := ReadUintFile(path)
+		assert.True(t, ok)
+		assert.Equal(t, uint64(12345), val)
+	})
+
+	t.Run("invalid uint", func(t *testing.T) {
+		path := filepath.Join(tmpDir, "invalid.txt")
+		os.WriteFile(path, []byte("abc"), 0644)
+		val, ok := ReadUintFile(path)
+		assert.False(t, ok)
+		assert.Equal(t, uint64(0), val)
+	})
+
+	t.Run("missing file", func(t *testing.T) {
+		path := filepath.Join(tmpDir, "missing.txt")
+		val, ok := ReadUintFile(path)
+		assert.False(t, ok)
+		assert.Equal(t, uint64(0), val)
+	})
+}
+
+func TestGetEnv(t *testing.T) {
+	key := "TEST_VAR"
+	prefixedKey := "BESZEL_AGENT_" + key
+
+	t.Run("prefixed variable exists", func(t *testing.T) {
+		os.Setenv(prefixedKey, "prefixed_val")
+		os.Setenv(key, "unprefixed_val")
+		defer os.Unsetenv(prefixedKey)
+		defer os.Unsetenv(key)
+
+		val, exists := GetEnv(key)
+		assert.True(t, exists)
+		assert.Equal(t, "prefixed_val", val)
+	})
+
+	t.Run("only unprefixed variable exists", func(t *testing.T) {
+		os.Unsetenv(prefixedKey)
+		os.Setenv(key, "unprefixed_val")
+		defer os.Unsetenv(key)
+
+		val, exists := GetEnv(key)
+		assert.True(t, exists)
+		assert.Equal(t, "unprefixed_val", val)
+	})
+
+	t.Run("neither variable exists", func(t *testing.T) {
+		os.Unsetenv(prefixedKey)
+		os.Unsetenv(key)
+
+		val, exists := GetEnv(key)
+		assert.False(t, exists)
+		assert.Empty(t, val)
+	})
+}

agent/zfs/zfs_freebsd.go

@@ -0,0 +1,11 @@
+//go:build freebsd
+
+package zfs
+
+import (
+	"golang.org/x/sys/unix"
+)
+
+func ARCSize() (uint64, error) {
+	return unix.SysctlUint64("kstat.zfs.misc.arcstats.size")
+}

agent/zfs/zfs_linux.go

@@ -0,0 +1,34 @@
+//go:build linux
+
+// Package zfs provides functions to read ZFS statistics.
+package zfs
+
+import (
+	"bufio"
+	"fmt"
+	"os"
+	"strconv"
+	"strings"
+)
+
+func ARCSize() (uint64, error) {
+	file, err := os.Open("/proc/spl/kstat/zfs/arcstats")
+	if err != nil {
+		return 0, err
+	}
+	defer file.Close()
+
+	scanner := bufio.NewScanner(file)
+	for scanner.Scan() {
+		line := scanner.Text()
+		if strings.HasPrefix(line, "size") {
+			fields := strings.Fields(line)
+			if len(fields) < 3 {
+				return 0, fmt.Errorf("unexpected arcstats size format: %s", line)
+			}
+			return strconv.ParseUint(fields[2], 10, 64)
+		}
+	}
+
+	return 0, fmt.Errorf("size field not found in arcstats")
+}

agent/zfs/zfs_unsupported.go

@@ -0,0 +1,9 @@
+//go:build !linux && !freebsd
+
+package zfs
+
+import "errors"
+
+func ARCSize() (uint64, error) {
+	return 0, errors.ErrUnsupported
+}

beszel.go

@@ -6,7 +6,7 @@ import "github.com/blang/semver"
 
 const (
 	// Version is the current version of the application.
-	Version = "0.18.3"
+	Version = "0.18.4"
 	// AppName is the name of the application.
 	AppName = "beszel"
 )

go.mod

@@ -1,6 +1,6 @@
 module github.com/henrygd/beszel
 
-go 1.25.7
+go 1.26.1
 
 require (
 	github.com/blang/semver v3.5.1+incompatible
@@ -11,17 +11,17 @@ require (
 	github.com/gliderlabs/ssh v0.3.8
 	github.com/google/uuid v1.6.0
 	github.com/lxzan/gws v1.8.9
-	github.com/nicholas-fedor/shoutrrr v0.13.1
-	github.com/pocketbase/dbx v1.11.0
-	github.com/pocketbase/pocketbase v0.36.2
+	github.com/nicholas-fedor/shoutrrr v0.13.2
+	github.com/pocketbase/dbx v1.12.0
+	github.com/pocketbase/pocketbase v0.36.4
 	github.com/shirou/gopsutil/v4 v4.26.1
 	github.com/spf13/cast v1.10.0
 	github.com/spf13/cobra v1.10.2
 	github.com/spf13/pflag v1.0.10
 	github.com/stretchr/testify v1.11.1
-	golang.org/x/crypto v0.47.0
-	golang.org/x/exp v0.0.0-20260112195511-716be5621a96
-	golang.org/x/sys v0.40.0
+	golang.org/x/crypto v0.48.0
+	golang.org/x/exp v0.0.0-20260218203240-3dfff04db8fa
+	golang.org/x/sys v0.41.0
 	gopkg.in/yaml.v3 v3.0.1
 )
 
@@ -42,8 +42,8 @@ require (
 	github.com/godbus/dbus/v5 v5.2.2 // indirect
 	github.com/golang-jwt/jwt/v5 v5.3.1 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // indirect
-	github.com/klauspost/compress v1.18.3 // indirect
-	github.com/lufia/plan9stats v0.0.0-20251013123823-9fd1530e3ec3 // indirect
+	github.com/klauspost/compress v1.18.4 // indirect
+	github.com/lufia/plan9stats v0.0.0-20260216142805-b3301c5f2a88 // indirect
 	github.com/mattn/go-colorable v0.1.14 // indirect
 	github.com/mattn/go-isatty v0.0.20 // indirect
 	github.com/ncruces/go-strftime v1.0.0 // indirect
@@ -54,15 +54,15 @@ require (
 	github.com/tklauser/numcpus v0.11.0 // indirect
 	github.com/x448/float16 v0.8.4 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // indirect
-	golang.org/x/image v0.35.0 // indirect
-	golang.org/x/net v0.49.0 // indirect
-	golang.org/x/oauth2 v0.34.0 // indirect
+	golang.org/x/image v0.36.0 // indirect
+	golang.org/x/net v0.50.0 // indirect
+	golang.org/x/oauth2 v0.35.0 // indirect
 	golang.org/x/sync v0.19.0 // indirect
-	golang.org/x/term v0.39.0 // indirect
-	golang.org/x/text v0.33.0 // indirect
+	golang.org/x/term v0.40.0 // indirect
+	golang.org/x/text v0.34.0 // indirect
 	howett.net/plist v1.0.1 // indirect
 	modernc.org/libc v1.67.6 // indirect
 	modernc.org/mathutil v1.7.1 // indirect
 	modernc.org/memory v1.11.0 // indirect
-	modernc.org/sqlite v1.44.3 // indirect
+	modernc.org/sqlite v1.45.0 // indirect
 )

go.sum

@@ -69,14 +69,14 @@ github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLf
 github.com/jarcoal/httpmock v1.4.1 h1:0Ju+VCFuARfFlhVXFc2HxlcQkfB+Xq12/EotHko+x2A=
 github.com/jarcoal/httpmock v1.4.1/go.mod h1:ftW1xULwo+j0R0JJkJIIi7UKigZUXCLLanykgjwBXL0=
 github.com/jessevdk/go-flags v1.4.0/go.mod h1:4FA24M0QyGHXBuZZK/XkWh8h0e1EYbRYJSGM75WSRxI=
-github.com/klauspost/compress v1.18.3 h1:9PJRvfbmTabkOX8moIpXPbMMbYN60bWImDDU7L+/6zw=
-github.com/klauspost/compress v1.18.3/go.mod h1:R0h/fSBs8DE4ENlcrlib3PsXS61voFxhIs2DeRhCvJ4=
+github.com/klauspost/compress v1.18.4 h1:RPhnKRAQ4Fh8zU2FY/6ZFDwTVTxgJ/EMydqSTzE9a2c=
+github.com/klauspost/compress v1.18.4/go.mod h1:R0h/fSBs8DE4ENlcrlib3PsXS61voFxhIs2DeRhCvJ4=
 github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
 github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
 github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
 github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
-github.com/lufia/plan9stats v0.0.0-20251013123823-9fd1530e3ec3 h1:PwQumkgq4/acIiZhtifTV5OUqqiP82UAl0h87xj/l9k=
-github.com/lufia/plan9stats v0.0.0-20251013123823-9fd1530e3ec3/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
+github.com/lufia/plan9stats v0.0.0-20260216142805-b3301c5f2a88 h1:PTw+yKnXcOFCR6+8hHTyWBeQ/P4Nb7dd4/0ohEcWQuM=
+github.com/lufia/plan9stats v0.0.0-20260216142805-b3301c5f2a88/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
 github.com/lxzan/gws v1.8.9 h1:VU3SGUeWlQrEwfUSfokcZep8mdg/BrUF+y73YYshdBM=
 github.com/lxzan/gws v1.8.9/go.mod h1:d9yHaR1eDTBHagQC6KY7ycUOaz5KWeqQtP3xu7aMK8Y=
 github.com/mattn/go-colorable v0.1.14 h1:9A9LHSqF/7dyVVX6g0U9cwm9pG3kP9gSzcuIPHPsaIE=
@@ -85,19 +85,19 @@ github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWE
 github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
 github.com/ncruces/go-strftime v1.0.0 h1:HMFp8mLCTPp341M/ZnA4qaf7ZlsbTc+miZjCLOFAw7w=
 github.com/ncruces/go-strftime v1.0.0/go.mod h1:Fwc5htZGVVkseilnfgOVb9mKy6w1naJmn9CehxcKcls=
-github.com/nicholas-fedor/shoutrrr v0.13.1 h1:llEoHNbnMM4GfQ9+2Ns3n6ssvNfi3NPWluM0AQiicoY=
-github.com/nicholas-fedor/shoutrrr v0.13.1/go.mod h1:kU4cFJpEAtTzl3iV0l+XUXmM90OlC5T01b7roM4/pYM=
-github.com/onsi/ginkgo/v2 v2.27.3 h1:ICsZJ8JoYafeXFFlFAG75a7CxMsJHwgKwtO+82SE9L8=
-github.com/onsi/ginkgo/v2 v2.27.3/go.mod h1:ArE1D/XhNXBXCBkKOLkbsb2c81dQHCRcF5zwn/ykDRo=
-github.com/onsi/gomega v1.38.3 h1:eTX+W6dobAYfFeGC2PV6RwXRu/MyT+cQguijutvkpSM=
-github.com/onsi/gomega v1.38.3/go.mod h1:ZCU1pkQcXDO5Sl9/VVEGlDyp+zm0m1cmeG5TOzLgdh4=
+github.com/nicholas-fedor/shoutrrr v0.13.2 h1:hfsYBIqSFYGg92pZP5CXk/g7/OJIkLYmiUnRl+AD1IA=
+github.com/nicholas-fedor/shoutrrr v0.13.2/go.mod h1:ZqzV3gY/Wj6AvWs1etlO7+yKbh4iptSbeL8avBpMQbA=
+github.com/onsi/ginkgo/v2 v2.28.1 h1:S4hj+HbZp40fNKuLUQOYLDgZLwNUVn19N3Atb98NCyI=
+github.com/onsi/ginkgo/v2 v2.28.1/go.mod h1:CLtbVInNckU3/+gC8LzkGUb9oF+e8W8TdUsxPwvdOgE=
+github.com/onsi/gomega v1.39.1 h1:1IJLAad4zjPn2PsnhH70V4DKRFlrCzGBNrNaru+Vf28=
+github.com/onsi/gomega v1.39.1/go.mod h1:hL6yVALoTOxeWudERyfppUcZXjMwIMLnuSfruD2lcfg=
 github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 h1:Jamvg5psRIccs7FGNTlIRMkT8wgtp5eCXdBlqhYGL6U=
 github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
-github.com/pocketbase/dbx v1.11.0 h1:LpZezioMfT3K4tLrqA55wWFw1EtH1pM4tzSVa7kgszU=
-github.com/pocketbase/dbx v1.11.0/go.mod h1:xXRCIAKTHMgUCyCKZm55pUOdvFziJjQfXaWKhu2vhMs=
-github.com/pocketbase/pocketbase v0.36.2 h1:mzrxnvXKc3yxKlvZdbwoYXkH8kfIETteD0hWdgj0VI4=
-github.com/pocketbase/pocketbase v0.36.2/go.mod h1:71vSF8whUDzC8mcLFE10+Qatf9JQdeOGIRWawOuLLKM=
+github.com/pocketbase/dbx v1.12.0 h1:/oLErM+A0b4xI0PWTGPqSDVjzix48PqI/bng2l0PzoA=
+github.com/pocketbase/dbx v1.12.0/go.mod h1:xXRCIAKTHMgUCyCKZm55pUOdvFziJjQfXaWKhu2vhMs=
+github.com/pocketbase/pocketbase v0.36.4 h1:zTjRZbp2WfTOJJfb+pFRWa200UaQwxZYt8RzkFMlAZ4=
+github.com/pocketbase/pocketbase v0.36.4/go.mod h1:9CiezhRudd9FZGa5xZa53QZBTNxc5vvw/FGG+diAECI=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 h1:o4JXh1EVt9k/+g42oCprj/FisM4qX9L3sZB3upGN2ZU=
 github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
@@ -129,20 +129,20 @@ github.com/yusufpapurcu/wmi v1.2.4/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQ
 go.yaml.in/yaml/v3 v3.0.4 h1:tfq32ie2Jv2UxXFdLJdh3jXuOzWiL1fo0bu/FbuKpbc=
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-golang.org/x/crypto v0.47.0 h1:V6e3FRj+n4dbpw86FJ8Fv7XVOql7TEwpHapKoMJ/GO8=
-golang.org/x/crypto v0.47.0/go.mod h1:ff3Y9VzzKbwSSEzWqJsJVBnWmRwRSHt/6Op5n9bQc4A=
-golang.org/x/exp v0.0.0-20260112195511-716be5621a96 h1:Z/6YuSHTLOHfNFdb8zVZomZr7cqNgTJvA8+Qz75D8gU=
-golang.org/x/exp v0.0.0-20260112195511-716be5621a96/go.mod h1:nzimsREAkjBCIEFtHiYkrJyT+2uy9YZJB7H1k68CXZU=
+golang.org/x/crypto v0.48.0 h1:/VRzVqiRSggnhY7gNRxPauEQ5Drw9haKdM0jqfcCFts=
+golang.org/x/crypto v0.48.0/go.mod h1:r0kV5h3qnFPlQnBSrULhlsRfryS2pmewsg+XfMgkVos=
+golang.org/x/exp v0.0.0-20260218203240-3dfff04db8fa h1:Zt3DZoOFFYkKhDT3v7Lm9FDMEV06GpzjG2jrqW+QTE0=
+golang.org/x/exp v0.0.0-20260218203240-3dfff04db8fa/go.mod h1:K79w1Vqn7PoiZn+TkNpx3BUWUQksGO3JcVX6qIjytmA=
 golang.org/x/image v0.0.0-20191009234506-e7c1f5e7dbb8/go.mod h1:FeLwcggjj3mMvU+oOTbSwawSJRM1uh48EjtB4UJZlP0=
-golang.org/x/image v0.35.0 h1:LKjiHdgMtO8z7Fh18nGY6KDcoEtVfsgLDPeLyguqb7I=
-golang.org/x/image v0.35.0/go.mod h1:MwPLTVgvxSASsxdLzKrl8BRFuyqMyGhLwmC+TO1Sybk=
-golang.org/x/mod v0.32.0 h1:9F4d3PHLljb6x//jOyokMv3eX+YDeepZSEo3mFJy93c=
-golang.org/x/mod v0.32.0/go.mod h1:SgipZ/3h2Ci89DlEtEXWUk/HteuRin+HHhN+WbNhguU=
+golang.org/x/image v0.36.0 h1:Iknbfm1afbgtwPTmHnS2gTM/6PPZfH+z2EFuOkSbqwc=
+golang.org/x/image v0.36.0/go.mod h1:YsWD2TyyGKiIX1kZlu9QfKIsQ4nAAK9bdgdrIsE7xy4=
+golang.org/x/mod v0.33.0 h1:tHFzIWbBifEmbwtGz65eaWyGiGZatSrT9prnU8DbVL8=
+golang.org/x/mod v0.33.0/go.mod h1:swjeQEj+6r7fODbD2cqrnje9PnziFuw4bmLbBZFrQ5w=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
-golang.org/x/net v0.49.0 h1:eeHFmOGUTtaaPSGNmjBKpbng9MulQsJURQUAfUwY++o=
-golang.org/x/net v0.49.0/go.mod h1:/ysNB2EvaqvesRkuLAyjI1ycPZlQHM3q01F02UY/MV8=
-golang.org/x/oauth2 v0.34.0 h1:hqK/t4AKgbqWkdkcAeI8XLmbK+4m4G5YeQRrmiotGlw=
-golang.org/x/oauth2 v0.34.0/go.mod h1:lzm5WQJQwKZ3nwavOZ3IS5Aulzxi68dUSgRHujetwEA=
+golang.org/x/net v0.50.0 h1:ucWh9eiCGyDR3vtzso0WMQinm2Dnt8cFMuQa9K33J60=
+golang.org/x/net v0.50.0/go.mod h1:UgoSli3F/pBgdJBHCTc+tp3gmrU4XswgGRgtnwWTfyM=
+golang.org/x/oauth2 v0.35.0 h1:Mv2mzuHuZuY2+bkyWXIHMfhNdJAdwW3FuWeCPYN5GVQ=
+golang.org/x/oauth2 v0.35.0/go.mod h1:lzm5WQJQwKZ3nwavOZ3IS5Aulzxi68dUSgRHujetwEA=
 golang.org/x/sync v0.19.0 h1:vV+1eWNmZ5geRlYjzm2adRgW2/mcpevXNg50YZtPCE4=
 golang.org/x/sync v0.19.0/go.mod h1:9KTHXmSnoGruLpwFjVSX0lNNA75CykiMECbovNTZqGI=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
@@ -150,20 +150,20 @@ golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7w
 golang.org/x/sys v0.0.0-20201204225414-ed752295db88/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 golang.org/x/sys v0.1.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
 golang.org/x/sys v0.6.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
-golang.org/x/sys v0.40.0 h1:DBZZqJ2Rkml6QMQsZywtnjnnGvHza6BTfYFWY9kjEWQ=
-golang.org/x/sys v0.40.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
-golang.org/x/term v0.39.0 h1:RclSuaJf32jOqZz74CkPA9qFuVTX7vhLlpfj/IGWlqY=
-golang.org/x/term v0.39.0/go.mod h1:yxzUCTP/U+FzoxfdKmLaA0RV1WgE0VY7hXBwKtY/4ww=
+golang.org/x/sys v0.41.0 h1:Ivj+2Cp/ylzLiEU89QhWblYnOE9zerudt9Ftecq2C6k=
+golang.org/x/sys v0.41.0/go.mod h1:OgkHotnGiDImocRcuBABYBEXf8A9a87e/uXjp9XT3ks=
+golang.org/x/term v0.40.0 h1:36e4zGLqU4yhjlmxEaagx2KuYbJq3EwY8K943ZsHcvg=
+golang.org/x/term v0.40.0/go.mod h1:w2P8uVp06p2iyKKuvXIm7N/y0UCRt3UfJTfZ7oOpglM=
 golang.org/x/text v0.3.0/go.mod h1:NqM8EUOU14njkJ3fqMW+pc6Ldnwhi/IjpwHt7yyuwOQ=
 golang.org/x/text v0.3.2/go.mod h1:bEr9sfX3Q8Zfm5fL9x+3itogRgK3+ptLWKqgva+5dAk=
-golang.org/x/text v0.33.0 h1:B3njUFyqtHDUI5jMn1YIr5B0IE2U0qck04r6d4KPAxE=
-golang.org/x/text v0.33.0/go.mod h1:LuMebE6+rBincTi9+xWTY8TztLzKHc/9C1uBCG27+q8=
+golang.org/x/text v0.34.0 h1:oL/Qq0Kdaqxa1KbNeMKwQq0reLCCaFtqu2eNuSeNHbk=
+golang.org/x/text v0.34.0/go.mod h1:homfLqTYRFyVYemLBFl5GgL/DWEiH5wcsQ5gSh1yziA=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.41.0 h1:a9b8iMweWG+S0OBnlU36rzLp20z1Rp10w+IY2czHTQc=
-golang.org/x/tools v0.41.0/go.mod h1:XSY6eDqxVNiYgezAVqqCeihT4j1U2CCsqvH3WhQpnlg=
+golang.org/x/tools v0.42.0 h1:uNgphsn75Tdz5Ji2q36v/nsFSfR/9BRFvqhGBaJGd5k=
+golang.org/x/tools v0.42.0/go.mod h1:Ma6lCIwGZvHK6XtgbswSoWroEkhugApmsXyrUmBhfr0=
 google.golang.org/appengine v1.6.5/go.mod h1:8WjMMxjGQR8xUklV/ARdw2HLXBOI7O7uCIDZVag1xfc=
-google.golang.org/protobuf v1.36.10 h1:AYd7cD/uASjIL6Q9LiTjz8JLcrh/88q5UObnmY3aOOE=
-google.golang.org/protobuf v1.36.10/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
+google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
+google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
 gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
 gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
@@ -195,8 +195,8 @@ modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8=
 modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns=
 modernc.org/sortutil v1.2.1 h1:+xyoGf15mM3NMlPDnFqrteY07klSFxLElE2PVuWIJ7w=
 modernc.org/sortutil v1.2.1/go.mod h1:7ZI3a3REbai7gzCLcotuw9AC4VZVpYMjDzETGsSMqJE=
-modernc.org/sqlite v1.44.3 h1:+39JvV/HWMcYslAwRxHb8067w+2zowvFOUrOWIy9PjY=
-modernc.org/sqlite v1.44.3/go.mod h1:CzbrU2lSB1DKUusvwGz7rqEKIq+NUd8GWuBBZDs9/nA=
+modernc.org/sqlite v1.45.0 h1:r51cSGzKpbptxnby+EIIz5fop4VuE4qFoVEjNvWoObs=
+modernc.org/sqlite v1.45.0/go.mod h1:CzbrU2lSB1DKUusvwGz7rqEKIq+NUd8GWuBBZDs9/nA=
 modernc.org/strutil v1.2.1 h1:UneZBkQA+DX2Rp35KcM69cSsNES9ly8mQWD71HKlOA0=
 modernc.org/strutil v1.2.1/go.mod h1:EHkiggD70koQxjVdSBM3JKM7k6L0FbGE5eymy9i3B9A=
 modernc.org/token v1.1.0 h1:Xl7Ap9dKaEs5kLoOQeQmPWevfnk/DM5qcLcYlA8ys6Y=

