update color

* fix(install): detect rc.conf.d override that blocks FreeBSD service start

* fix log file removal

.github/dependabot.yml

@@ -0,0 +1,12 @@
+version: 2
+updates:
+  - package-ecosystem: gomod
+    directory: /
+    schedule:
+      interval: weekly
+
+  - package-ecosystem: github-actions
+    directory: /
+    schedule:
+      interval: weekly
+

.github/workflows/docker-images.yml

@@ -41,7 +41,7 @@ jobs:
           # henrygd/beszel-agent-nvidia
           - image: henrygd/beszel-agent-nvidia
             dockerfile: ./internal/dockerfile_agent_nvidia
-            platforms: linux/amd64
+            platforms: linux/amd64,linux/arm64
             registry: docker.io
             username_secret: DOCKERHUB_USERNAME
             password_secret: DOCKERHUB_TOKEN
@@ -96,7 +96,7 @@ jobs:
           # ghcr.io/henrygd/beszel-agent-nvidia
           - image: ghcr.io/${{ github.repository }}/beszel-agent-nvidia
             dockerfile: ./internal/dockerfile_agent_nvidia
-            platforms: linux/amd64
+            platforms: linux/amd64,linux/arm64
             registry: ghcr.io
             username: ${{ github.actor }}
             password_secret: GITHUB_TOKEN
@@ -152,7 +152,7 @@ jobs:
 
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
 
       - name: Set up bun
         uses: oven-sh/setup-bun@v2
@@ -164,14 +164,14 @@ jobs:
         run: bun run --cwd ./internal/site build
 
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v3
+        uses: docker/setup-qemu-action@v4
 
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
 
       - name: Docker metadata
         id: metadata
-        uses: docker/metadata-action@v5
+        uses: docker/metadata-action@v6
         with:
           images: ${{ matrix.image }}
           tags: ${{ matrix.tags }}
@@ -181,7 +181,7 @@ jobs:
         env:
           password_secret_exists: ${{ secrets[matrix.password_secret] != '' && 'true' || 'false' }}
         if: github.event_name != 'pull_request' && env.password_secret_exists == 'true'
-        uses: docker/login-action@v3
+        uses: docker/login-action@v4
         with:
           username: ${{ matrix.username || secrets[matrix.username_secret] }}
           password: ${{ secrets[matrix.password_secret] }}
@@ -190,11 +190,13 @@ jobs:
       # Build and push Docker image with Buildx (don't push on PR)
       # https://github.com/docker/build-push-action
       - name: Build and push Docker image
-        uses: docker/build-push-action@v5
+        uses: docker/build-push-action@v7
         with:
           context: ./
           file: ${{ matrix.dockerfile }}
           platforms: ${{ matrix.platforms || 'linux/amd64,linux/arm64,linux/arm/v7' }}
           push: ${{ github.ref_type == 'tag' && secrets[matrix.password_secret] != '' }}
+          provenance: mode=max
+          sbom: true
           tags: ${{ steps.metadata.outputs.tags }}
           labels: ${{ steps.metadata.outputs.labels }}

.github/workflows/inactivity-actions.yml

@@ -15,7 +15,7 @@ jobs:
     name: Lock Inactive Issues
     runs-on: ubuntu-24.04
     steps:
-      - uses: klaasnicolaas/action-inactivity-lock@v1.1.3
+      - uses: klaasnicolaas/action-inactivity-lock@v2.0.1
         id: lock
         with:
           days-inactive-issues: 14
@@ -29,7 +29,7 @@ jobs:
     runs-on: ubuntu-24.04
     steps:
       - name: Close Stale Issues
-        uses: actions/stale@v10
+        uses: actions/stale@v11
         with:
           repo-token: ${{ secrets.GITHUB_TOKEN }}
 

.github/workflows/release.yml

@@ -13,7 +13,7 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
         with:
           fetch-depth: 0
 
@@ -27,12 +27,12 @@ jobs:
         run: bun run --cwd ./internal/site build
 
       - name: Set up Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
           go-version: "^1.22.1"
 
       - name: Set up .NET
-        uses: actions/setup-dotnet@v4
+        uses: actions/setup-dotnet@v6
         with:
           dotnet-version: "9.0.x"
 
@@ -42,7 +42,7 @@ jobs:
         shell: bash
 
       - name: GoReleaser beszel
-        uses: goreleaser/goreleaser-action@v6
+        uses: goreleaser/goreleaser-action@v7
         with:
           workdir: ./
           distribution: goreleaser

.github/workflows/vulncheck.yml

@@ -19,11 +19,11 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Check out code into the Go module directory
-        uses: actions/checkout@v4
+        uses: actions/checkout@v7
       - name: Set up Go
-        uses: actions/setup-go@v5
+        uses: actions/setup-go@v7
         with:
-          go-version: 1.25.x
+          go-version: stable
           # cached: false
       - name: Get official govulncheck
         run: go install golang.org/x/vuln/cmd/govulncheck@latest

Makefile

@@ -52,7 +52,7 @@ lint:
 	golangci-lint run
 
 test:
-	go test -tags=testing ./...
+	go test -tags='testing no_ui' ./...
 
 tidy:
 	go mod tidy

agent/agent.go

@@ -19,6 +19,8 @@ import (
 	gossh "golang.org/x/crypto/ssh"
 )
 
+const defaultDataCacheTimeMs uint16 = 60_000
+
 type Agent struct {
 	sync.Mutex                                                                      // Used to lock agent while collecting data
 	debug                     bool                                                  // true if LOG_LEVEL is set to debug
@@ -36,6 +38,7 @@ type Agent struct {
 	sensorConfig              *SensorConfig                                         // Sensors config
 	systemInfo                system.Info                                           // Host system info (dynamic)
 	systemDetails             system.Details                                        // Host system details (static, once-per-connection)
+	detailsDirty              bool                                                  // Whether system details have changed and need to be resent
 	gpuManager                *GPUManager                                           // Manages GPU data
 	cache                     *systemDataCache                                      // Cache for system stats based on cache time
 	connectionManager         *ConnectionManager                                    // Channel to signal connection events
@@ -97,7 +100,7 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 	slog.Debug(beszel.Version)
 
 	// initialize docker manager
-	agent.dockerManager = newDockerManager()
+	agent.dockerManager = newDockerManager(agent)
 
 	// initialize system info
 	agent.refreshSystemDetails()
@@ -142,7 +145,7 @@ func NewAgent(dataDir ...string) (agent *Agent, err error) {
 
 	// if debugging, print stats
 	if agent.debug {
-		slog.Debug("Stats", "data", agent.gatherStats(common.DataRequestOptions{CacheTimeMs: 60_000, IncludeDetails: true}))
+		slog.Debug("Stats", "data", agent.gatherStats(common.DataRequestOptions{CacheTimeMs: defaultDataCacheTimeMs, IncludeDetails: true}))
 	}
 
 	return agent, nil
@@ -164,11 +167,6 @@ func (a *Agent) gatherStats(options common.DataRequestOptions) *system.CombinedD
 		Info:  a.systemInfo,
 	}
 
-	// Include static system details only when requested
-	if options.IncludeDetails {
-		data.Details = &a.systemDetails
-	}
-
 	// slog.Info("System data", "data", data, "cacheTimeMs", cacheTimeMs)
 
 	if a.dockerManager != nil {
@@ -181,7 +179,7 @@ func (a *Agent) gatherStats(options common.DataRequestOptions) *system.CombinedD
 	}
 
 	// skip updating systemd services if cache time is not the default 60sec interval
-	if a.systemdManager != nil && cacheTimeMs == 60_000 {
+	if a.systemdManager != nil && cacheTimeMs == defaultDataCacheTimeMs {
 		totalCount := uint16(a.systemdManager.getServiceStatsCount())
 		if totalCount > 0 {
 			numFailed := a.systemdManager.getFailedServiceCount()
@@ -212,7 +210,8 @@ func (a *Agent) gatherStats(options common.DataRequestOptions) *system.CombinedD
 	slog.Debug("Extra FS", "data", data.Stats.ExtraFs)
 
 	a.cache.Set(data, cacheTimeMs)
-	return data
+
+	return a.attachSystemDetails(data, cacheTimeMs, options.IncludeDetails)
 }
 
 // Start initializes and starts the agent with optional WebSocket connection

agent/battery/battery.go

@@ -1,84 +1,70 @@
-//go:build !freebsd
-
-// Package battery provides functions to check if the system has a battery and to get the battery stats.
+// Package battery provides battery information for the host and connected devices.
 package battery
 
 import (
 	"errors"
-	"log/slog"
-	"math"
-
-	"github.com/distatus/battery"
+	"sort"
+	"strconv"
+	"strings"
 )
 