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"`
+}
+
+// Values pulled from system_stats.stats that are relevant to alerts.
 type SystemAlertStats struct {
 	Cpu          float64                       `json:"cpu"`
 	Mem          float64                       `json:"mp"`
 	Disk         float64                       `json:"dp"`
-	NetSent      float64                       `json:"ns"`
-	NetRecv      float64                       `json:"nr"`
+	Bandwidth    [2]uint64                     `json:"b"`
 	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 {
@@ -259,13 +265,14 @@ func (am *AlertManager) SendShoutrrrAlert(notificationUrl, title, message, link,
 	}
 
 	// Add link
-	if scheme == "ntfy" {
+	switch scheme {
+	case "ntfy":
 		queryParams.Add("Actions", fmt.Sprintf("view, %s, %s", linkText, link))
-	} else if scheme == "lark" {
+	case "lark":
 		queryParams.Add("link", link)
-	} else if scheme == "bark" {
+	case "bark":
 		queryParams.Add("url", link)
-	} else {
+	default:
 		message += "\n\n" + link
 	}
 

internal/alerts/alerts_battery_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package alerts_test
 

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_quiet_hours_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package alerts_test
 

internal/alerts/alerts_smart_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package alerts_test
 

internal/alerts/alerts_system.go

@@ -11,7 +11,6 @@ import (
 	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/pocketbase/pocketbase/tools/types"
-	"github.com/spf13/cast"
 )
 
 func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error {
@@ -38,7 +37,7 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 		case "Memory":
 			val = data.Info.MemPct
 		case "Bandwidth":
-			val = data.Info.Bandwidth
+			val = float64(data.Info.BandwidthBytes) / (1024 * 1024)
 			unit = " MB/s"
 		case "Disk":
 			maxUsedPct := data.Info.DiskPct
@@ -92,7 +91,7 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 			}
 		}
 
-		min := max(1, cast.ToUint8(alertRecord.Get("min")))
+		min := max(1, uint8(alertRecord.GetInt("min")))
 
 		alert := SystemAlertData{
 			systemRecord: systemRecord,
@@ -192,22 +191,24 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 			case "Memory":
 				alert.val += stats.Mem
 			case "Bandwidth":
-				alert.val += stats.NetSent + stats.NetRecv
+				alert.val += float64(stats.Bandwidth[0]+stats.Bandwidth[1]) / (1024 * 1024)
 			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 {

internal/alerts/alerts_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package alerts_test
 

internal/alerts/alerts_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package alerts
 

internal/cmd/agent/agent.go

@@ -9,6 +9,7 @@ import (
 	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/agent"
 	"github.com/henrygd/beszel/agent/health"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/spf13/pflag"
 	"golang.org/x/crypto/ssh"
 )
@@ -116,12 +117,12 @@ func (opts *cmdOptions) loadPublicKeys() ([]ssh.PublicKey, error) {
 	}
 
 	// Try environment variable
-	if key, ok := agent.GetEnv("KEY"); ok && key != "" {
+	if key, ok := utils.GetEnv("KEY"); ok && key != "" {
 		return agent.ParseKeys(key)
 	}
 
 	// Try key file
-	keyFile, ok := agent.GetEnv("KEY_FILE")
+	keyFile, ok := utils.GetEnv("KEY_FILE")
 	if !ok {
 		return nil, fmt.Errorf("no key provided: must set -key flag, KEY env var, or KEY_FILE env var. Use 'beszel-agent help' for usage")
 	}

internal/entities/smart/smart.go

@@ -143,8 +143,8 @@ type AtaDeviceStatisticsPage struct {
 }
 
 type AtaDeviceStatisticsEntry struct {
-	Name  string  `json:"name"`
-	Value *uint64 `json:"value,omitempty"`
+	Name  string `json:"name"`
+	Value *int64 `json:"value,omitempty"`
 }
 
 type AtaSmartAttribute struct {
@@ -356,8 +356,8 @@ type SmartInfoForSata struct {
 	SmartStatus SmartStatusInfo `json:"smart_status"`
 	// AtaSmartData                 AtaSmartData                 `json:"ata_smart_data"`
 	// AtaSctCapabilities           AtaSctCapabilities           `json:"ata_sct_capabilities"`
-	AtaSmartAttributes  AtaSmartAttributes  `json:"ata_smart_attributes"`
-	AtaDeviceStatistics AtaDeviceStatistics `json:"ata_device_statistics"`
+	AtaSmartAttributes  AtaSmartAttributes `json:"ata_smart_attributes"`
+	AtaDeviceStatistics json.RawMessage    `json:"ata_device_statistics"`
 	// PowerOnTime                  PowerOnTimeInfo              `json:"power_on_time"`
 	// PowerCycleCount              uint16                       `json:"power_cycle_count"`
 	Temperature TemperatureInfo `json:"temperature"`

internal/entities/system/system.go

@@ -12,8 +12,9 @@ import (
 
 type Stats struct {
 	Cpu            float64             `json:"cpu" cbor:"0,keyasint"`
-	MaxCpu         float64             `json:"cpum,omitempty" cbor:"1,keyasint,omitempty"`
+	MaxCpu         float64             `json:"cpum,omitempty" cbor:"-"`
 	Mem            float64             `json:"m" cbor:"2,keyasint"`
+	MaxMem         float64             `json:"mm,omitempty" cbor:"-"`
 	MemUsed        float64             `json:"mu" cbor:"3,keyasint"`
 	MemPct         float64             `json:"mp" cbor:"4,keyasint"`
 	MemBuffCache   float64             `json:"mb" cbor:"5,keyasint"`
@@ -23,26 +24,25 @@ type Stats struct {
 	DiskTotal      float64             `json:"d" cbor:"9,keyasint"`
 	DiskUsed       float64             `json:"du" cbor:"10,keyasint"`
 	DiskPct        float64             `json:"dp" cbor:"11,keyasint"`
-	DiskReadPs     float64             `json:"dr" cbor:"12,keyasint"`
-	DiskWritePs    float64             `json:"dw" cbor:"13,keyasint"`
-	MaxDiskReadPs  float64             `json:"drm,omitempty" cbor:"14,keyasint,omitempty"`
-	MaxDiskWritePs float64             `json:"dwm,omitempty" cbor:"15,keyasint,omitempty"`
+	DiskReadPs     float64             `json:"dr,omitzero" cbor:"12,keyasint,omitzero"`
+	DiskWritePs    float64             `json:"dw,omitzero" cbor:"13,keyasint,omitzero"`
+	MaxDiskReadPs  float64             `json:"drm,omitempty" cbor:"-"`
+	MaxDiskWritePs float64             `json:"dwm,omitempty" cbor:"-"`
 	NetworkSent    float64             `json:"ns,omitzero" cbor:"16,keyasint,omitzero"`
 	NetworkRecv    float64             `json:"nr,omitzero" cbor:"17,keyasint,omitzero"`
-	MaxNetworkSent float64             `json:"nsm,omitempty" cbor:"18,keyasint,omitempty"`
-	MaxNetworkRecv float64             `json:"nrm,omitempty" cbor:"19,keyasint,omitempty"`
+	MaxNetworkSent float64             `json:"nsm,omitempty" cbor:"-"`
+	MaxNetworkRecv float64             `json:"nrm,omitempty" cbor:"-"`
 	Temperatures   map[string]float64  `json:"t,omitempty" cbor:"20,keyasint,omitempty"`
 	ExtraFs        map[string]*FsStats `json:"efs,omitempty" cbor:"21,keyasint,omitempty"`
 	GPUData        map[string]GPUData  `json:"g,omitempty" cbor:"22,keyasint,omitempty"`
-	LoadAvg1       float64             `json:"l1,omitempty" cbor:"23,keyasint,omitempty"`
-	LoadAvg5       float64             `json:"l5,omitempty" cbor:"24,keyasint,omitempty"`
-	LoadAvg15      float64             `json:"l15,omitempty" cbor:"25,keyasint,omitempty"`
-	Bandwidth      [2]uint64           `json:"b,omitzero" cbor:"26,keyasint,omitzero"`  // [sent bytes, recv bytes]
-	MaxBandwidth   [2]uint64           `json:"bm,omitzero" cbor:"27,keyasint,omitzero"` // [sent bytes, recv bytes]
+	// LoadAvg1       float64             `json:"l1,omitempty" cbor:"23,keyasint,omitempty"`
+	// LoadAvg5       float64             `json:"l5,omitempty" cbor:"24,keyasint,omitempty"`
+	// LoadAvg15      float64             `json:"l15,omitempty" cbor:"25,keyasint,omitempty"`
+	Bandwidth    [2]uint64 `json:"b,omitzero" cbor:"26,keyasint,omitzero"` // [sent bytes, recv bytes]
+	MaxBandwidth [2]uint64 `json:"bm,omitzero" cbor:"-"`                   // [sent bytes, recv bytes]
 	// TODO: remove other load fields in future release in favor of load avg array
 	LoadAvg           [3]float64           `json:"la,omitempty" cbor:"28,keyasint"`
-	Battery           [2]uint8             `json:"bat,omitzero" cbor:"29,keyasint,omitzero"` // [percent, charge state, current]
-	MaxMem            float64              `json:"mm,omitempty" cbor:"30,keyasint,omitempty"`
+	Battery           [2]uint8             `json:"bat,omitzero" cbor:"29,keyasint,omitzero"`    // [percent, charge state, current]
 	NetworkInterfaces map[string][4]uint64 `json:"ni,omitempty" cbor:"31,keyasint,omitempty"`   // [upload bytes, download bytes, total upload, total download]
 	DiskIO            [2]uint64            `json:"dio,omitzero" cbor:"32,keyasint,omitzero"`    // [read bytes, write bytes]
 	MaxDiskIO         [2]uint64            `json:"diom,omitzero" cbor:"-"`                      // [max read bytes, max write bytes]
@@ -90,8 +90,8 @@ type FsStats struct {
 	TotalWrite     uint64    `json:"-"`
 	DiskReadPs     float64   `json:"r" cbor:"2,keyasint"`
 	DiskWritePs    float64   `json:"w" cbor:"3,keyasint"`
-	MaxDiskReadPS  float64   `json:"rm,omitempty" cbor:"4,keyasint,omitempty"`
-	MaxDiskWritePS float64   `json:"wm,omitempty" cbor:"5,keyasint,omitempty"`
+	MaxDiskReadPS  float64   `json:"rm,omitempty" cbor:"-"`
+	MaxDiskWritePS float64   `json:"wm,omitempty" cbor:"-"`
 	// TODO: remove DiskReadPs and DiskWritePs in future release in favor of DiskReadBytes and DiskWriteBytes
 	DiskReadBytes     uint64 `json:"rb" cbor:"6,keyasint,omitempty"`
 	DiskWriteBytes    uint64 `json:"wb" cbor:"7,keyasint,omitempty"`
@@ -129,23 +129,23 @@ type Info struct {
 	KernelVersion string `json:"k,omitempty" cbor:"1,keyasint,omitempty"` // deprecated - moved to Details struct
 	Cores         int    `json:"c,omitzero" cbor:"2,keyasint,omitzero"`   // deprecated - moved to Details struct
 	// Threads is needed in Info struct to calculate load average thresholds
-	Threads        int     `json:"t,omitempty" cbor:"3,keyasint,omitempty"`
-	CpuModel       string  `json:"m,omitempty" cbor:"4,keyasint,omitempty"` // deprecated - moved to Details struct
-	Uptime         uint64  `json:"u" cbor:"5,keyasint"`
-	Cpu            float64 `json:"cpu" cbor:"6,keyasint"`
-	MemPct         float64 `json:"mp" cbor:"7,keyasint"`
-	DiskPct        float64 `json:"dp" cbor:"8,keyasint"`
-	Bandwidth      float64 `json:"b" cbor:"9,keyasint"`
-	AgentVersion   string  `json:"v" cbor:"10,keyasint"`
-	Podman         bool    `json:"p,omitempty" cbor:"11,keyasint,omitempty"` // deprecated - moved to Details struct
-	GpuPct         float64 `json:"g,omitempty" cbor:"12,keyasint,omitempty"`
-	DashboardTemp  float64 `json:"dt,omitempty" cbor:"13,keyasint,omitempty"`
-	Os             Os      `json:"os,omitempty" cbor:"14,keyasint,omitempty"`  // deprecated - moved to Details struct
-	LoadAvg1       float64 `json:"l1,omitempty" cbor:"15,keyasint,omitempty"`  // deprecated - use `la` array instead
-	LoadAvg5       float64 `json:"l5,omitempty" cbor:"16,keyasint,omitempty"`  // deprecated - use `la` array instead
-	LoadAvg15      float64 `json:"l15,omitempty" cbor:"17,keyasint,omitempty"` // deprecated - use `la` array instead
-	BandwidthBytes uint64  `json:"bb" cbor:"18,keyasint"`
+	Threads       int     `json:"t,omitempty" cbor:"3,keyasint,omitempty"`
+	CpuModel      string  `json:"m,omitempty" cbor:"4,keyasint,omitempty"` // deprecated - moved to Details struct
+	Uptime        uint64  `json:"u" cbor:"5,keyasint"`
+	Cpu           float64 `json:"cpu" cbor:"6,keyasint"`
+	MemPct        float64 `json:"mp" cbor:"7,keyasint"`
+	DiskPct       float64 `json:"dp" cbor:"8,keyasint"`
+	Bandwidth     float64 `json:"b,omitzero" cbor:"9,keyasint"` // deprecated in favor of BandwidthBytes
+	AgentVersion  string  `json:"v" cbor:"10,keyasint"`
+	Podman        bool    `json:"p,omitempty" cbor:"11,keyasint,omitempty"` // deprecated - moved to Details struct
+	GpuPct        float64 `json:"g,omitempty" cbor:"12,keyasint,omitempty"`
+	DashboardTemp float64 `json:"dt,omitempty" cbor:"13,keyasint,omitempty"`
+	Os            Os      `json:"os,omitempty" cbor:"14,keyasint,omitempty"` // deprecated - moved to Details struct
+	// LoadAvg1       float64 `json:"l1,omitempty" cbor:"15,keyasint,omitempty"`  // deprecated - use `la` array instead
+	// LoadAvg5       float64 `json:"l5,omitempty" cbor:"16,keyasint,omitempty"`  // deprecated - use `la` array instead
+	// LoadAvg15      float64 `json:"l15,omitempty" cbor:"17,keyasint,omitempty"` // deprecated - use `la` array instead
 
+	BandwidthBytes uint64             `json:"bb" cbor:"18,keyasint"`
 	LoadAvg        [3]float64         `json:"la,omitempty" cbor:"19,keyasint"`
 	ConnectionType ConnectionType     `json:"ct,omitempty" cbor:"20,keyasint,omitempty,omitzero"`
 	ExtraFsPct     map[string]float64 `json:"efs,omitempty" cbor:"21,keyasint,omitempty"`

internal/ghupdate/ghupdate.go

@@ -34,7 +34,7 @@ func ColorPrint(color, text string) {
 	fmt.Println(color + text + colorReset)
 }
 
-func ColorPrintf(color, format string, args ...interface{}) {
+func ColorPrintf(color, format string, args ...any) {
 	fmt.Printf(color+format+colorReset+"\n", args...)
 }
 

internal/hub/agent_connect_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package hub
 
@@ -10,6 +9,7 @@ import (
 	"net/http/httptest"
 	"os"
 	"path/filepath"
+	"runtime"
 	"strings"
 	"testing"
 	"time"
@@ -35,6 +35,26 @@ func createTestHub(t testing.TB) (*Hub, *pbtests.TestApp, error) {
 	return NewHub(testApp), testApp, nil
 }
 
+// cleanupTestHub stops background system goroutines before tearing down the app.
+func cleanupTestHub(hub *Hub, testApp *pbtests.TestApp) {
+	if hub != nil {
+		sm := hub.GetSystemManager()
+		sm.RemoveAllSystems()
+		// Give updater goroutines a brief window to observe cancellation before DB teardown.
+		for range 20 {
+			if sm.GetSystemCount() == 0 {
+				break
+			}
+			runtime.Gosched()
+			time.Sleep(5 * time.Millisecond)
+		}
+		time.Sleep(20 * time.Millisecond)
+	}
+	if testApp != nil {
+		testApp.Cleanup()
+	}
+}
+
 // Helper function to create a test record
 func createTestRecord(app core.App, collection string, data map[string]any) (*core.Record, error) {
 	col, err := app.FindCachedCollectionByNameOrId(collection)
@@ -64,7 +84,7 @@ func TestValidateAgentHeaders(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	testCases := []struct {
 		name          string
@@ -145,7 +165,7 @@ func TestGetAllFingerprintRecordsByToken(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// create test user
 	userRecord, err := createTestUser(testApp)
@@ -235,7 +255,7 @@ func TestSetFingerprint(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)
@@ -315,7 +335,7 @@ func TestCreateSystemFromAgentData(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)
@@ -425,7 +445,7 @@ func TestUniversalTokenFlow(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(nil, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)
@@ -493,7 +513,7 @@ func TestAgentConnect(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)
@@ -652,7 +672,7 @@ func TestHandleAgentConnect(t *testing.T) {
 	if err != nil {
 		t.Fatal(err)
 	}
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)
@@ -737,7 +757,7 @@ func TestAgentWebSocketIntegration(t *testing.T) {
 	// Create hub and test app
 	hub, testApp, err := createTestHub(t)
 	require.NoError(t, err)
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Get the hub's SSH key
 	hubSigner, err := hub.GetSSHKey("")
@@ -897,7 +917,7 @@ func TestAgentWebSocketIntegration(t *testing.T) {
 
 			// Wait for connection result
 			maxWait := 2 * time.Second
-			time.Sleep(20 * time.Millisecond)
+			time.Sleep(40 * time.Millisecond)
 			checkInterval := 20 * time.Millisecond
 			timeout := time.After(maxWait)
 			ticker := time.Tick(checkInterval)
@@ -942,6 +962,8 @@ func TestAgentWebSocketIntegration(t *testing.T) {
 				}
 			}
 
+			time.Sleep(20 * time.Millisecond)
+
 			// Verify fingerprint state by re-reading the specific record
 			updatedFingerprintRecord, err := testApp.FindRecordById("fingerprints", fingerprintRecord.Id)
 			require.NoError(t, err)
@@ -976,7 +998,7 @@ func TestMultipleSystemsWithSameUniversalToken(t *testing.T) {
 	// Create hub and test app
 	hub, testApp, err := createTestHub(t)
 	require.NoError(t, err)
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Get the hub's SSH key
 	hubSigner, err := hub.GetSSHKey("")
@@ -1144,6 +1166,8 @@ func TestMultipleSystemsWithSameUniversalToken(t *testing.T) {
 					assert.Equal(t, systemCount, systemsAfterCount, "Total system count should remain the same")
 				}
 
+				time.Sleep(20 * time.Millisecond)
+
 				// Verify that a fingerprint record exists for this fingerprint
 				fingerprints, err := testApp.FindRecordsByFilter("fingerprints", "token = {:token} && fingerprint = {:fingerprint}", "", -1, 0, map[string]any{
 					"token":       universalToken,
@@ -1176,7 +1200,7 @@ func TestPermanentUniversalTokenFromDB(t *testing.T) {
 	// Create hub and test app
 	hub, testApp, err := createTestHub(t)
 	require.NoError(t, err)
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Get the hub's SSH key
 	hubSigner, err := hub.GetSSHKey("")
@@ -1273,7 +1297,7 @@ verify:
 func TestFindOrCreateSystemForToken(t *testing.T) {
 	hub, testApp, err := createTestHub(t)
 	require.NoError(t, err)
-	defer testApp.Cleanup()
+	defer cleanupTestHub(hub, testApp)
 
 	// Create test user
 	userRecord, err := createTestUser(testApp)

internal/hub/config/config_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package config_test
 

internal/hub/expirymap/expirymap.go

@@ -1,35 +1,39 @@
+// Package expirymap provides a thread-safe map with expiring entries.
+// It supports TTL-based expiration with both lazy cleanup on access
+// and periodic background cleanup.
 package expirymap
 
 import (
-	"reflect"
+	"sync"
 	"time"
 
 	"github.com/pocketbase/pocketbase/tools/store"
 )
 
-type val[T any] struct {
+type val[T comparable] struct {
 	value   T
 	expires time.Time
 }
 
-type ExpiryMap[T any] struct {
-	store           *store.Store[string, *val[T]]
-	cleanupInterval time.Duration
+type ExpiryMap[T comparable] struct {
+	store    store.Store[string, val[T]]
+	stopChan chan struct{}
+	stopOnce sync.Once
 }
 
 // New creates a new expiry map with custom cleanup interval
-func New[T any](cleanupInterval time.Duration) *ExpiryMap[T] {
+func New[T comparable](cleanupInterval time.Duration) *ExpiryMap[T] {
 	m := &ExpiryMap[T]{
-		store:           store.New(map[string]*val[T]{}),
-		cleanupInterval: cleanupInterval,
+		store:    *store.New(map[string]val[T]{}),
+		stopChan: make(chan struct{}),
 	}
-	m.startCleaner()
+	go m.startCleaner(cleanupInterval)
 	return m
 }
 
 // Set stores a value with the given TTL
 func (m *ExpiryMap[T]) Set(key string, value T, ttl time.Duration) {
-	m.store.Set(key, &val[T]{
+	m.store.Set(key, val[T]{
 		value:   value,
 		expires: time.Now().Add(ttl),
 	})
@@ -55,7 +59,7 @@ func (m *ExpiryMap[T]) GetOk(key string) (T, bool) {
 // GetByValue retrieves a value by value
 func (m *ExpiryMap[T]) GetByValue(val T) (key string, value T, ok bool) {
 	for key, v := range m.store.GetAll() {
-		if reflect.DeepEqual(v.value, val) {
+		if v.value == val {
 			// check if expired
 			if v.expires.Before(time.Now()) {
 				m.store.Remove(key)
@@ -75,7 +79,7 @@ func (m *ExpiryMap[T]) Remove(key string) {
 // RemovebyValue removes a value by value
 func (m *ExpiryMap[T]) RemovebyValue(value T) (T, bool) {
 	for key, val := range m.store.GetAll() {
-		if reflect.DeepEqual(val.value, value) {
+		if val.value == value {
 			m.store.Remove(key)
 			return val.value, true
 		}
@@ -84,13 +88,23 @@ func (m *ExpiryMap[T]) RemovebyValue(value T) (T, bool) {
 }
 
 // startCleaner runs the background cleanup process
-func (m *ExpiryMap[T]) startCleaner() {
-	go func() {
-		tick := time.Tick(m.cleanupInterval)
-		for range tick {
+func (m *ExpiryMap[T]) startCleaner(interval time.Duration) {
+	tick := time.Tick(interval)
+	for {
+		select {
+		case <-tick:
 			m.cleanup()
+		case <-m.stopChan:
+			return
 		}
-	}()
+	}
+}
+
+// StopCleaner stops the background cleanup process
+func (m *ExpiryMap[T]) StopCleaner() {
+	m.stopOnce.Do(func() {
+		close(m.stopChan)
+	})
 }
 
 // cleanup removes all expired entries
@@ -102,3 +116,12 @@ func (m *ExpiryMap[T]) cleanup() {
 		}
 	}
 }
+
+// UpdateExpiration updates the expiration time of a key
+func (m *ExpiryMap[T]) UpdateExpiration(key string, ttl time.Duration) {
+	value, ok := m.store.GetOk(key)
+	if ok {
+		value.expires = time.Now().Add(ttl)
+		m.store.Set(key, value)
+	}
+}

internal/hub/expirymap/expirymap_test.go

@@ -1,10 +1,10 @@
 //go:build testing
-// +build testing
 
 package expirymap
 
 import (
 	"testing"
+	"testing/synctest"
 	"time"
 
 	"github.com/stretchr/testify/assert"
@@ -178,6 +178,33 @@ func TestExpiryMap_GenericTypes(t *testing.T) {
 	})
 }
 
+func TestExpiryMap_UpdateExpiration(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Set a value with short TTL
+	em.Set("key1", "value1", time.Millisecond*50)
+
+	// Verify it exists
+	assert.True(t, em.Has("key1"))
+
+	// Update expiration to a longer TTL
+	em.UpdateExpiration("key1", time.Hour)
+
+	// Wait for the original TTL to pass
+	time.Sleep(time.Millisecond * 100)
+
+	// Should still exist because expiration was updated
+	assert.True(t, em.Has("key1"))
+	value, ok := em.GetOk("key1")
+	assert.True(t, ok)
+	assert.Equal(t, "value1", value)
+
+	// Try updating non-existent key (should not panic)
+	assert.NotPanics(t, func() {
+		em.UpdateExpiration("nonexistent", time.Hour)
+	})
+}
+
 func TestExpiryMap_ZeroValues(t *testing.T) {
 	em := New[string](time.Hour)
 
@@ -474,3 +501,52 @@ func TestExpiryMap_ValueOperations_Integration(t *testing.T) {
 	assert.Equal(t, "unique", value)
 	assert.Equal(t, "key2", key)
 }
+
+func TestExpiryMap_Cleaner(t *testing.T) {
+	synctest.Test(t, func(t *testing.T) {
+		em := New[string](time.Second)
+		defer em.StopCleaner()
+
+		em.Set("test", "value", 500*time.Millisecond)
+
+		// Wait 600ms, value is expired but cleaner hasn't run yet (interval is 1s)
+		time.Sleep(600 * time.Millisecond)
+		synctest.Wait()
+
+		// Map should still hold the value in its internal store before lazy access or cleaner
+		assert.Equal(t, 1, len(em.store.GetAll()), "store should still have 1 item before cleaner runs")
+
+		// Wait another 500ms so cleaner (1s interval) runs
+		time.Sleep(500 * time.Millisecond)
+		synctest.Wait() // Wait for background goroutine to process the tick
+
+		assert.Equal(t, 0, len(em.store.GetAll()), "store should be empty after cleaner runs")
+	})
+}
+
+func TestExpiryMap_StopCleaner(t *testing.T) {
+	em := New[string](time.Hour)
+
+	// Initially, stopChan is open, reading would block
+	select {
+	case <-em.stopChan:
+		t.Fatal("stopChan should be open initially")
+	default:
+		// success
+	}
+
+	em.StopCleaner()
+
+	// After StopCleaner, stopChan is closed, reading returns immediately
+	select {
+	case <-em.stopChan:
+		// success
+	default:
+		t.Fatal("stopChan was not closed by StopCleaner")
+	}
+
+	// Calling StopCleaner again should NOT panic thanks to sync.Once
+	assert.NotPanics(t, func() {
+		em.StopCleaner()
+	})
+}