-var (
-	systemHasBattery   = false
-	haveCheckedBattery = false
+const (
+	stateUnknown uint8 = iota
+	stateEmpty
+	stateFull
+	stateCharging
+	stateDischarging
+	stateIdle
 )
 
-// HasReadableBattery checks if the system has a battery and returns true if it does.
-func HasReadableBattery() bool {
-	if haveCheckedBattery {
-		return systemHasBattery
-	}
-	haveCheckedBattery = true
-	batteries, err := battery.GetAll()
-	for _, bat := range batteries {
-		if bat != nil && (bat.Full > 0 || bat.Design > 0) {
-			systemHasBattery = true
-			break
-		}
-	}
-	if !systemHasBattery {
-		slog.Debug("No battery found", "err", err)
-	}
-	return systemHasBattery
+// Battery is a readable battery reported by the operating system.
+type Battery struct {
+	Name                  string
+	Percent               uint8
+	State                 uint8
+	FullChargeCapacity    uint64
+	HasFullChargeCapacity bool
+	System                bool
 }
 
-// GetBatteryStats returns the current battery percent and charge state
-// percent = (current charge of all batteries) / (sum of designed/full capacity of all batteries)
-func GetBatteryStats() (batteryPercent uint8, batteryState uint8, err error) {
-	if !HasReadableBattery() {
-		return batteryPercent, batteryState, errors.ErrUnsupported
+var errNoBatteries = errors.New("no readable batteries")
+
+// normalizeBatteries supplies stable fallback names and disambiguates duplicates.
+func normalizeBatteries(batteries []Battery) []Battery {
+	nameCounts := make(map[string]int, len(batteries))
+	for i := range batteries {
+		name := strings.TrimSpace(batteries[i].Name)
+		if name == "" {
+			name = "Battery " + strconv.Itoa(i+1)
+		}
+		nameCounts[name]++
+		if nameCounts[name] > 1 {
+			name += " (" + strconv.Itoa(nameCounts[name]) + ")"
+		}
+		batteries[i].Name = name
 	}
-	batteries, err := battery.GetAll()
-	// we'll handle errors later by skipping batteries with errors, rather
-	// than skipping everything because of the presence of some errors.
+	return batteries
+}
+
+// Primary returns the representative battery. Reported full-charge capacity wins,
+// then system-scoped devices, then name for deterministic ties.
+func Primary(batteries []Battery) (Battery, bool) {
 	if len(batteries) == 0 {
-		return batteryPercent, batteryState, errors.New("no batteries")
+		return Battery{}, false
 	}
-
-	totalCapacity := float64(0)
-	totalCharge := float64(0)
-	errs, partialErrs := err.(battery.Errors)
-
-	batteryState = math.MaxUint8
-
-	for i, bat := range batteries {
-		if partialErrs && errs[i] != nil {
-			// if there were some errors, like missing data, skip it
-			continue
+	ordered := append([]Battery(nil), batteries...)
+	sort.SliceStable(ordered, func(i, j int) bool {
+		a, b := ordered[i], ordered[j]
+		if a.HasFullChargeCapacity != b.HasFullChargeCapacity {
+			return a.HasFullChargeCapacity
 		}
-		if bat == nil || bat.Full == 0 {
-			// skip batteries with no capacity. Charge is unlikely to ever be zero, but
-			// we can't guarantee that, so don't skip based on charge.
-			continue
+		if a.HasFullChargeCapacity && a.FullChargeCapacity != b.FullChargeCapacity {
+			return a.FullChargeCapacity > b.FullChargeCapacity
 		}
-		totalCapacity += bat.Full
-		totalCharge += min(bat.Current, bat.Full)
-		if bat.State.Raw >= 0 {
-			batteryState = uint8(bat.State.Raw)
+		if a.System != b.System {
+			return a.System
 		}
-	}
-
-	if totalCapacity == 0 || batteryState == math.MaxUint8 {
-		// for macs there's sometimes a ghost battery with 0 capacity
-		// https://github.com/distatus/battery/issues/34
-		// Instead of skipping over those batteries, we'll check for total 0 capacity
-		// and return an error. This also prevents a divide by zero.
-		return batteryPercent, batteryState, errors.New("no battery capacity")
-	}
-
-	batteryPercent = uint8(totalCharge / totalCapacity * 100)
-	return batteryPercent, batteryState, nil
+		return a.Name < b.Name
+	})
+	return ordered[0], true
 }