internal/hub/heartbeat/heartbeat.go

@@ -0,0 +1,303 @@
+// Package heartbeat sends periodic outbound pings to an external monitoring
+// endpoint (e.g. BetterStack, Uptime Kuma, Healthchecks.io) so operators can
+// monitor Beszel without exposing it to the internet.
+package heartbeat
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"net/url"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/henrygd/beszel"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// Default values for heartbeat configuration.
+const (
+	defaultInterval = 60 // seconds
+	httpTimeout     = 10 * time.Second
+)
+
+// Payload is the JSON body sent with each heartbeat request.
+type Payload struct {
+	// Status is "ok" when all non-paused systems are up, "warn" when alerts
+	// are triggered but no systems are down, and "error" when any system is down.
+	Status    string         `json:"status"`
+	Timestamp string         `json:"timestamp"`
+	Msg       string         `json:"msg"`
+	Systems   SystemsSummary `json:"systems"`
+	Down      []SystemInfo   `json:"down_systems,omitempty"`
+	Alerts    []AlertInfo    `json:"triggered_alerts,omitempty"`
+	Version   string         `json:"beszel_version"`
+}
+
+// SystemsSummary contains counts of systems by status.
+type SystemsSummary struct {
+	Total   int `json:"total"`
+	Up      int `json:"up"`
+	Down    int `json:"down"`
+	Paused  int `json:"paused"`
+	Pending int `json:"pending"`
+}
+
+// SystemInfo identifies a system that is currently down.
+type SystemInfo struct {
+	ID   string `json:"id" db:"id"`
+	Name string `json:"name" db:"name"`
+	Host string `json:"host" db:"host"`
+}
+
+// AlertInfo describes a currently triggered alert.
+type AlertInfo struct {
+	SystemID   string  `json:"system_id"`
+	SystemName string  `json:"system_name"`
+	AlertName  string  `json:"alert_name"`
+	Threshold  float64 `json:"threshold"`
+}
+
+// Config holds heartbeat settings read from environment variables.
+type Config struct {
+	URL      string // endpoint to ping
+	Interval int    // seconds between pings
+	Method   string // HTTP method (GET or POST, default POST)
+}
+
+// Heartbeat manages the periodic outbound health check.
+type Heartbeat struct {
+	app    core.App
+	config Config
+	client *http.Client
+}
+
+// New creates a Heartbeat if configuration is present.
+// Returns nil if HEARTBEAT_URL is not set (feature disabled).
+func New(app core.App, getEnv func(string) (string, bool)) *Heartbeat {
+	url, _ := getEnv("HEARTBEAT_URL")
+	url = strings.TrimSpace(url)
+	if app == nil || url == "" {
+		return nil
+	}
+
+	interval := defaultInterval
+	if v, ok := getEnv("HEARTBEAT_INTERVAL"); ok {
+		if parsed, err := strconv.Atoi(v); err == nil && parsed > 0 {
+			interval = parsed
+		}
+	}
+
+	method := http.MethodPost
+	if v, ok := getEnv("HEARTBEAT_METHOD"); ok {
+		v = strings.ToUpper(strings.TrimSpace(v))
+		if v == http.MethodGet || v == http.MethodHead {
+			method = v
+		}
+	}
+
+	return &Heartbeat{
+		app: app,
+		config: Config{
+			URL:      url,
+			Interval: interval,
+			Method:   method,
+		},
+		client: &http.Client{Timeout: httpTimeout},
+	}
+}
+
+// Start begins the heartbeat loop. It blocks and should be called in a goroutine.
+// The loop runs until the provided stop channel is closed.
+func (hb *Heartbeat) Start(stop <-chan struct{}) {
+	sanitizedURL := sanitizeHeartbeatURL(hb.config.URL)
+	hb.app.Logger().Info("Heartbeat enabled",
+		"url", sanitizedURL,
+		"interval", fmt.Sprintf("%ds", hb.config.Interval),
+		"method", hb.config.Method,
+	)
+
+	// Send an initial heartbeat immediately on startup.
+	hb.send()
+
+	ticker := time.NewTicker(time.Duration(hb.config.Interval) * time.Second)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-stop:
+			return
+		case <-ticker.C:
+			hb.send()
+		}
+	}
+}
+
+// Send performs a single heartbeat ping. Exposed for the test-heartbeat API endpoint.
+func (hb *Heartbeat) Send() error {
+	return hb.send()
+}
+
+// GetConfig returns the current heartbeat configuration.
+func (hb *Heartbeat) GetConfig() Config {
+	return hb.config
+}
+
+func (hb *Heartbeat) send() error {
+	var req *http.Request
+	var err error
+	method := normalizeMethod(hb.config.Method)
+
+	if method == http.MethodGet || method == http.MethodHead {
+		req, err = http.NewRequest(method, hb.config.URL, nil)
+	} else {
+		payload, payloadErr := hb.buildPayload()
+		if payloadErr != nil {
+			hb.app.Logger().Error("Heartbeat: failed to build payload", "err", payloadErr)
+			return payloadErr
+		}
+
+		body, jsonErr := json.Marshal(payload)
+		if jsonErr != nil {
+			hb.app.Logger().Error("Heartbeat: failed to marshal payload", "err", jsonErr)
+			return jsonErr
+		}
+		req, err = http.NewRequest(http.MethodPost, hb.config.URL, bytes.NewReader(body))
+		if err == nil {
+			req.Header.Set("Content-Type", "application/json")
+		}
+	}
+
+	if err != nil {
+		hb.app.Logger().Error("Heartbeat: failed to create request", "err", err)
+		return err
+	}
+
+	req.Header.Set("User-Agent", "Beszel-Heartbeat")
+
+	resp, err := hb.client.Do(req)
+	if err != nil {
+		hb.app.Logger().Error("Heartbeat: request failed", "url", sanitizeHeartbeatURL(hb.config.URL), "err", err)
+		return err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode >= 400 {
+		hb.app.Logger().Warn("Heartbeat: non-success response",
+			"url", sanitizeHeartbeatURL(hb.config.URL),
+			"status", resp.StatusCode,
+		)
+		return fmt.Errorf("heartbeat endpoint returned status %d", resp.StatusCode)
+	}
+
+	return nil
+}
+
+func (hb *Heartbeat) buildPayload() (*Payload, error) {
+	db := hb.app.DB()
+
+	// Count systems by status.
+	var systemCounts []struct {
+		Status string `db:"status"`
+		Count  int    `db:"cnt"`
+	}
+	err := db.NewQuery("SELECT status, COUNT(*) as cnt FROM systems GROUP BY status").All(&systemCounts)
+	if err != nil {
+		return nil, fmt.Errorf("query system counts: %w", err)
+	}
+
+	summary := SystemsSummary{}
+	for _, sc := range systemCounts {
+		switch sc.Status {
+		case "up":
+			summary.Up = sc.Count
+		case "down":
+			summary.Down = sc.Count
+		case "paused":
+			summary.Paused = sc.Count
+		case "pending":
+			summary.Pending = sc.Count
+		}
+		summary.Total += sc.Count
+	}
+
+	// Get names of down systems.
+	var downSystems []SystemInfo
+	if summary.Down > 0 {
+		err = db.NewQuery("SELECT id, name, host FROM systems WHERE status = 'down'").All(&downSystems)
+		if err != nil {
+			return nil, fmt.Errorf("query down systems: %w", err)
+		}
+	}
+
+	// Get triggered alerts with system names.
+	var triggeredAlerts []struct {
+		SystemID   string  `db:"system"`
+		SystemName string  `db:"system_name"`
+		AlertName  string  `db:"name"`
+		Value      float64 `db:"value"`
+	}
+	err = db.NewQuery(`
+		SELECT a.system, s.name as system_name, a.name, a.value
+		FROM alerts a
+		JOIN systems s ON a.system = s.id
+		WHERE a.triggered = true
+	`).All(&triggeredAlerts)
+	if err != nil {
+		// Non-fatal: alerts info is supplementary.
+		triggeredAlerts = nil
+	}
+
+	alerts := make([]AlertInfo, 0, len(triggeredAlerts))
+	for _, ta := range triggeredAlerts {
+		alerts = append(alerts, AlertInfo{
+			SystemID:   ta.SystemID,
+			SystemName: ta.SystemName,
+			AlertName:  ta.AlertName,
+			Threshold:  ta.Value,
+		})
+	}
+
+	// Determine overall status.
+	status := "ok"
+	msg := "All systems operational"
+	if summary.Down > 0 {
+		status = "error"
+		names := make([]string, len(downSystems))
+		for i, ds := range downSystems {
+			names[i] = ds.Name
+		}
+		msg = fmt.Sprintf("%d system(s) down: %s", summary.Down, strings.Join(names, ", "))
+	} else if len(alerts) > 0 {
+		status = "warn"
+		msg = fmt.Sprintf("%d alert(s) triggered", len(alerts))
+	}
+
+	return &Payload{
+		Status:    status,
+		Timestamp: time.Now().UTC().Format(time.RFC3339),
+		Msg:       msg,
+		Systems:   summary,
+		Down:      downSystems,
+		Alerts:    alerts,
+		Version:   beszel.Version,
+	}, nil
+}
+
+func normalizeMethod(method string) string {
+	upper := strings.ToUpper(strings.TrimSpace(method))
+	if upper == http.MethodGet || upper == http.MethodHead || upper == http.MethodPost {
+		return upper
+	}
+	return http.MethodPost
+}
+
+func sanitizeHeartbeatURL(rawURL string) string {
+	parsed, err := url.Parse(strings.TrimSpace(rawURL))
+	if err != nil || parsed.Scheme == "" || parsed.Host == "" {
+		return "<invalid-url>"
+	}
+	return parsed.Scheme + "://" + parsed.Host
+}

internal/hub/heartbeat/heartbeat_test.go

@@ -0,0 +1,257 @@
+//go:build testing
+
+package heartbeat_test
+
+import (
+	"encoding/json"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"testing"
+
+	"github.com/henrygd/beszel/internal/hub/heartbeat"
+	beszeltests "github.com/henrygd/beszel/internal/tests"
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestNew(t *testing.T) {
+	t.Run("returns nil when app is missing", func(t *testing.T) {
+		hb := heartbeat.New(nil, envGetter(map[string]string{
+			"HEARTBEAT_URL": "https://heartbeat.example.com/ping",
+		}))
+		assert.Nil(t, hb)
+	})
+
+	t.Run("returns nil when URL is missing", func(t *testing.T) {
+		app := newTestHub(t)
+		hb := heartbeat.New(app.App, func(string) (string, bool) {
+			return "", false
+		})
+		assert.Nil(t, hb)
+	})
+
+	t.Run("parses and normalizes config values", func(t *testing.T) {
+		app := newTestHub(t)
+		env := map[string]string{
+			"HEARTBEAT_URL":      "  https://heartbeat.example.com/ping  ",
+			"HEARTBEAT_INTERVAL": "90",
+			"HEARTBEAT_METHOD":   "head",
+		}
+		getEnv := func(key string) (string, bool) {
+			v, ok := env[key]
+			return v, ok
+		}
+
+		hb := heartbeat.New(app.App, getEnv)
+		require.NotNil(t, hb)
+		cfg := hb.GetConfig()
+		assert.Equal(t, "https://heartbeat.example.com/ping", cfg.URL)
+		assert.Equal(t, 90, cfg.Interval)
+		assert.Equal(t, http.MethodHead, cfg.Method)
+	})
+}
+
+func TestSendGETDoesNotRequireAppOrDB(t *testing.T) {
+	app := newTestHub(t)
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		assert.Equal(t, http.MethodGet, r.Method)
+		assert.Equal(t, "Beszel-Heartbeat", r.Header.Get("User-Agent"))
+		w.WriteHeader(http.StatusOK)
+	}))
+	defer server.Close()
+
+	hb := heartbeat.New(app.App, envGetter(map[string]string{
+		"HEARTBEAT_URL":    server.URL,
+		"HEARTBEAT_METHOD": "GET",
+	}))
+	require.NotNil(t, hb)
+
+	require.NoError(t, hb.Send())
+}
+
+func TestSendReturnsErrorOnHTTPFailureStatus(t *testing.T) {
+	app := newTestHub(t)
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		w.WriteHeader(http.StatusInternalServerError)
+	}))
+	defer server.Close()
+
+	hb := heartbeat.New(app.App, envGetter(map[string]string{
+		"HEARTBEAT_URL":    server.URL,
+		"HEARTBEAT_METHOD": "GET",
+	}))
+	require.NotNil(t, hb)
+
+	err := hb.Send()
+	require.Error(t, err)
+	assert.ErrorContains(t, err, "heartbeat endpoint returned status 500")
+}
+
+func TestSendPOSTBuildsExpectedStatuses(t *testing.T) {
+	tests := []struct {
+		name           string
+		setup          func(t *testing.T, app *beszeltests.TestHub, user *core.Record)
+		expectStatus   string
+		expectMsgPart  string
+		expectDown     int
+		expectAlerts   int
+		expectTotal    int
+		expectUp       int
+		expectPaused   int
+		expectPending  int
+		expectDownSumm int
+	}{
+		{
+			name: "error when at least one system is down",
+			setup: func(t *testing.T, app *beszeltests.TestHub, user *core.Record) {
+				downSystem := createTestSystem(t, app, user.Id, "db-1", "10.0.0.1", "down")
+				_ = createTestSystem(t, app, user.Id, "web-1", "10.0.0.2", "up")
+				createTriggeredAlert(t, app, user.Id, downSystem.Id, "CPU", 95)
+			},
+			expectStatus:   "error",
+			expectMsgPart:  "1 system(s) down",
+			expectDown:     1,
+			expectAlerts:   1,
+			expectTotal:    2,
+			expectUp:       1,
+			expectDownSumm: 1,
+		},
+		{
+			name: "warn when only alerts are triggered",
+			setup: func(t *testing.T, app *beszeltests.TestHub, user *core.Record) {
+				system := createTestSystem(t, app, user.Id, "api-1", "10.1.0.1", "up")
+				createTriggeredAlert(t, app, user.Id, system.Id, "CPU", 90)
+			},
+			expectStatus:   "warn",
+			expectMsgPart:  "1 alert(s) triggered",
+			expectDown:     0,
+			expectAlerts:   1,
+			expectTotal:    1,
+			expectUp:       1,
+			expectDownSumm: 0,
+		},
+		{
+			name: "ok when no down systems and no alerts",
+			setup: func(t *testing.T, app *beszeltests.TestHub, user *core.Record) {
+				_ = createTestSystem(t, app, user.Id, "node-1", "10.2.0.1", "up")
+				_ = createTestSystem(t, app, user.Id, "node-2", "10.2.0.2", "paused")
+				_ = createTestSystem(t, app, user.Id, "node-3", "10.2.0.3", "pending")
+			},
+			expectStatus:   "ok",
+			expectMsgPart:  "All systems operational",
+			expectDown:     0,
+			expectAlerts:   0,
+			expectTotal:    3,
+			expectUp:       1,
+			expectPaused:   1,
+			expectPending:  1,
+			expectDownSumm: 0,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			app := newTestHub(t)
+			user := createTestUser(t, app)
+			tt.setup(t, app, user)
+
+			type requestCapture struct {
+				method      string
+				userAgent   string
+				contentType string
+				payload     heartbeat.Payload
+			}
+
+			captured := make(chan requestCapture, 1)
+			server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				defer r.Body.Close()
+				body, err := io.ReadAll(r.Body)
+				require.NoError(t, err)
+
+				var payload heartbeat.Payload
+				require.NoError(t, json.Unmarshal(body, &payload))
+				captured <- requestCapture{
+					method:      r.Method,
+					userAgent:   r.Header.Get("User-Agent"),
+					contentType: r.Header.Get("Content-Type"),
+					payload:     payload,
+				}
+				w.WriteHeader(http.StatusNoContent)
+			}))
+			defer server.Close()
+
+			hb := heartbeat.New(app.App, envGetter(map[string]string{
+				"HEARTBEAT_URL":    server.URL,
+				"HEARTBEAT_METHOD": "POST",
+			}))
+			require.NotNil(t, hb)
+			require.NoError(t, hb.Send())
+
+			req := <-captured
+			assert.Equal(t, http.MethodPost, req.method)
+			assert.Equal(t, "Beszel-Heartbeat", req.userAgent)
+			assert.Equal(t, "application/json", req.contentType)
+
+			assert.Equal(t, tt.expectStatus, req.payload.Status)
+			assert.Contains(t, req.payload.Msg, tt.expectMsgPart)
+			assert.Equal(t, tt.expectDown, len(req.payload.Down))
+			assert.Equal(t, tt.expectAlerts, len(req.payload.Alerts))
+			assert.Equal(t, tt.expectTotal, req.payload.Systems.Total)
+			assert.Equal(t, tt.expectUp, req.payload.Systems.Up)
+			assert.Equal(t, tt.expectDownSumm, req.payload.Systems.Down)
+			assert.Equal(t, tt.expectPaused, req.payload.Systems.Paused)
+			assert.Equal(t, tt.expectPending, req.payload.Systems.Pending)
+		})
+	}
+}
+
+func newTestHub(t *testing.T) *beszeltests.TestHub {
+	t.Helper()
+	app, err := beszeltests.NewTestHub(t.TempDir())
+	require.NoError(t, err)
+	t.Cleanup(app.Cleanup)
+	return app
+}
+
+func createTestUser(t *testing.T, app *beszeltests.TestHub) *core.Record {
+	t.Helper()
+	user, err := beszeltests.CreateUser(app.App, "admin@example.com", "password123")
+	require.NoError(t, err)
+	return user
+}
+
+func createTestSystem(t *testing.T, app *beszeltests.TestHub, userID, name, host, status string) *core.Record {
+	t.Helper()
+	system, err := beszeltests.CreateRecord(app.App, "systems", map[string]any{
+		"name":   name,
+		"host":   host,
+		"port":   "45876",
+		"users":  []string{userID},
+		"status": status,
+	})
+	require.NoError(t, err)
+	return system
+}
+
+func createTriggeredAlert(t *testing.T, app *beszeltests.TestHub, userID, systemID, name string, threshold float64) *core.Record {
+	t.Helper()
+	alert, err := beszeltests.CreateRecord(app.App, "alerts", map[string]any{
+		"name":      name,
+		"system":    systemID,
+		"user":      userID,
+		"value":     threshold,
+		"min":       0,
+		"triggered": true,
+	})
+	require.NoError(t, err)
+	return alert
+}
+
+func envGetter(values map[string]string) func(string) (string, bool) {
+	return func(key string) (string, bool) {
+		v, ok := values[key]
+		return v, ok
+	}
+}

internal/hub/hub.go

@@ -9,12 +9,14 @@ import (
 	"net/url"
 	"os"
 	"path"
+	"regexp"
 	"strings"
 	"time"
 
 	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/internal/alerts"
 	"github.com/henrygd/beszel/internal/hub/config"
+	"github.com/henrygd/beszel/internal/hub/heartbeat"
 	"github.com/henrygd/beszel/internal/hub/systems"
 	"github.com/henrygd/beszel/internal/records"
 	"github.com/henrygd/beszel/internal/users"
@@ -33,11 +35,15 @@ type Hub struct {
 	um     *users.UserManager
 	rm     *records.RecordManager
 	sm     *systems.SystemManager
+	hb     *heartbeat.Heartbeat
+	hbStop chan struct{}
 	pubKey string
 	signer ssh.Signer
 	appURL string
 }
 
+var containerIDPattern = regexp.MustCompile(`^[a-fA-F0-9]{12,64}$`)
+
 // NewHub creates a new Hub instance with default configuration
 func NewHub(app core.App) *Hub {
 	hub := &Hub{}
@@ -48,6 +54,10 @@ func NewHub(app core.App) *Hub {
 	hub.rm = records.NewRecordManager(hub)
 	hub.sm = systems.NewSystemManager(hub)
 	hub.appURL, _ = GetEnv("APP_URL")
+	hub.hb = heartbeat.New(app, GetEnv)
+	if hub.hb != nil {
+		hub.hbStop = make(chan struct{})
+	}
 	return hub
 }
 
@@ -88,6 +98,10 @@ func (h *Hub) StartHub() error {
 		if err := h.sm.Initialize(); err != nil {
 			return err
 		}
+		// start heartbeat if configured
+		if h.hb != nil {
+			go h.hb.Start(h.hbStop)
+		}
 		return e.Next()
 	})
 
@@ -287,6 +301,9 @@ func (h *Hub) registerApiRoutes(se *core.ServeEvent) error {
 	})
 	// send test notification
 	apiAuth.POST("/test-notification", h.SendTestNotification)
+	// heartbeat status and test
+	apiAuth.GET("/heartbeat-status", h.getHeartbeatStatus)
+	apiAuth.POST("/test-heartbeat", h.testHeartbeat)
 	// get config.yml content
 	apiAuth.GET("/config-yaml", config.GetYamlConfig)
 	// handle agent websocket connection
@@ -403,6 +420,42 @@ func (h *Hub) getUniversalToken(e *core.RequestEvent) error {
 	return e.JSON(http.StatusOK, response)
 }
 
+// getHeartbeatStatus returns current heartbeat configuration and whether it's enabled
+func (h *Hub) getHeartbeatStatus(e *core.RequestEvent) error {
+	if e.Auth.GetString("role") != "admin" {
+		return e.ForbiddenError("Requires admin role", nil)
+	}
+	if h.hb == nil {
+		return e.JSON(http.StatusOK, map[string]any{
+			"enabled": false,
+			"msg":     "Set HEARTBEAT_URL to enable outbound heartbeat monitoring",
+		})
+	}
+	cfg := h.hb.GetConfig()
+	return e.JSON(http.StatusOK, map[string]any{
+		"enabled":  true,
+		"url":      cfg.URL,
+		"interval": cfg.Interval,
+		"method":   cfg.Method,
+	})
+}
+
+// testHeartbeat triggers a single heartbeat ping and returns the result
+func (h *Hub) testHeartbeat(e *core.RequestEvent) error {
+	if e.Auth.GetString("role") != "admin" {
+		return e.ForbiddenError("Requires admin role", nil)
+	}
+	if h.hb == nil {
+		return e.JSON(http.StatusOK, map[string]any{
+			"err": "Heartbeat not configured. Set HEARTBEAT_URL environment variable.",
+		})
+	}
+	if err := h.hb.Send(); err != nil {
+		return e.JSON(http.StatusOK, map[string]any{"err": err.Error()})
+	}
+	return e.JSON(http.StatusOK, map[string]any{"err": false})
+}
+
 // containerRequestHandler handles both container logs and info requests
 func (h *Hub) containerRequestHandler(e *core.RequestEvent, fetchFunc func(*systems.System, string) (string, error), responseKey string) error {
 	systemID := e.Request.URL.Query().Get("system")
@@ -411,6 +464,9 @@ func (h *Hub) containerRequestHandler(e *core.RequestEvent, fetchFunc func(*syst
 	if systemID == "" || containerID == "" {
 		return e.JSON(http.StatusBadRequest, map[string]string{"error": "system and container parameters are required"})
 	}
+	if !containerIDPattern.MatchString(containerID) {
+		return e.JSON(http.StatusBadRequest, map[string]string{"error": "invalid container parameter"})
+	}
 
 	system, err := h.sm.GetSystem(systemID)
 	if err != nil {

internal/hub/hub_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package hub_test
 
@@ -362,6 +361,58 @@ func TestApiRoutesAuthentication(t *testing.T) {
 			ExpectedContent: []string{"test-system"},
 			TestAppFactory:  testAppFactory,
 		},
+		{
+			Name:            "GET /heartbeat-status - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/heartbeat-status",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /heartbeat-status - with user auth should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/heartbeat-status",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Requires admin role"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /heartbeat-status - with admin auth should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/heartbeat-status",
+			Headers: map[string]string{
+				"Authorization": adminUserToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{`"enabled":false`},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /test-heartbeat - with user auth should fail",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/test-heartbeat",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Requires admin role"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /test-heartbeat - with admin auth should report disabled state",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/test-heartbeat",
+			Headers: map[string]string{
+				"Authorization": adminUserToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"Heartbeat not configured"},
+			TestAppFactory:  testAppFactory,
+		},
 		{
 			Name:            "GET /universal-token - no auth should fail",
 			Method:          http.MethodGet,
@@ -493,7 +544,7 @@ func TestApiRoutesAuthentication(t *testing.T) {
 		{
 			Name:   "GET /containers/logs - with auth but invalid system should fail",
 			Method: http.MethodGet,
-			URL:    "/api/beszel/containers/logs?system=invalid-system&container=test-container",
+			URL:    "/api/beszel/containers/logs?system=invalid-system&container=0123456789ab",
 			Headers: map[string]string{
 				"Authorization": userToken,
 			},
@@ -501,6 +552,39 @@ func TestApiRoutesAuthentication(t *testing.T) {
 			ExpectedContent: []string{"system not found"},
 			TestAppFactory:  testAppFactory,
 		},
+		{
+			Name:   "GET /containers/logs - traversal container should fail validation",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/logs?system=" + system.Id + "&container=..%2F..%2Fversion",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"invalid container parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/info - traversal container should fail validation",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/info?system=" + system.Id + "&container=../../version?x=",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"invalid container parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/info - non-hex container should fail validation",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/info?system=" + system.Id + "&container=container_name",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"invalid container parameter"},
+			TestAppFactory:  testAppFactory,
+		},
 
 		// Auth Optional Routes - Should work without authentication
 		{

internal/hub/hub_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package hub
 

internal/hub/systems/system.go

@@ -48,7 +48,6 @@ type System struct {
 	detailsFetched atomic.Bool             // True if static system details have been fetched and saved
 	smartFetching  atomic.Bool             // True if SMART devices are currently being fetched
 	smartInterval  time.Duration           // Interval for periodic SMART data updates
-	lastSmartFetch atomic.Int64            // Unix milliseconds of last SMART data fetch
 }
 
 func (sm *SystemManager) NewSystem(systemId string) *System {
@@ -134,19 +133,34 @@ func (sys *System) update() error {
 		return err
 	}
 
+	// ensure deprecated fields from older agents are migrated to current fields
+	migrateDeprecatedFields(data, !sys.detailsFetched.Load())
+
 	// create system records
 	_, err = sys.createRecords(data)
 
+	// if details were included and fetched successfully, mark details as fetched and update smart interval if set by agent
+	if err == nil && data.Details != nil {
+		sys.detailsFetched.Store(true)
+		// update smart interval if it's set on the agent side
+		if data.Details.SmartInterval > 0 {
+			sys.smartInterval = data.Details.SmartInterval
+			// make sure we reset expiration of lastFetch to remain as long as the new smart interval
+			// to prevent premature expiration leading to new fetch if interval is different.
+			sys.manager.smartFetchMap.UpdateExpiration(sys.Id, sys.smartInterval+time.Minute)
+		}
+	}
+
 	// Fetch and save SMART devices when system first comes online or at intervals
-	if backgroundSmartFetchEnabled() {
+	if backgroundSmartFetchEnabled() && sys.detailsFetched.Load() {
 		if sys.smartInterval <= 0 {
 			sys.smartInterval = time.Hour
 		}
-		lastFetch := sys.lastSmartFetch.Load()
-		if time.Since(time.UnixMilli(lastFetch)) >= sys.smartInterval && sys.smartFetching.CompareAndSwap(false, true) {
+		lastFetch, _ := sys.manager.smartFetchMap.GetOk(sys.Id)
+		if time.Since(time.UnixMilli(lastFetch-1e4)) >= sys.smartInterval && sys.smartFetching.CompareAndSwap(false, true) {
 			go func() {
 				defer sys.smartFetching.Store(false)
-				sys.lastSmartFetch.Store(time.Now().UnixMilli())
+				sys.manager.smartFetchMap.Set(sys.Id, time.Now().UnixMilli(), sys.smartInterval+time.Minute)
 				_ = sys.FetchAndSaveSmartDevices()
 			}()
 		}
@@ -221,11 +235,6 @@ func (sys *System) createRecords(data *system.CombinedData) (*core.Record, error
 			if err := createSystemDetailsRecord(txApp, data.Details, sys.Id); err != nil {
 				return err
 			}
-			sys.detailsFetched.Store(true)
-			// update smart interval if it's set on the agent side
-			if data.Details.SmartInterval > 0 {
-				sys.smartInterval = data.Details.SmartInterval
-			}
 		}
 
 		// update system record (do this last because it triggers alerts and we need above records to be inserted first)
@@ -703,3 +712,50 @@ func getJitter() <-chan time.Time {
 	msDelay := (interval * minPercent / 100) + rand.Intn(interval*jitterRange/100)
 	return time.After(time.Duration(msDelay) * time.Millisecond)
 }
+
+// migrateDeprecatedFields moves values from deprecated fields to their new locations if the new
+// fields are not already populated. Deprecated fields and refs may be removed at least 30 days
+// and one minor version release after the release that includes the migration.
+//
+// This is run when processing incoming system data from agents, which may be on older versions.
+func migrateDeprecatedFields(cd *system.CombinedData, createDetails bool) {
+	// migration added 0.19.0
+	if cd.Stats.Bandwidth[0] == 0 && cd.Stats.Bandwidth[1] == 0 {
+		cd.Stats.Bandwidth[0] = uint64(cd.Stats.NetworkSent * 1024 * 1024)
+		cd.Stats.Bandwidth[1] = uint64(cd.Stats.NetworkRecv * 1024 * 1024)
+		cd.Stats.NetworkSent, cd.Stats.NetworkRecv = 0, 0
+	}
+	// migration added 0.19.0
+	if cd.Info.BandwidthBytes == 0 {
+		cd.Info.BandwidthBytes = uint64(cd.Info.Bandwidth * 1024 * 1024)
+		cd.Info.Bandwidth = 0
+	}
+	// migration added 0.19.0
+	if cd.Stats.DiskIO[0] == 0 && cd.Stats.DiskIO[1] == 0 {
+		cd.Stats.DiskIO[0] = uint64(cd.Stats.DiskReadPs * 1024 * 1024)
+		cd.Stats.DiskIO[1] = uint64(cd.Stats.DiskWritePs * 1024 * 1024)
+		cd.Stats.DiskReadPs, cd.Stats.DiskWritePs = 0, 0
+	}
+	// migration added 0.19.0 - Move deprecated Info fields to Details struct
+	if cd.Details == nil && cd.Info.Hostname != "" {
+		if createDetails {
+			cd.Details = &system.Details{
+				Hostname:    cd.Info.Hostname,
+				Kernel:      cd.Info.KernelVersion,
+				Cores:       cd.Info.Cores,
+				Threads:     cd.Info.Threads,
+				CpuModel:    cd.Info.CpuModel,
+				Podman:      cd.Info.Podman,
+				Os:          cd.Info.Os,
+				MemoryTotal: uint64(cd.Stats.Mem * 1024 * 1024 * 1024),
+			}
+		}
+		// zero the deprecated fields to prevent saving them in systems.info DB json payload
+		cd.Info.Hostname = ""
+		cd.Info.KernelVersion = ""
+		cd.Info.Cores = 0
+		cd.Info.CpuModel = ""
+		cd.Info.Podman = false
+		cd.Info.Os = 0
+	}
+}

internal/hub/systems/system_manager.go

@@ -8,6 +8,7 @@ import (
 	"github.com/henrygd/beszel/internal/hub/ws"
 
 	"github.com/henrygd/beszel/internal/entities/system"
+	"github.com/henrygd/beszel/internal/hub/expirymap"
 
 	"github.com/henrygd/beszel/internal/common"
 
@@ -40,9 +41,10 @@ var errSystemExists = errors.New("system exists")
 // SystemManager manages a collection of monitored systems and their connections.
 // It handles system lifecycle, status updates, and maintains both SSH and WebSocket connections.
 type SystemManager struct {
-	hub       hubLike                       // Hub interface for database and alert operations
-	systems   *store.Store[string, *System] // Thread-safe store of active systems
-	sshConfig *ssh.ClientConfig             // SSH client configuration for system connections
+	hub           hubLike                       // Hub interface for database and alert operations
+	systems       *store.Store[string, *System] // Thread-safe store of active systems
+	sshConfig     *ssh.ClientConfig             // SSH client configuration for system connections
+	smartFetchMap *expirymap.ExpiryMap[int64]   // Stores last SMART fetch time per system ID
 }
 
 // hubLike defines the interface requirements for the hub dependency.
@@ -58,8 +60,9 @@ type hubLike interface {
 // The hub must implement the hubLike interface to provide database and alert functionality.
 func NewSystemManager(hub hubLike) *SystemManager {
 	return &SystemManager{
-		systems: store.New(map[string]*System{}),
-		hub:     hub,
+		systems:       store.New(map[string]*System{}),
+		hub:           hub,
+		smartFetchMap: expirymap.New[int64](time.Hour),
 	}
 }
 

internal/hub/systems/system_test.go

@@ -0,0 +1,159 @@
+//go:build testing
+
+package systems
+
+import (
+	"testing"
+
+	"github.com/henrygd/beszel/internal/entities/system"
+)
+
+func TestCombinedData_MigrateDeprecatedFields(t *testing.T) {
+	t.Run("Migrate NetworkSent and NetworkRecv to Bandwidth", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Stats: system.Stats{
+				NetworkSent: 1.5, // 1.5 MB
+				NetworkRecv: 2.5, // 2.5 MB
+			},
+		}
+		migrateDeprecatedFields(cd, true)
+
+		expectedSent := uint64(1.5 * 1024 * 1024)
+		expectedRecv := uint64(2.5 * 1024 * 1024)
+
+		if cd.Stats.Bandwidth[0] != expectedSent {
+			t.Errorf("expected Bandwidth[0] %d, got %d", expectedSent, cd.Stats.Bandwidth[0])
+		}
+		if cd.Stats.Bandwidth[1] != expectedRecv {
+			t.Errorf("expected Bandwidth[1] %d, got %d", expectedRecv, cd.Stats.Bandwidth[1])
+		}
+		if cd.Stats.NetworkSent != 0 || cd.Stats.NetworkRecv != 0 {
+			t.Errorf("expected NetworkSent and NetworkRecv to be reset, got %f, %f", cd.Stats.NetworkSent, cd.Stats.NetworkRecv)
+		}
+	})
+
+	t.Run("Migrate Info.Bandwidth to Info.BandwidthBytes", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Info: system.Info{
+				Bandwidth: 10.0, // 10 MB
+			},
+		}
+		migrateDeprecatedFields(cd, true)
+
+		expected := uint64(10 * 1024 * 1024)
+		if cd.Info.BandwidthBytes != expected {
+			t.Errorf("expected BandwidthBytes %d, got %d", expected, cd.Info.BandwidthBytes)
+		}
+		if cd.Info.Bandwidth != 0 {
+			t.Errorf("expected Info.Bandwidth to be reset, got %f", cd.Info.Bandwidth)
+		}
+	})
+
+	t.Run("Migrate DiskReadPs and DiskWritePs to DiskIO", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Stats: system.Stats{
+				DiskReadPs:  3.0, // 3 MB
+				DiskWritePs: 4.0, // 4 MB
+			},
+		}
+		migrateDeprecatedFields(cd, true)
+
+		expectedRead := uint64(3 * 1024 * 1024)
+		expectedWrite := uint64(4 * 1024 * 1024)
+
+		if cd.Stats.DiskIO[0] != expectedRead {
+			t.Errorf("expected DiskIO[0] %d, got %d", expectedRead, cd.Stats.DiskIO[0])
+		}
+		if cd.Stats.DiskIO[1] != expectedWrite {
+			t.Errorf("expected DiskIO[1] %d, got %d", expectedWrite, cd.Stats.DiskIO[1])
+		}
+		if cd.Stats.DiskReadPs != 0 || cd.Stats.DiskWritePs != 0 {
+			t.Errorf("expected DiskReadPs and DiskWritePs to be reset, got %f, %f", cd.Stats.DiskReadPs, cd.Stats.DiskWritePs)
+		}
+	})
+
+	t.Run("Migrate Info fields to Details struct", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Stats: system.Stats{
+				Mem: 16.0, // 16 GB
+			},
+			Info: system.Info{
+				Hostname:      "test-host",
+				KernelVersion: "6.8.0",
+				Cores:         8,
+				Threads:       16,
+				CpuModel:      "Intel i7",
+				Podman:        true,
+				Os:            system.Linux,
+			},
+		}
+		migrateDeprecatedFields(cd, true)
+
+		if cd.Details == nil {
+			t.Fatal("expected Details struct to be created")
+		}
+		if cd.Details.Hostname != "test-host" {
+			t.Errorf("expected Hostname 'test-host', got '%s'", cd.Details.Hostname)
+		}
+		if cd.Details.Kernel != "6.8.0" {
+			t.Errorf("expected Kernel '6.8.0', got '%s'", cd.Details.Kernel)
+		}
+		if cd.Details.Cores != 8 {
+			t.Errorf("expected Cores 8, got %d", cd.Details.Cores)
+		}
+		if cd.Details.Threads != 16 {
+			t.Errorf("expected Threads 16, got %d", cd.Details.Threads)
+		}
+		if cd.Details.CpuModel != "Intel i7" {
+			t.Errorf("expected CpuModel 'Intel i7', got '%s'", cd.Details.CpuModel)
+		}
+		if cd.Details.Podman != true {
+			t.Errorf("expected Podman true, got %v", cd.Details.Podman)
+		}
+		if cd.Details.Os != system.Linux {
+			t.Errorf("expected Os Linux, got %d", cd.Details.Os)
+		}
+		expectedMem := uint64(16 * 1024 * 1024 * 1024)
+		if cd.Details.MemoryTotal != expectedMem {
+			t.Errorf("expected MemoryTotal %d, got %d", expectedMem, cd.Details.MemoryTotal)
+		}
+
+		if cd.Info.Hostname != "" || cd.Info.KernelVersion != "" || cd.Info.Cores != 0 || cd.Info.CpuModel != "" || cd.Info.Podman != false || cd.Info.Os != 0 {
+			t.Errorf("expected Info fields to be reset, got %+v", cd.Info)
+		}
+	})
+
+	t.Run("Do not migrate if Details already exists", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Details: &system.Details{Hostname: "existing-host"},
+			Info: system.Info{
+				Hostname: "deprecated-host",
+			},
+		}
+		migrateDeprecatedFields(cd, true)
+
+		if cd.Details.Hostname != "existing-host" {
+			t.Errorf("expected Hostname 'existing-host', got '%s'", cd.Details.Hostname)
+		}
+		if cd.Info.Hostname != "deprecated-host" {
+			t.Errorf("expected Info.Hostname to remain 'deprecated-host', got '%s'", cd.Info.Hostname)
+		}
+	})
+
+	t.Run("Do not create details if migrateDetails is false", func(t *testing.T) {
+		cd := &system.CombinedData{
+			Info: system.Info{
+				Hostname: "deprecated-host",
+			},
+		}
+		migrateDeprecatedFields(cd, false)
+
+		if cd.Details != nil {
+			t.Fatal("expected Details struct to not be created")
+		}
+
+		if cd.Info.Hostname != "" {
+			t.Errorf("expected Info.Hostname to be reset, got '%s'", cd.Info.Hostname)
+		}
+	})
+}

internal/hub/systems/systems_production.go

@@ -1,5 +1,4 @@
 //go:build !testing
-// +build !testing
 
 package systems
 

internal/hub/systems/systems_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package systems_test
 

internal/hub/systems/systems_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package systems
 