agent/battery/battery_darwin.go

@@ -0,0 +1,76 @@
+//go:build darwin
+
+package battery
+
+import (
+	"os/exec"
+
+	"howett.net/plist"
+)
+
+type macBattery struct {
+	CurrentCapacity   int  `plist:"CurrentCapacity"`
+	MaxCapacity       int  `plist:"MaxCapacity"`
+	FullyCharged      bool `plist:"FullyCharged"`
+	IsCharging        bool `plist:"IsCharging"`
+	ExternalConnected bool `plist:"ExternalConnected"`
+}
+
+func readMacBatteries() ([]macBattery, error) {
+	out, err := exec.Command("ioreg", "-n", "AppleSmartBattery", "-r", "-a").Output()
+	if err != nil {
+		return nil, err
+	}
+	if len(out) == 0 {
+		return nil, nil
+	}
+	var batteries []macBattery
+	if _, err := plist.Unmarshal(out, &batteries); err != nil {
+		return nil, err
+	}
+	return batteries, nil
+}
+
+func HasReadableBattery() bool {
+	batteries, _ := GetBatteryStats()
+	return len(batteries) > 0
+}
+
+// GetBatteryStats returns every readable battery reported by macOS.
+func GetBatteryStats() ([]Battery, error) {
+	batteries, err := readMacBatteries()
+	if err != nil {
+		return nil, err
+	}
+	if len(batteries) == 0 {
+		return nil, errNoBatteries
+	}
+	result := make([]Battery, 0, len(batteries))
+	for _, bat := range batteries {
+		if bat.MaxCapacity <= 0 {
+			// skip ghost batteries with 0 capacity
+			// https://github.com/distatus/battery/issues/34
+			continue
+		}
+		percent := min(max(float64(bat.CurrentCapacity)/float64(bat.MaxCapacity)*100, 0), 100)
+		state := stateUnknown
+		switch {
+		case !bat.ExternalConnected:
+			state = stateDischarging
+		case bat.IsCharging:
+			state = stateCharging
+		case bat.CurrentCapacity == 0:
+			state = stateEmpty
+		case !bat.FullyCharged:
+			state = stateIdle
+		default:
+			state = stateFull
+		}
+		result = append(result, Battery{Name: "Primary", Percent: uint8(percent), State: state,
+			FullChargeCapacity: uint64(bat.MaxCapacity), HasFullChargeCapacity: true, System: true})
+	}
+	if len(result) == 0 {
+		return nil, errNoBatteries
+	}
+	return normalizeBatteries(result), nil
+}

agent/battery/battery_freebsd.go

@@ -1,13 +0,0 @@
-//go:build freebsd
-
-package battery
-
-import "errors"
-
-func HasReadableBattery() bool {
-	return false
-}
-
-func GetBatteryStats() (uint8, uint8, error) {
-	return 0, 0, errors.ErrUnsupported
-}