@@ -114,4 +113,5 @@ func (sm *SystemManager) RemoveAllSystems() {
 	for _, system := range sm.systems.GetAll() {
 		sm.RemoveSystem(system.Id)
 	}
+	sm.smartFetchMap.StopCleaner()
 }

internal/hub/ws/request_manager_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package ws
 

internal/hub/ws/ws_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package ws
 

internal/hub/ws/ws_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package ws
 

internal/records/records_test.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package records_test
 

internal/records/records_test_helpers.go

@@ -1,5 +1,4 @@
 //go:build testing
-// +build testing
 
 package records
 

internal/site/biome.json

@@ -1,5 +1,5 @@
 {
-	"$schema": "https://biomejs.dev/schemas/2.2.3/schema.json",
+	"$schema": "https://biomejs.dev/schemas/2.2.4/schema.json",
 	"vcs": {
 		"enabled": true,
 		"clientKind": "git",
@@ -12,7 +12,7 @@
 		"lineWidth": 120,
 		"formatWithErrors": true
 	},
-	"assist": { "actions": { "source": { "organizeImports": "on" } } },
+	"assist": { "actions": { "source": { "organizeImports": "off" } } },
 	"linter": {
 		"enabled": true,
 		"rules": {

internal/site/package.json

@@ -1,13 +1,14 @@
 {
 	"name": "beszel",
 	"private": true,
-	"version": "0.18.3",
+	"version": "0.18.4",
 	"type": "module",
 	"scripts": {
 		"dev": "vite --host",
 		"build": "lingui extract --overwrite && lingui compile && vite build",
 		"preview": "vite preview",
 		"sync": "lingui extract --overwrite && lingui compile",
+		"sync_no_compile": "lingui extract --overwrite --clean",
 		"sync_and_purge": "lingui extract --overwrite --clean && lingui compile",
 		"format": "biome format --write .",
 		"lint": "biome lint .",

internal/site/src/components/alerts/alert-button.tsx

@@ -26,7 +26,7 @@ export default memo(function AlertsButton({ system }: { system: SystemRecord })
 						/>
 					</Button>
 				</SheetTrigger>
-				<SheetContent className="max-h-full overflow-auto w-150 !max-w-full p-4 sm:p-6">
+				<SheetContent className="max-h-full overflow-auto w-160 !max-w-full p-4 sm:p-6">
 					{opened && <AlertDialogContent system={system} />}
 				</SheetContent>
 			</Sheet>

internal/site/src/components/alerts/alerts-sheet.tsx

@@ -7,6 +7,7 @@ import { lazy, memo, Suspense, useMemo, useState } from "react"
 import { $router, Link } from "@/components/router"
 import { Checkbox } from "@/components/ui/checkbox"
 import { DialogDescription, DialogHeader, DialogTitle } from "@/components/ui/dialog"
+import { Input } from "@/components/ui/input"
 import { Switch } from "@/components/ui/switch"
 import { Tabs, TabsContent, TabsList, TabsTrigger } from "@/components/ui/tabs"
 import { toast } from "@/components/ui/use-toast"
@@ -20,7 +21,7 @@ const Slider = lazy(() => import("@/components/ui/slider"))
 
 const endpoint = "/api/beszel/user-alerts"
 
-const alertDebounce = 100
+const alertDebounce = 400
 
 const alertKeys = Object.keys(alertInfo) as (keyof typeof alertInfo)[]
 
@@ -244,7 +245,7 @@ export function AlertContent({
 					<Suspense fallback={<div className="h-10" />}>
 						{!singleDescription && (
 							<div>
-								<p id={`v${name}`} className="text-sm block h-8">
+								<p id={`v${name}`} className="text-sm block h-6">
 									{alertData.invert ? (
 										<Trans>
 											Average drops below{" "}
@@ -263,21 +264,38 @@ export function AlertContent({
 										</Trans>
 									)}
 								</p>
-								<div className="flex gap-3">
+								<div className="flex gap-3 items-center">
 									<Slider
 										aria-labelledby={`v${name}`}
-										defaultValue={[value]}
+										value={[value]}
 										onValueCommit={(val) => sendUpsert(min, val[0])}
 										onValueChange={(val) => setValue(val[0])}
 										step={alertData.step ?? 1}
 										min={alertData.min ?? 1}
 										max={alertData.max ?? 99}
 									/>
+									<Input
+										type="number"
+										value={value}
+										onChange={(e) => {
+											let val = parseFloat(e.target.value)
+											if (!Number.isNaN(val)) {
+												if (alertData.max != null) val = Math.min(val, alertData.max)
+												if (alertData.min != null) val = Math.max(val, alertData.min)
+												setValue(val)
+												sendUpsert(min, val)
+											}
+										}}
+										step={alertData.step ?? 1}
+										min={alertData.min ?? 1}
+										max={alertData.max ?? 99}
+										className="w-16 h-8 text-center px-1"
+									/>
 								</div>
 							</div>
 						)}
 						<div className={cn(singleDescription && "col-span-full lowercase")}>
-							<p id={`t${name}`} className="text-sm block h-8 first-letter:uppercase">
+							<p id={`t${name}`} className="text-sm block h-6 first-letter:uppercase">
 								{singleDescription && (
 									<>
 										{singleDescription}
@@ -289,15 +307,30 @@ export function AlertContent({
 									<Plural value={min} one="minute" other="minutes" />
 								</Trans>
 							</p>
-							<div className="flex gap-3">
+							<div className="flex gap-3 items-center">
 								<Slider
-									aria-labelledby={`v${name}`}
-									defaultValue={[min]}
-									onValueCommit={(minVal) => sendUpsert(minVal[0], value)}
+									aria-labelledby={`t${name}`}
+									value={[min]}
+									onValueCommit={(val) => sendUpsert(val[0], value)}
 									onValueChange={(val) => setMin(val[0])}
 									min={1}
 									max={60}
 								/>
+								<Input
+									type="number"
+									value={min}
+									onChange={(e) => {
+										let val = parseInt(e.target.value, 10)
+										if (!Number.isNaN(val)) {
+											val = Math.max(1, Math.min(val, 60))
+											setMin(val)
+											sendUpsert(val, value)
+										}
+									}}
+									min={1}
+									max={60}
+									className="w-16 h-8 text-center px-1"
+								/>
 							</div>
 						</div>
 					</Suspense>

internal/site/src/components/charts/load-average-chart.tsx

@@ -16,19 +16,16 @@ import { useYAxisWidth } from "./hooks"
 export default memo(function LoadAverageChart({ chartData }: { chartData: ChartData }) {
 	const { yAxisWidth, updateYAxisWidth } = useYAxisWidth()
 
-	const keys: { legacy: keyof SystemStats; color: string; label: string }[] = [
+	const keys: { color: string; label: string }[] = [
 		{
-			legacy: "l1",
 			color: "hsl(271, 81%, 60%)", // Purple
 			label: t({ message: `1 min`, comment: "Load average" }),
 		},
 		{
-			legacy: "l5",
 			color: "hsl(217, 91%, 60%)", // Blue
 			label: t({ message: `5 min`, comment: "Load average" }),
 		},
 		{
-			legacy: "l15",
 			color: "hsl(25, 95%, 53%)", // Orange
 			label: t({ message: `15 min`, comment: "Load average" }),
 		},
@@ -66,27 +63,18 @@ export default memo(function LoadAverageChart({ chartData }: { chartData: ChartD
 							/>
 						}
 					/>
-					{keys.map(({ legacy, color, label }, i) => {
-						const dataKey = (value: { stats: SystemStats }) => {
-							const { minor, patch } = chartData.agentVersion
-							if (minor <= 12 && patch < 1) {
-								return value.stats?.[legacy]
-							}
-							return value.stats?.la?.[i] ?? value.stats?.[legacy]
-						}
-						return (
-							<Line
-								key={label}
-								dataKey={dataKey}
-								name={label}
-								type="monotoneX"
-								dot={false}
-								strokeWidth={1.5}
-								stroke={color}
-								isAnimationActive={false}
-							/>
-						)
-					})}
+					{keys.map(({ color, label }, i) => (
+						<Line
+							key={label}
+							dataKey={(value: { stats: SystemStats }) => value.stats?.la?.[i]}
+							name={label}
+							type="monotoneX"
+							dot={false}
+							strokeWidth={1.5}
+							stroke={color}
+							isAnimationActive={false}
+						/>
+					))}
 					<ChartLegend content={<ChartLegendContent />} />
 				</LineChart>
 			</ChartContainer>

internal/site/src/components/install-dropdowns.tsx

@@ -43,7 +43,7 @@ export function copyDockerCompose(port = "45876", publicKey: string, token: stri
 
 export function copyDockerRun(port = "45876", publicKey: string, token: string) {
 	copyToClipboard(
-		`docker run -d --name beszel-agent --network host --restart unless-stopped -v /var/run/docker.sock:/var/run/docker.sock:ro -v ./beszel_agent_data:/var/lib/beszel-agent -e KEY="${publicKey}" -e LISTEN=${port} -e TOKEN="${token}" -e HUB_URL="${getHubURL()}" henrygd/beszel-agent`
+		`docker run -d --name beszel-agent --network host --restart unless-stopped -v /var/run/docker.sock:/var/run/docker.sock:ro -v beszel_agent_data:/var/lib/beszel-agent -e KEY="${publicKey}" -e LISTEN=${port} -e TOKEN="${token}" -e HUB_URL="${getHubURL()}" henrygd/beszel-agent`
 	)
 }
 

internal/site/src/components/lang-toggle.tsx

@@ -32,7 +32,10 @@ export function LangToggle() {
 						className={cn("px-2.5 flex gap-2.5 cursor-pointer", lang === i18n.locale && "bg-accent/70 font-medium")}
 						onClick={() => dynamicActivate(lang)}
 					>
-						<span>{e}</span> {label}
+						<span>
+							{e || <code className="font-mono bg-muted text-[.65em] w-5 h-4 grid place-items-center">{lang}</code>}
+						</span>{" "}
+						{label}
 					</DropdownMenuItem>
 				))}
 			</DropdownMenuContent>

internal/site/src/components/routes/settings/general.tsx

@@ -70,7 +70,16 @@ export default function SettingsProfilePage({ userSettings }: { userSettings: Us
 						<SelectContent>
 							{languages.map(([lang, label, e]) => (
 								<SelectItem key={lang} value={lang}>
-									<span className="me-2.5">{e}</span>
+									<span className="me-2.5">
+										{e || (
+											<code
+												aria-hidden="true"
+												className="font-mono bg-muted text-[.65em] w-5 h-4 inline-grid place-items-center"
+											>
+												{lang}
+											</code>
+										)}
+									</span>
 									{label}
 								</SelectItem>
 							))}

internal/site/src/components/routes/settings/heartbeat.tsx

@@ -0,0 +1,219 @@
+import { t } from "@lingui/core/macro"
+import { Trans } from "@lingui/react/macro"
+import { redirectPage } from "@nanostores/router"
+import { LoaderCircleIcon, SendIcon } from "lucide-react"
+import { useEffect, useState } from "react"
+import { $router } from "@/components/router"
+import { Badge } from "@/components/ui/badge"
+import { Button } from "@/components/ui/button"
+import { Separator } from "@/components/ui/separator"
+import { toast } from "@/components/ui/use-toast"
+import { isAdmin, pb } from "@/lib/api"
+import { cn } from "@/lib/utils"
+
+interface HeartbeatStatus {
+	enabled: boolean
+	url?: string
+	interval?: number
+	method?: string
+	msg?: string
+}
+
+export default function HeartbeatSettings() {
+	const [status, setStatus] = useState<HeartbeatStatus | null>(null)
+	const [isLoading, setIsLoading] = useState(true)
+	const [isTesting, setIsTesting] = useState(false)
+
+	if (!isAdmin()) {
+		redirectPage($router, "settings", { name: "general" })
+	}
+
+	useEffect(() => {
+		fetchStatus()
+	}, [])
+
+	async function fetchStatus() {
+		try {
+			setIsLoading(true)
+			const res = await pb.send<HeartbeatStatus>("/api/beszel/heartbeat-status", {})
+			setStatus(res)
+		} catch (error: unknown) {
+			toast({
+				title: t`Error`,
+				description: (error as Error).message,
+				variant: "destructive",
+			})
+		} finally {
+			setIsLoading(false)
+		}
+	}
+
+	async function sendTestHeartbeat() {
+		setIsTesting(true)
+		try {
+			const res = await pb.send<{ err: string | false }>("/api/beszel/test-heartbeat", {
+				method: "POST",
+			})
+			if ("err" in res && !res.err) {
+				toast({
+					title: t`Heartbeat sent successfully`,
+					description: t`Check your monitoring service`,
+				})
+			} else {
+				toast({
+					title: t`Error`,
+					description: (res.err as string) ?? t`Failed to send heartbeat`,
+					variant: "destructive",
+				})
+			}
+		} catch (error: unknown) {
+			toast({
+				title: t`Error`,
+				description: (error as Error).message,
+				variant: "destructive",
+			})
+		} finally {
+			setIsTesting(false)
+		}
+	}
+
+	return (
+		<div>
+			<div>
+				<h3 className="text-xl font-medium mb-2">
+					<Trans>Heartbeat Monitoring</Trans>
+				</h3>
+				<p className="text-sm text-muted-foreground leading-relaxed">
+					<Trans>
+						Send periodic outbound pings to an external monitoring service so you can monitor Beszel without exposing it
+						to the internet.
+					</Trans>
+				</p>
+			</div>
+			<Separator className="my-4" />
+
+			{status?.enabled ? (
+				<EnabledState status={status} isTesting={isTesting} sendTestHeartbeat={sendTestHeartbeat} />
+			) : (
+				<NotEnabledState isLoading={isLoading} />
+			)}
+		</div>
+	)
+}
+
+function EnabledState({
+	status,
+	isTesting,
+	sendTestHeartbeat,
+}: {
+	status: HeartbeatStatus
+	isTesting: boolean
+	sendTestHeartbeat: () => void
+}) {
+	const TestIcon = isTesting ? LoaderCircleIcon : SendIcon
+	return (
+		<div className="space-y-5">
+			<div className="flex items-center gap-2">
+				<Badge variant="success">
+					<Trans>Active</Trans>
+				</Badge>
+			</div>
+			<div className="grid gap-4 sm:grid-cols-2">
+				<ConfigItem label={t`Endpoint URL`} value={status.url ?? ""} mono />
+				<ConfigItem label={t`Interval`} value={`${status.interval}s`} />
+				<ConfigItem label={t`HTTP Method`} value={status.method ?? "POST"} />
+			</div>
+
+			<Separator />
+
+			<div>
+				<h4 className="text-base font-medium mb-1">
+					<Trans>Test heartbeat</Trans>
+				</h4>
+				<p className="text-sm text-muted-foreground leading-relaxed mb-3">
+					<Trans>Send a single heartbeat ping to verify your endpoint is working.</Trans>
+				</p>
+				<Button
+					type="button"
+					variant="outline"
+					className="flex items-center gap-1.5"
+					onClick={sendTestHeartbeat}
+					disabled={isTesting}
+				>
+					<TestIcon className={cn("size-4", isTesting && "animate-spin")} />
+					<Trans>Send test heartbeat</Trans>
+				</Button>
+			</div>
+
+			<Separator />
+
+			<div>
+				<h4 className="text-base font-medium mb-2">
+					<Trans>Payload format</Trans>
+				</h4>
+				<p className="text-sm text-muted-foreground leading-relaxed mb-2">
+					<Trans>
+						When using POST, each heartbeat includes a JSON payload with system status summary, list of down systems,
+						and triggered alerts.
+					</Trans>
+				</p>
+				<p className="text-sm text-muted-foreground leading-relaxed">
+					<Trans>
+						The overall status is <code className="bg-muted rounded-sm px-1 text-primary">ok</code> when all systems are
+						up, <code className="bg-muted rounded-sm px-1 text-primary">warn</code> when alerts are triggered, and{" "}
+						<code className="bg-muted rounded-sm px-1 text-primary">error</code> when any system is down.
+					</Trans>
+				</p>
+			</div>
+		</div>
+	)
+}
+
+function NotEnabledState({ isLoading }: { isLoading?: boolean }) {
+	return (
+		<div className={cn("grid gap-4", isLoading && "animate-pulse")}>
+			<div>
+				<p className="text-sm text-muted-foreground leading-relaxed mb-3">
+					<Trans>Set the following environment variables on your Beszel hub to enable heartbeat monitoring:</Trans>
+				</p>
+				<div className="grid gap-2.5">
+					<EnvVarItem
+						name="HEARTBEAT_URL"
+						description={t`Endpoint URL to ping (required)`}
+						example="https://uptime.betterstack.com/api/v1/heartbeat/xxxx"
+					/>
+					<EnvVarItem name="HEARTBEAT_INTERVAL" description={t`Seconds between pings (default: 60)`} example="60" />
+					<EnvVarItem
+						name="HEARTBEAT_METHOD"
+						description={t`HTTP method: POST, GET, or HEAD (default: POST)`}
+						example="POST"
+					/>
+				</div>
+			</div>
+			<p className="text-sm text-muted-foreground leading-relaxed">
+				<Trans>After setting the environment variables, restart your Beszel hub for changes to take effect.</Trans>
+			</p>
+		</div>
+	)
+}
+
+function ConfigItem({ label, value, mono }: { label: string; value: string; mono?: boolean }) {
+	return (
+		<div>
+			<p className="text-sm font-medium mb-0.5">{label}</p>
+			<p className={cn("text-sm text-muted-foreground break-all", mono && "font-mono")}>{value}</p>
+		</div>
+	)
+}
+
+function EnvVarItem({ name, description, example }: { name: string; description: string; example: string }) {
+	return (
+		<div className="bg-muted/50 rounded-md px-3 py-2.5 grid gap-1.5">
+			<code className="text-sm font-mono text-primary font-medium leading-tight">{name}</code>
+			<p className="text-sm text-muted-foreground">{description}</p>
+			<p className="text-xs text-muted-foreground">
+				<Trans>Example:</Trans> <code className="font-mono">{example}</code>
+			</p>
+		</div>
+	)
+}

internal/site/src/components/routes/settings/layout.tsx

@@ -2,7 +2,14 @@ import { t } from "@lingui/core/macro"
 import { Trans, useLingui } from "@lingui/react/macro"
 import { useStore } from "@nanostores/react"
 import { getPagePath, redirectPage } from "@nanostores/router"
-import { AlertOctagonIcon, BellIcon, FileSlidersIcon, FingerprintIcon, SettingsIcon } from "lucide-react"
+import {
+	AlertOctagonIcon,
+	BellIcon,
+	FileSlidersIcon,
+	FingerprintIcon,
+	HeartPulseIcon,
+	SettingsIcon,
+} from "lucide-react"
 import { lazy, useEffect } from "react"
 import { $router } from "@/components/router.tsx"
 import { Card, CardContent, CardDescription, CardHeader, CardTitle } from "@/components/ui/card.tsx"
@@ -18,12 +25,14 @@ const notificationsSettingsImport = () => import("./notifications.tsx")
 const configYamlSettingsImport = () => import("./config-yaml.tsx")
 const fingerprintsSettingsImport = () => import("./tokens-fingerprints.tsx")
 const alertsHistoryDataTableSettingsImport = () => import("./alerts-history-data-table.tsx")
+const heartbeatSettingsImport = () => import("./heartbeat.tsx")
 
 const GeneralSettings = lazy(generalSettingsImport)
 const NotificationsSettings = lazy(notificationsSettingsImport)
 const ConfigYamlSettings = lazy(configYamlSettingsImport)
 const FingerprintsSettings = lazy(fingerprintsSettingsImport)
 const AlertsHistoryDataTableSettings = lazy(alertsHistoryDataTableSettingsImport)
+const HeartbeatSettings = lazy(heartbeatSettingsImport)
 
 export async function saveSettings(newSettings: Partial<UserSettings>) {
 	try {
@@ -81,6 +90,13 @@ export default function SettingsLayout() {
 			icon: AlertOctagonIcon,
 			preload: alertsHistoryDataTableSettingsImport,
 		},
+		{
+			title: t`Heartbeat`,
+			href: getPagePath($router, "settings", { name: "heartbeat" }),
+			icon: HeartPulseIcon,
+			admin: true,
+			preload: heartbeatSettingsImport,
+		},
 		{
 			title: t`YAML Config`,
 			href: getPagePath($router, "settings", { name: "config" }),
@@ -141,5 +157,7 @@ function SettingsContent({ name }: { name: string }) {
 			return <FingerprintsSettings />
 		case "alert-history":
 			return <AlertsHistoryDataTableSettings />
+		case "heartbeat":
+			return <HeartbeatSettings />
 	}
 }

internal/site/src/components/routes/system.tsx

@@ -593,7 +593,7 @@ export default memo(function SystemDetail({ id }: { id: string }) {
 										if (showMax) {
 											return data?.stats?.bm?.[0] ?? (data?.stats?.nsm ?? 0) * 1024 * 1024
 										}
-										return data?.stats?.b?.[0] ?? data?.stats?.ns * 1024 * 1024
+										return data?.stats?.b?.[0] ?? (data?.stats?.ns ?? 0) * 1024 * 1024
 									},
 									color: 5,
 									opacity: 0.2,
@@ -604,7 +604,7 @@ export default memo(function SystemDetail({ id }: { id: string }) {
 										if (showMax) {
 											return data?.stats?.bm?.[1] ?? (data?.stats?.nrm ?? 0) * 1024 * 1024
 										}
-										return data?.stats?.b?.[1] ?? data?.stats?.nr * 1024 * 1024
+										return data?.stats?.b?.[1] ?? (data?.stats?.nr ?? 0) * 1024 * 1024
 									},
 									color: 2,
 									opacity: 0.2,
@@ -654,7 +654,7 @@ export default memo(function SystemDetail({ id }: { id: string }) {
 					)}
 
 					{/* Load Average chart */}
-					{chartData.agentVersion?.minor >= 12 && (
+					{chartData.agentVersion?.minor > 12 && (
 						<ChartCard
 							empty={dataEmpty}
 							grid={grid}

internal/site/src/components/routes/system/smart-table.tsx

@@ -621,8 +621,8 @@ function DiskSheet({
 	const deviceName = disk?.name || unknown
 	const model = disk?.model || unknown
 	const capacity = disk?.capacity ? formatCapacity(disk.capacity) : unknown
-	const serialNumber = disk?.serial || unknown
-	const firmwareVersion = disk?.firmware || unknown
+	const serialNumber = disk?.serial
+	const firmwareVersion = disk?.firmware
 	const status = disk?.state || unknown
 
 	return (
@@ -636,34 +636,42 @@ function DiskSheet({
 						{model}
 						<Separator orientation="vertical" className="h-2.5 bg-muted-foreground opacity-70" />
 						{capacity}
-						<Separator orientation="vertical" className="h-2.5 bg-muted-foreground opacity-70" />
-						<Tooltip>
-							<TooltipTrigger asChild>
-								<span>{serialNumber}</span>
-							</TooltipTrigger>
-							<TooltipContent>
-								<Trans>Serial Number</Trans>
-							</TooltipContent>
-						</Tooltip>
-						<Separator orientation="vertical" className="h-2.5 bg-muted-foreground opacity-70" />
-						<Tooltip>
-							<TooltipTrigger asChild>
-								<span>{firmwareVersion}</span>
-							</TooltipTrigger>
-							<TooltipContent>
-								<Trans>Firmware</Trans>
-							</TooltipContent>
-						</Tooltip>
+						{serialNumber && (
+							<>
+								<Separator orientation="vertical" className="h-2.5 bg-muted-foreground opacity-70" />
+								<Tooltip>
+									<TooltipTrigger asChild>
+										<span>{serialNumber}</span>
+									</TooltipTrigger>
+									<TooltipContent>
+										<Trans>Serial Number</Trans>
+									</TooltipContent>
+								</Tooltip>
+							</>
+						)}
+						{firmwareVersion && (
+							<>
+								<Separator orientation="vertical" className="h-2.5 bg-muted-foreground opacity-70" />
+								<Tooltip>
+									<TooltipTrigger asChild>
+										<span>{firmwareVersion}</span>
+									</TooltipTrigger>
+									<TooltipContent>
+										<Trans>Firmware</Trans>
+									</TooltipContent>
+								</Tooltip>
+							</>
+						)}
 					</SheetDescription>
 				</SheetHeader>
-				<div className="flex-1 overflow-auto p-4 flex flex-col gap-4">
+				<div className="flex-1 overflow-hidden p-4 flex flex-col gap-4">
 					{isLoading ? (
 						<div className="flex justify-center py-8">
 							<LoaderCircleIcon className="animate-spin size-10 opacity-60" />
 						</div>
 					) : (
 						<>
-							<Alert className="pb-3">
+							<Alert className="pb-3 shrink-0">
 								{status === "PASSED" ? <CheckCircle2Icon className="size-4" /> : <XCircleIcon className="size-4" />}
 								<AlertTitle>
 									<Trans>S.M.A.R.T. Self-Test</Trans>: {status}
@@ -675,9 +683,9 @@ function DiskSheet({
 								)}
 							</Alert>
 							{smartAttributes.length > 0 ? (
-								<div className="rounded-md border overflow-auto">
+								<div className="rounded-md border min-h-0 flex flex-col">
 									<Table>
-										<TableHeader>
+										<TableHeader className="sticky top-0 z-10">
 											{table.getHeaderGroups().map((headerGroup) => (
 												<TableRow key={headerGroup.id}>
 													{headerGroup.headers.map((header) => (

internal/site/src/components/systems-table/systems-table-columns.tsx

@@ -33,7 +33,6 @@ import {
 	decimalString,
 	formatBytes,
 	formatTemperature,
-	getMeterState,
 	parseSemVer,
 	secondsToUptimeString,
 } from "@/lib/utils"
@@ -81,6 +80,10 @@ const STATUS_COLORS = {
 	[SystemStatus.Pending]: "bg-yellow-500",
 } as const
 
+function getMeterStateByThresholds(value: number, warn = 65, crit = 90): MeterState {
+	return value >= crit ? MeterState.Crit : value >= warn ? MeterState.Warn : MeterState.Good
+}
+
 /**
  * @param viewMode - "table" or "grid"
  * @returns - Column definitions for the systems table
@@ -195,36 +198,24 @@ export function SystemsTableColumns(viewMode: "table" | "grid"): ColumnDef<Syste
 		},
 		{
 			id: "loadAverage",
-			accessorFn: ({ info }) => {
-				const sum = info.la?.reduce((acc, curr) => acc + curr, 0)
-				// TODO: remove this in future release in favor of la array
-				if (!sum) {
-					return (info.l1 ?? 0) + (info.l5 ?? 0) + (info.l15 ?? 0) || undefined
-				}
-				return sum || undefined
-			},
+			accessorFn: ({ info }) => info.la?.reduce((acc, curr) => acc + curr, 0),
 			name: () => t({ message: "Load Avg", comment: "Short label for load average" }),
 			size: 0,
 			Icon: HourglassIcon,
 			header: sortableHeader,
 			cell(info: CellContext<SystemRecord, unknown>) {
 				const { info: sysInfo, status } = info.row.original
-				// agent version
-				const { minor, patch } = parseSemVer(sysInfo.v)
-				let loadAverages = sysInfo.la
-
-				// use legacy load averages if agent version is less than 12.1.0
-				if (!loadAverages || (minor === 12 && patch < 1)) {
-					loadAverages = [sysInfo.l1 ?? 0, sysInfo.l5 ?? 0, sysInfo.l15 ?? 0]
-				}
+				const { major, minor } = parseSemVer(sysInfo.v)
+				const { colorWarn = 65, colorCrit = 90 } = useStore($userSettings, { keys: ["colorWarn", "colorCrit"] })
+				const loadAverages = sysInfo.la || []
 
 				const max = Math.max(...loadAverages)
-				if (max === 0 && (status === SystemStatus.Paused || minor < 12)) {
+				if (max === 0 && (status === SystemStatus.Paused || (major < 1 && minor < 13))) {
 					return null
 				}
 
 				const normalizedLoad = max / (sysInfo.t ?? 1)
-				const threshold = getMeterState(normalizedLoad * 100)
+				const threshold = getMeterStateByThresholds(normalizedLoad * 100, colorWarn, colorCrit)
 
 				return (
 					<div className="flex items-center gap-[.35em] w-full tabular-nums tracking-tight">
@@ -244,19 +235,20 @@ export function SystemsTableColumns(viewMode: "table" | "grid"): ColumnDef<Syste
 			},
 		},
 		{
-			accessorFn: ({ info }) => info.bb || (info.b || 0) * 1024 * 1024 || undefined,
+			accessorFn: ({ info, status }) => (status !== SystemStatus.Up ? undefined : info.bb),
 			id: "net",
 			name: () => t`Net`,
 			size: 0,
 			Icon: EthernetIcon,
 			header: sortableHeader,
+			sortUndefined: "last",
 			cell(info) {
-				const sys = info.row.original
-				const userSettings = useStore($userSettings, { keys: ["unitNet"] })
-				if (sys.status === SystemStatus.Paused) {
+				const val = info.getValue() as number | undefined
+				if (val === undefined) {
 					return null
 				}
-				const { value, unit } = formatBytes((info.getValue() || 0) as number, true, userSettings.unitNet, false)
+				const userSettings = useStore($userSettings, { keys: ["unitNet"] })
+				const { value, unit } = formatBytes(val, true, userSettings.unitNet, false)
 				return (
 					<span className="tabular-nums whitespace-nowrap">
 						{decimalString(value, value >= 100 ? 1 : 2)} {unit}
@@ -463,8 +455,9 @@ function sortableHeader(context: HeaderContext<SystemRecord, unknown>) {
 }
 
 function TableCellWithMeter(info: CellContext<SystemRecord, unknown>) {
+	const { colorWarn = 65, colorCrit = 90 } = useStore($userSettings, { keys: ["colorWarn", "colorCrit"] })
 	const val = Number(info.getValue()) || 0
-	const threshold = getMeterState(val)
+	const threshold = getMeterStateByThresholds(val, colorWarn, colorCrit)
 	const meterClass = cn(
 		"h-full",
 		(info.row.original.status !== SystemStatus.Up && STATUS_COLORS.paused) ||
@@ -483,6 +476,7 @@ function TableCellWithMeter(info: CellContext<SystemRecord, unknown>) {
 }
 
 function DiskCellWithMultiple(info: CellContext<SystemRecord, unknown>) {
+	const { colorWarn = 65, colorCrit = 90 } = useStore($userSettings, { keys: ["colorWarn", "colorCrit"] })
 	const { info: sysInfo, status, id } = info.row.original
 	const extraFs = Object.entries(sysInfo.efs ?? {})
 
@@ -496,7 +490,7 @@ function DiskCellWithMultiple(info: CellContext<SystemRecord, unknown>) {
 	extraFs.sort((a, b) => b[1] - a[1])
 
 	function getIndicatorColor(pct: number) {
-		const threshold = getMeterState(pct)
+		const threshold = getMeterStateByThresholds(pct, colorWarn, colorCrit)
 		return (
 			(status !== SystemStatus.Up && STATUS_COLORS.paused) ||
 			(threshold === MeterState.Good && STATUS_COLORS.up) ||
@@ -514,7 +508,9 @@ function DiskCellWithMultiple(info: CellContext<SystemRecord, unknown>) {
 	const extraDiskIndicators =
 		status !== SystemStatus.Up
 			? []
-			: [...new Set(extraFs.map(([, pct]) => getMeterState(pct)))].sort().map((state) => stateColors[state])
+			: [...new Set(extraFs.map(([, pct]) => getMeterStateByThresholds(pct, colorWarn, colorCrit)))]
+					.sort()
+					.map((state) => stateColors[state])
 
 	return (
 		<Tooltip>

internal/site/src/lib/alerts.ts

@@ -40,7 +40,7 @@ export const alertInfo: Record<string, AlertInfo> = {
 		unit: " MB/s",
 		icon: EthernetIcon,
 		desc: () => t`Triggers when combined up/down exceeds a threshold`,
-		max: 125,
+		max: 250,
 	},
 	GPU: {
 		name: () => t`GPU Usage`,

internal/site/src/lib/languages.ts

@@ -8,7 +8,7 @@ export default [
 	["es", "Español", "🇪🇸"],
 	["fa", "فارسی", "🇮🇷"],
 	["fr", "Français", "🇫🇷"],
-	["he", "עברית", "🕎"],
+	["he", "עברית", ""],
 	["hr", "Hrvatski", "🇭🇷"],
 	["hu", "Magyar", "🇭🇺"],
 	["id", "Indonesia", "🇮🇩"],

internal/site/src/lib/utils.ts

@@ -6,7 +6,7 @@ import { useEffect, useState } from "react"
 import { twMerge } from "tailwind-merge"
 import { toast } from "@/components/ui/use-toast"
 import type { ChartTimeData, FingerprintRecord, SemVer, SystemRecord } from "@/types"
-import { HourFormat, MeterState, Unit } from "./enums"
+import { HourFormat, Unit } from "./enums"
 import { $copyContent, $userSettings } from "./stores"
 
 export function cn(...inputs: ClassValue[]) {
@@ -210,7 +210,6 @@ export function useBrowserStorage<T>(key: string, defaultValue: T, storageInterf
 	const [value, setValue] = useState(() => {
 		return getStorageValue(key, defaultValue, storageInterface)
 	})
-	// biome-ignore lint/correctness/useExhaustiveDependencies: storageInterface won't change
 	useEffect(() => {
 		storageInterface?.setItem(key, JSON.stringify(value))
 	}, [key, value])
@@ -394,12 +393,6 @@ export function compareSemVer(a: SemVer, b: SemVer) {
 	return a.patch - b.patch
 }
 
-/** Get meter state from 0-100 value. Used for color coding meters. */
-export function getMeterState(value: number): MeterState {
-	const { colorWarn = 65, colorCrit = 90 } = $userSettings.get()
-	return value >= colorCrit ? MeterState.Crit : value >= colorWarn ? MeterState.Warn : MeterState.Good
-}
-
 // biome-ignore lint/suspicious/noExplicitAny: any is used to allow any function to be passed in
 export function debounce<T extends (...args: any[]) => any>(func: T, wait: number): (...args: Parameters<T>) => void {
 	let timeout: ReturnType<typeof setTimeout>

internal/site/src/locales/ar/ar.po

@@ -93,6 +93,7 @@ msgstr "إجراءات"
 