agent/battery/battery_linux.go

@@ -0,0 +1,85 @@
+//go:build linux
+
+package battery
+
+import (
+	"os"
+	"path/filepath"
+	"strconv"
+
+	"github.com/henrygd/beszel/agent/utils"
+)
+
+var batteryRoot = "/sys/class/power_supply"
+
+// HasReadableBattery reports whether collection currently finds a readable battery.
+func HasReadableBattery() bool {
+	batteries, _ := GetBatteryStats()
+	return len(batteries) > 0
+}
+
+func parseSysfsState(status string) uint8 {
+	switch status {
+	case "Empty":
+		return stateEmpty
+	case "Full":
+		return stateFull
+	case "Charging":
+		return stateCharging
+	case "Discharging":
+		return stateDischarging
+	case "Not charging":
+		return stateIdle
+	default:
+		return stateUnknown
+	}
+}
+
+// GetBatteryStats re-enumerates power supplies and returns every readable battery.
+func GetBatteryStats() ([]Battery, error) {
+	entries, err := os.ReadDir(batteryRoot)
+	if err != nil {
+		return nil, err
+	}
+	batteries := make([]Battery, 0, len(entries))
+	for _, entry := range entries {
+		path := filepath.Join(batteryRoot, entry.Name())
+		if utils.ReadStringFile(filepath.Join(path, "type")) != "Battery" {
+			continue
+		}
+		capStr, ok := utils.ReadStringFileOK(filepath.Join(path, "capacity"))
+		if !ok {
+			continue
+		}
+		cap, parseErr := strconv.Atoi(capStr)
+		if parseErr != nil {
+			continue
+		}
+		cap = min(max(cap, 0), 100)
+		name := utils.ReadStringFile(filepath.Join(path, "model_name"))
+		if name == "" {
+			name = utils.ReadStringFile(filepath.Join(path, "model"))
+		}
+		if name == "" {
+			name = entry.Name()
+		}
+		battery := Battery{
+			Name:    name,
+			Percent: uint8(cap),
+			State:   parseSysfsState(utils.ReadStringFile(filepath.Join(path, "status"))),
+			System:  utils.ReadStringFile(filepath.Join(path, "scope")) != "Device",
+		}
+		for _, fullName := range []string{"charge_full", "energy_full"} {
+			if parsed, ok := utils.ReadUintFile(filepath.Join(path, fullName)); ok && parsed > 0 {
+				battery.FullChargeCapacity = parsed
+				battery.HasFullChargeCapacity = true
+				break
+			}
+		}
+		batteries = append(batteries, battery)
+	}
+	if len(batteries) == 0 {
+		return nil, errNoBatteries
+	}
+	return normalizeBatteries(batteries), nil
+}

agent/battery/battery_linux_test.go

@@ -0,0 +1,109 @@
+//go:build testing && linux
+
+package battery
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+type fakeBattery struct{ id, name, capacity, status, full, scope string }
+
+func setupFakeSysfs(t *testing.T) (string, func(fakeBattery)) {
+	t.Helper()
+	root := t.TempDir()
+	previousRoot := batteryRoot
+	batteryRoot = root
+	t.Cleanup(func() { batteryRoot = previousRoot })
+	write := func(path, value string) {
+		t.Helper()
+		require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755))
+		require.NoError(t, os.WriteFile(path, []byte(value), 0o644))
+	}
+	add := func(b fakeBattery) {
+		t.Helper()
+		dir := filepath.Join(root, b.id)
+		write(filepath.Join(dir, "type"), "Battery")
+		if b.capacity != "" {
+			write(filepath.Join(dir, "capacity"), b.capacity)
+		}
+		write(filepath.Join(dir, "status"), b.status)
+		if b.name != "" {
+			write(filepath.Join(dir, "model_name"), b.name)
+		}
+		if b.full != "" {
+			write(filepath.Join(dir, "energy_full"), b.full)
+		}
+		if b.scope != "" {
+			write(filepath.Join(dir, "scope"), b.scope)
+		}
+	}
+	return root, add
+}
+
+func TestParseSysfsState(t *testing.T) {
+	assert.Equal(t, stateEmpty, parseSysfsState("Empty"))
+	assert.Equal(t, stateFull, parseSysfsState("Full"))
+	assert.Equal(t, stateCharging, parseSysfsState("Charging"))
+	assert.Equal(t, stateDischarging, parseSysfsState("Discharging"))
+	assert.Equal(t, stateIdle, parseSysfsState("Not charging"))
+	assert.Equal(t, stateUnknown, parseSysfsState("SomethingElse"))
+}
+
+func TestGetBatteryStatsMultipleNamedAndPrimary(t *testing.T) {
+	_, add := setupFakeSysfs(t)
+	add(fakeBattery{id: "BAT0", name: "Primary", capacity: "105", status: "Charging", full: "5000", scope: "System"})
+	add(fakeBattery{id: "hidpp_battery_0", name: "MX Keys S", capacity: "55", status: "Unknown", full: "900", scope: "Device"})
+	batteries, err := GetBatteryStats()
+	require.NoError(t, err)
+	require.Len(t, batteries, 2)
+	assert.Equal(t, "Primary", batteries[0].Name)
+	assert.Equal(t, uint8(100), batteries[0].Percent)
+	assert.Equal(t, stateUnknown, batteries[1].State)
+	primary, ok := Primary(batteries)
+	require.True(t, ok)
+	assert.Equal(t, "Primary", primary.Name)
+}
+
+func TestGetBatteryStatsFallbackDuplicatesAndUnreadable(t *testing.T) {
+	root, add := setupFakeSysfs(t)
+	add(fakeBattery{id: "BAT0", name: "Keyboard", capacity: "80", status: "Discharging"})
+	add(fakeBattery{id: "BAT1", name: "Keyboard", capacity: "-4", status: "SomethingWeird"})
+	add(fakeBattery{id: "BAT2", capacity: "not-a-number", status: "Charging"})
+	add(fakeBattery{id: "BAT3", capacity: "42", status: "Full"})
+	ac := filepath.Join(root, "AC0")
+	require.NoError(t, os.MkdirAll(ac, 0o755))
+	require.NoError(t, os.WriteFile(filepath.Join(ac, "type"), []byte("Mains"), 0o644))
+	batteries, err := GetBatteryStats()
+	require.NoError(t, err)
+	require.Len(t, batteries, 3)
+	assert.Equal(t, "Keyboard", batteries[0].Name)
+	assert.Equal(t, "Keyboard (2)", batteries[1].Name)
+	assert.Equal(t, uint8(0), batteries[1].Percent)
+	assert.Equal(t, "BAT3", batteries[2].Name)
+}
+
+func TestGetBatteryStatsHotPlugReenumerates(t *testing.T) {
+	_, add := setupFakeSysfs(t)
+	_, err := GetBatteryStats()
+	assert.Error(t, err)
+	assert.False(t, HasReadableBattery())
+	add(fakeBattery{id: "BAT0", capacity: "64", status: "Discharging"})
+	batteries, err := GetBatteryStats()
+	require.NoError(t, err)
+	assert.True(t, HasReadableBattery())
+	require.Len(t, batteries, 1)
+	assert.Equal(t, uint8(64), batteries[0].Percent)
+}
+
+func TestGetBatteryStatsNoReadableCapacity(t *testing.T) {
+	_, add := setupFakeSysfs(t)
+	add(fakeBattery{id: "BAT0", status: "Charging"})
+	_, err := GetBatteryStats()
+	assert.Error(t, err)
+	assert.False(t, HasReadableBattery())
+}