 #: src/components/alerts-history-columns.tsx
 #: src/components/routes/settings/alerts-history-data-table.tsx
+#: src/components/routes/settings/heartbeat.tsx
 #: src/components/routes/settings/quiet-hours.tsx
 msgid "Active"
 msgstr "نشط"
@@ -140,6 +141,10 @@ msgstr "مسؤول"
 msgid "After"
 msgstr "بعد"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "After setting the environment variables, restart your Beszel hub for changes to take effect."
+msgstr "بعد تعيين متغيرات البيئة، أعد تشغيل مركز Beszel الخاص بك لتصبح التغييرات سارية المفعول."
+
 #: src/components/systems-table/systems-table-columns.tsx
 msgid "Agent"
 msgstr "وكيل"
@@ -350,6 +355,10 @@ msgstr "تحقق من {email} للحصول على رابط إعادة التعي
 msgid "Check logs for more details."
 msgstr "تحقق من السجلات لمزيد من التفاصيل."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Check your monitoring service"
+msgstr "تحقق من خدمة المراقبة الخاصة بك"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Check your notification service"
 msgstr "تحقق من خدمة الإشعارات الخاصة بك"
@@ -643,6 +652,14 @@ msgstr "فارغة"
 msgid "End Time"
 msgstr "وقت النهاية"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Endpoint URL"
+msgstr "عنوان URL للنقطة النهائية"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Endpoint URL to ping (required)"
+msgstr "عنوان URL للنقطة النهائية لل ping (مطلوب)"
+
 #: src/components/login/login.tsx
 msgid "Enter email address to reset password"
 msgstr "أدخل عنوان البريد الإشباكي لإعادة تعيين كلمة المرور"
@@ -662,6 +679,9 @@ msgstr "مؤقت"
 #: src/components/login/auth-form.tsx
 #: src/components/routes/settings/alerts-history-data-table.tsx
 #: src/components/routes/settings/config-yaml.tsx
+#: src/components/routes/settings/heartbeat.tsx
+#: src/components/routes/settings/heartbeat.tsx
+#: src/components/routes/settings/heartbeat.tsx
 #: src/components/routes/settings/notifications.tsx
 #: src/components/routes/settings/quiet-hours.tsx
 #: src/components/routes/settings/quiet-hours.tsx
@@ -670,6 +690,10 @@ msgstr "مؤقت"
 msgid "Error"
 msgstr "خطأ"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Example:"
+msgstr "مثال:"
+
 #. placeholder {0}: alert.value
 #. placeholder {1}: info.unit
 #. placeholder {2}: alert.min
@@ -727,6 +751,10 @@ msgstr "فشل في المصادقة"
 msgid "Failed to save settings"
 msgstr "فشل في حفظ الإعدادات"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Failed to send heartbeat"
+msgstr "فشل في إرسال نبضة القلب"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Failed to send test notification"
 msgstr "فشل في إرسال إشعار الاختبار"
@@ -806,6 +834,18 @@ msgstr "شبكة"
 msgid "Health"
 msgstr "الصحة"
 
+#: src/components/routes/settings/layout.tsx
+msgid "Heartbeat"
+msgstr "نبضة القلب"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Heartbeat Monitoring"
+msgstr "مراقبة نبضة القلب"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Heartbeat sent successfully"
+msgstr "تم إرسال نبضة القلب بنجاح"
+
 #: src/components/add-system.tsx
 #: src/components/routes/settings/tokens-fingerprints.tsx
 msgctxt "Button to copy install command"
@@ -816,6 +856,14 @@ msgstr "أمر Homebrew"
 msgid "Host / IP"
 msgstr "مضيف / IP"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "HTTP Method"
+msgstr "طريقة HTTP"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "HTTP method: POST, GET, or HEAD (default: POST)"
+msgstr "طريقة HTTP: POST، GET، أو HEAD (الافتراضي: POST)"
+
 #. Context: Battery state
 #: src/lib/i18n.ts
 msgid "Idle"
@@ -834,6 +882,10 @@ msgstr "صورة"
 msgid "Inactive"
 msgstr "غير نشط"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Interval"
+msgstr "الفاصل الزمني"
+
 #: src/components/login/auth-form.tsx
 msgid "Invalid email address."
 msgstr "عنوان البريد الإشباكي غير صالح."
@@ -1110,6 +1162,10 @@ msgstr "متوقف مؤقتا"
 msgid "Paused ({pausedSystemsLength})"
 msgstr "متوقف مؤقتا ({pausedSystemsLength})"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Payload format"
+msgstr "تنسيق الحمولة"
+
 #: src/components/routes/system/cpu-sheet.tsx
 #: src/components/routes/system/cpu-sheet.tsx
 msgid "Per-core average utilization"
@@ -1307,6 +1363,10 @@ msgstr "بحث"
 msgid "Search for systems or settings..."
 msgstr "البحث عن الأنظمة أو الإعدادات..."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Seconds between pings (default: 60)"
+msgstr "ثواني بين ال pings (الافتراضي: 60)"
+
 #: src/components/alerts/alerts-sheet.tsx
 msgid "See <0>notification settings</0> to configure how you receive alerts."
 msgstr "راجع <0>إعدادات الإشعارات</0> لتكوين كيفية تلقي التنبيهات."
@@ -1315,6 +1375,18 @@ msgstr "راجع <0>إعدادات الإشعارات</0> لتكوين كيفي
 msgid "Select {foo}"
 msgstr "تحديد {foo}"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send a single heartbeat ping to verify your endpoint is working."
+msgstr "أرسل ping نبضة قلب واحدة للتحقق من أن نقطة النهاية الخاصة بك تعمل."
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send periodic outbound pings to an external monitoring service so you can monitor Beszel without exposing it to the internet."
+msgstr "أرسل pings صادرة دورية إلى خدمة مراقبة خارجية حتى تتمكن من مراقبة Beszel دون تعريضه للإنترنت."
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send test heartbeat"
+msgstr "إرسال نبضة قلب اختبارية"
+
 #: src/components/routes/system.tsx
 msgid "Sent"
 msgstr "تم الإرسال"
@@ -1335,6 +1407,10 @@ msgstr "الخدمات"
 msgid "Set percentage thresholds for meter colors."
 msgstr "تعيين عتبات النسبة المئوية لألوان العداد."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Set the following environment variables on your Beszel hub to enable heartbeat monitoring:"
+msgstr "قم بتعيين متغيرات البيئة التالية على مركز Beszel الخاص بك لتمكين مراقبة نبضة القلب:"
+
 #: src/components/command-palette.tsx
 #: src/components/command-palette.tsx
 #: src/components/navbar.tsx
@@ -1452,10 +1528,18 @@ msgstr "درجات حرارة مستشعرات النظام"
 msgid "Test <0>URL</0>"
 msgstr "اختبار <0>URL</0>"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Test heartbeat"
+msgstr "اختبار نبضة القلب"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Test notification sent"
 msgstr "تم إرسال إشعار الاختبار"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "The overall status is <0>ok</0> when all systems are up, <1>warn</1> when alerts are triggered, and <2>error</2> when any system is down."
+msgstr "الحالة العامة هي <0>موافق</0> عندما تكون جميع الأنظمة تعمل، و<1>تحذير</1> عند تشغيل التنبيهات، و<2>خطأ</2> عندما يكون أي نظام معطلاً."
+
 #: src/components/login/forgot-pass-form.tsx
 msgid "Then log into the backend and reset your user account password in the users table."
 msgstr "ثم قم بتسجيل الدخول إلى الواجهة الخلفية وأعد تعيين كلمة مرور حساب المستخدم الخاص بك في جدول المستخدمين."
@@ -1642,6 +1726,7 @@ msgid "Upload"
 msgstr "رفع"
 
 #: src/components/routes/system/info-bar.tsx
+#: src/components/systems-table/systems-table-columns.tsx
 msgid "Uptime"
 msgstr "مدة التشغيل"
 
@@ -1716,6 +1801,10 @@ msgstr "إشعارات Webhook / Push"
 msgid "When enabled, this token allows agents to self-register without prior system creation."
 msgstr "عند التفعيل، يسمح هذا الرمز المميز للوكلاء بالتسجيل الذاتي دون إنشاء نظام مسبق."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "When using POST, each heartbeat includes a JSON payload with system status summary, list of down systems, and triggered alerts."
+msgstr "عند استخدام POST، تتضمن كل نبضة قلب حمولة JSON مع ملخص حالة النظام وقائمة الأنظمة المعطلة والتنبيهات التي تم تشغيلها."
+
 #: src/components/add-system.tsx
 #: src/components/routes/settings/tokens-fingerprints.tsx
 msgctxt "Button to copy install command"

internal/site/src/locales/bg/bg.po

@@ -93,6 +93,7 @@ msgstr "Действия"
 
 #: src/components/alerts-history-columns.tsx
 #: src/components/routes/settings/alerts-history-data-table.tsx
+#: src/components/routes/settings/heartbeat.tsx
 #: src/components/routes/settings/quiet-hours.tsx
 msgid "Active"
 msgstr "Активен"
@@ -140,6 +141,10 @@ msgstr "Администратор"
 msgid "After"
 msgstr "След"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "After setting the environment variables, restart your Beszel hub for changes to take effect."
+msgstr "След като настроите променливите на средата, рестартирайте вашия Beszel hub, за да влязат промените в сила."
+
 #: src/components/systems-table/systems-table-columns.tsx
 msgid "Agent"
 msgstr "Агент"
@@ -231,7 +236,7 @@ msgstr "Bandwidth на мрежата"
 #. Battery label in systems table header
 #: src/components/systems-table/systems-table-columns.tsx
 msgid "Bat"
-msgstr ""
+msgstr "Бат"
 
 #: src/components/routes/system.tsx
 #: src/lib/alerts.ts
@@ -350,6 +355,10 @@ msgstr "Провери {email} за линк за нулиране."
 msgid "Check logs for more details."
 msgstr "Провери log-овете за повече информация."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Check your monitoring service"
+msgstr "Проверете вашата услуга за мониторинг"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Check your notification service"
 msgstr "Провери услугата си за удостоверяване"
@@ -621,7 +630,7 @@ msgstr "Редактирай"
 #: src/components/add-system.tsx
 #: src/components/routes/settings/quiet-hours.tsx
 msgid "Edit {foo}"
-msgstr ""
+msgstr "Редактиране на {foo}"
 
 #: src/components/login/auth-form.tsx
 #: src/components/login/forgot-pass-form.tsx
@@ -643,6 +652,14 @@ msgstr "Празна"
 msgid "End Time"
 msgstr "Крайно време"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Endpoint URL"
+msgstr "URL адрес на крайната точка"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Endpoint URL to ping (required)"
+msgstr "URL адрес на крайната точка за пинг (задължително)"
+
 #: src/components/login/login.tsx
 msgid "Enter email address to reset password"
 msgstr "Въведи имейл адрес за да нулираш паролата"
@@ -662,6 +679,9 @@ msgstr "Ефимерен"
 #: src/components/login/auth-form.tsx
 #: src/components/routes/settings/alerts-history-data-table.tsx
 #: src/components/routes/settings/config-yaml.tsx
+#: src/components/routes/settings/heartbeat.tsx
+#: src/components/routes/settings/heartbeat.tsx
+#: src/components/routes/settings/heartbeat.tsx
 #: src/components/routes/settings/notifications.tsx
 #: src/components/routes/settings/quiet-hours.tsx
 #: src/components/routes/settings/quiet-hours.tsx
@@ -670,6 +690,10 @@ msgstr "Ефимерен"
 msgid "Error"
 msgstr "Грешка"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Example:"
+msgstr "Пример:"
+
 #. placeholder {0}: alert.value
 #. placeholder {1}: info.unit
 #. placeholder {2}: alert.min
@@ -679,7 +703,7 @@ msgstr "Надвишава {0}{1} в последните {2, plural, one {# м
 
 #: src/components/systemd-table/systemd-table.tsx
 msgid "Exec main PID"
-msgstr ""
+msgstr "PID на главния изпълнителен процес"
 
 #: src/components/routes/settings/config-yaml.tsx
 msgid "Existing systems not defined in <0>config.yml</0> will be deleted. Please make regular backups."
@@ -727,6 +751,10 @@ msgstr "Неуспешно удостоверяване"
 msgid "Failed to save settings"
 msgstr "Неуспешно запазване на настройки"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Failed to send heartbeat"
+msgstr "Неуспешно изпращане на heartbeat"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Failed to send test notification"
 msgstr "Неуспешно изпрати тестова нотификация"
@@ -806,6 +834,18 @@ msgstr "Мрежово"
 msgid "Health"
 msgstr "Здраве"
 
+#: src/components/routes/settings/layout.tsx
+msgid "Heartbeat"
+msgstr "Heartbeat"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Heartbeat Monitoring"
+msgstr "Мониторинг на heartbeat"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Heartbeat sent successfully"
+msgstr "Heartbeat е изпратен успешно"
+
 #: src/components/add-system.tsx
 #: src/components/routes/settings/tokens-fingerprints.tsx
 msgctxt "Button to copy install command"
@@ -816,6 +856,14 @@ msgstr "Команда Homebrew"
 msgid "Host / IP"
 msgstr "Хост / IP"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "HTTP Method"
+msgstr "HTTP метод"
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "HTTP method: POST, GET, or HEAD (default: POST)"
+msgstr "HTTP метод: POST, GET или HEAD (по подразбиране: POST)"
+
 #. Context: Battery state
 #: src/lib/i18n.ts
 msgid "Idle"
@@ -834,6 +882,10 @@ msgstr "Образ"
 msgid "Inactive"
 msgstr "Неактивен"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Interval"
+msgstr "Интервал"
+
 #: src/components/login/auth-form.tsx
 msgid "Invalid email address."
 msgstr "Невалиден имейл адрес."
@@ -914,7 +966,7 @@ msgstr "Търсиш къде да създадеш тревоги? Натисн
 
 #: src/components/systemd-table/systemd-table.tsx
 msgid "Main PID"
-msgstr ""
+msgstr "Главен PID"
 
 #: src/components/routes/settings/layout.tsx
 msgid "Manage display and notification preferences."
@@ -1110,6 +1162,10 @@ msgstr "На пауза"
 msgid "Paused ({pausedSystemsLength})"
 msgstr "На пауза ({pausedSystemsLength})"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Payload format"
+msgstr "Формат на полезния товар"
+
 #: src/components/routes/system/cpu-sheet.tsx
 #: src/components/routes/system/cpu-sheet.tsx
 msgid "Per-core average utilization"
@@ -1307,6 +1363,10 @@ msgstr "Търси"
 msgid "Search for systems or settings..."
 msgstr "Търси за системи или настройки..."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Seconds between pings (default: 60)"
+msgstr "Секунди между пинговете (по подразбиране: 60)"
+
 #: src/components/alerts/alerts-sheet.tsx
 msgid "See <0>notification settings</0> to configure how you receive alerts."
 msgstr "Виж <0>настройките за нотификациите</0> за да конфигурираш как получаваш тревоги."
@@ -1315,6 +1375,18 @@ msgstr "Виж <0>настройките за нотификациите</0> з
 msgid "Select {foo}"
 msgstr "Избери {foo}"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send a single heartbeat ping to verify your endpoint is working."
+msgstr "Изпратете единичен heartbeat пинг, за да проверите дали вашата крайна точка работи."
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send periodic outbound pings to an external monitoring service so you can monitor Beszel without exposing it to the internet."
+msgstr "Изпращайте периодични изходящи пингове към външна услуга за мониторинг, за да можете да наблюдавате Beszel, без да го излагате на интернет."
+
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Send test heartbeat"
+msgstr "Изпращане на тестов heartbeat"
+
 #: src/components/routes/system.tsx
 msgid "Sent"
 msgstr "Изпратени"
@@ -1335,6 +1407,10 @@ msgstr "Услуги"
 msgid "Set percentage thresholds for meter colors."
 msgstr "Задайте процентни прагове за цветовете на измервателните уреди."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Set the following environment variables on your Beszel hub to enable heartbeat monitoring:"
+msgstr "Задайте следните променливи на средата на вашия Beszel hub, за да активирате мониторинга на heartbeat:"
+
 #: src/components/command-palette.tsx
 #: src/components/command-palette.tsx
 #: src/components/navbar.tsx
@@ -1452,10 +1528,18 @@ msgstr "Температири на системни сензори"
 msgid "Test <0>URL</0>"
 msgstr "Тествай <0>URL</0>"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "Test heartbeat"
+msgstr "Тестов heartbeat"
+
 #: src/components/routes/settings/notifications.tsx
 msgid "Test notification sent"
 msgstr "Тестова нотификация изпратена"
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "The overall status is <0>ok</0> when all systems are up, <1>warn</1> when alerts are triggered, and <2>error</2> when any system is down."
+msgstr "Общият статус е <0>ok</0>, когато всички системи работят, <1>warn</1>, когато са задействани предупреждения, и <2>error</2>, когато някоя система е спряла."
+
 #: src/components/login/forgot-pass-form.tsx
 msgid "Then log into the backend and reset your user account password in the users table."
 msgstr "След това влез в backend-а и нулирай паролата за потребителския акаунт в таблицата за потребители."
@@ -1642,8 +1726,9 @@ msgid "Upload"
 msgstr "Качване"
 
 #: src/components/routes/system/info-bar.tsx
+#: src/components/systems-table/systems-table-columns.tsx
 msgid "Uptime"
-msgstr "Време на работа"
+msgstr "Uptime"
 
 #: src/components/routes/system.tsx
 #: src/components/routes/system.tsx
@@ -1716,6 +1801,10 @@ msgstr "Webhook / Пуш нотификации"
 msgid "When enabled, this token allows agents to self-register without prior system creation."
 msgstr "Когато е активиран, този символ позволява на агентите да се регистрират сами без предварително създаване на система."
 
+#: src/components/routes/settings/heartbeat.tsx
+msgid "When using POST, each heartbeat includes a JSON payload with system status summary, list of down systems, and triggered alerts."
+msgstr "При използване на POST всеки heartbeat включва JSON полезен товар с резюме на състоянието на системата, списък на спрените системи и задействаните предупреждения."
+
 #: src/components/add-system.tsx
 #: src/components/routes/settings/tokens-fingerprints.tsx
 msgctxt "Button to copy install command"