agent/battery/battery_stub.go

@@ -0,0 +1,13 @@
+//go:build !darwin && !linux && !windows
+
+package battery
+
+import "errors"
+
+func HasReadableBattery() bool {
+	return false
+}
+
+func GetBatteryStats() ([]Battery, error) {
+	return nil, errors.ErrUnsupported
+}

agent/battery/battery_test.go

@@ -0,0 +1,35 @@
+package battery
+
+import (
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestPrimarySelection(t *testing.T) {
+	tests := []struct {
+		name string
+		bats []Battery
+		want string
+	}{
+		{"largest reported capacity", []Battery{{Name: "Small", FullChargeCapacity: 20, HasFullChargeCapacity: true, System: true}, {Name: "Large", FullChargeCapacity: 80, HasFullChargeCapacity: true}}, "Large"},
+		{"reported ranks over missing", []Battery{{Name: "Unknown", System: true}, {Name: "Known", FullChargeCapacity: 1, HasFullChargeCapacity: true}}, "Known"},
+		{"system wins capacity tie", []Battery{{Name: "Peripheral", FullChargeCapacity: 50, HasFullChargeCapacity: true}, {Name: "System", FullChargeCapacity: 50, HasFullChargeCapacity: true, System: true}}, "System"},
+		{"name resolves final tie", []Battery{{Name: "Zed"}, {Name: "Alpha"}}, "Alpha"},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got, ok := Primary(tt.bats)
+			require.True(t, ok)
+			assert.Equal(t, tt.want, got.Name)
+		})
+	}
+	_, ok := Primary(nil)
+	assert.False(t, ok)
+}
+
+func TestNormalizeBatteriesFallbackNames(t *testing.T) {
+	bats := normalizeBatteries([]Battery{{}, {}, {Name: "Mouse"}, {Name: "Mouse"}})
+	assert.Equal(t, []string{"Battery 1", "Battery 2", "Mouse", "Mouse (2)"}, []string{bats[0].Name, bats[1].Name, bats[2].Name, bats[3].Name})
+}

agent/battery/battery_windows.go

@@ -0,0 +1,291 @@
+//go:build windows
+
+// Most of the Windows battery code is based on
+// distatus/battery by Karol 'Kenji Takahashi' Woźniak
+
+package battery
+
+import (
+	"errors"
+	"syscall"
+	"unsafe"
+
+	"golang.org/x/sys/windows"
+)
+
+type batteryQueryInformation struct {
+	BatteryTag       uint32
+	InformationLevel int32
+	AtRate           int32
+}
+
+type batteryInformation struct {
+	Capabilities        uint32
+	Technology          uint8
+	Reserved            [3]uint8
+	Chemistry           [4]uint8
+	DesignedCapacity    uint32
+	FullChargedCapacity uint32
+	DefaultAlert1       uint32
+	DefaultAlert2       uint32
+	CriticalBias        uint32
+	CycleCount          uint32
+}
+
+type batteryWaitStatus struct {
+	BatteryTag   uint32
+	Timeout      uint32
+	PowerState   uint32
+	LowCapacity  uint32
+	HighCapacity uint32
+}
+
+type batteryStatus struct {
+	PowerState uint32
+	Capacity   uint32
+	Voltage    uint32
+	Rate       int32
+}
+
+type winGUID struct {
+	Data1 uint32
+	Data2 uint16
+	Data3 uint16
+	Data4 [8]byte
+}
+
+type spDeviceInterfaceData struct {
+	cbSize             uint32
+	InterfaceClassGuid winGUID
+	Flags              uint32
+	Reserved           uint
+}
+
+var guidDeviceBattery = winGUID{
+	0x72631e54,
+	0x78A4,
+	0x11d0,
+	[8]byte{0xbc, 0xf7, 0x00, 0xaa, 0x00, 0xb7, 0xb3, 0x2a},
+}
+
+var (
+	setupapi                         = &windows.LazyDLL{Name: "setupapi.dll", System: true}
+	setupDiGetClassDevsW             = setupapi.NewProc("SetupDiGetClassDevsW")
+	setupDiEnumDeviceInterfaces      = setupapi.NewProc("SetupDiEnumDeviceInterfaces")
+	setupDiGetDeviceInterfaceDetailW = setupapi.NewProc("SetupDiGetDeviceInterfaceDetailW")
+	setupDiDestroyDeviceInfoList     = setupapi.NewProc("SetupDiDestroyDeviceInfoList")
+)
+
+// winBatteryGet reads one battery by index.
+// Returns error == errNotFound when there are no more batteries.
+var errNotFound = errors.New("no more batteries")
+
+func setupDiSetup(proc *windows.LazyProc, nargs, a1, a2, a3, a4, a5, a6 uintptr) (uintptr, error) {
+	_ = nargs
+	r1, _, errno := syscall.SyscallN(proc.Addr(), a1, a2, a3, a4, a5, a6)
+	if windows.Handle(r1) == windows.InvalidHandle {
+		if errno != 0 {
+			return 0, error(errno)
+		}
+		return 0, syscall.EINVAL
+	}
+	return r1, nil
+}
+
+func setupDiCall(proc *windows.LazyProc, nargs, a1, a2, a3, a4, a5, a6 uintptr) syscall.Errno {
+	_ = nargs
+	r1, _, errno := syscall.SyscallN(proc.Addr(), a1, a2, a3, a4, a5, a6)
+	if r1 == 0 {
+		if errno != 0 {
+			return errno
+		}
+		return syscall.EINVAL
+	}
+	return 0
+}
+
+func readWinBatteryState(powerState uint32) uint8 {
+	switch {
+	case powerState&0x00000004 != 0:
+		return stateCharging
+	case powerState&0x00000008 != 0:
+		return stateEmpty
+	case powerState&0x00000002 != 0:
+		return stateDischarging
+	case powerState&0x00000001 != 0:
+		return stateFull
+	default:
+		return stateUnknown
+	}
+}
+
+func winBatteryGet(idx int) (Battery, error) {
+	hdev, err := setupDiSetup(
+		setupDiGetClassDevsW,
+		4,
+		uintptr(unsafe.Pointer(&guidDeviceBattery)),
+		0, 0,
+		2|16, // DIGCF_PRESENT|DIGCF_DEVICEINTERFACE
+		0, 0,
+	)
+	if err != nil {
+		return Battery{}, err
+	}
+	defer syscall.SyscallN(setupDiDestroyDeviceInfoList.Addr(), hdev)
+
+	var did spDeviceInterfaceData
+	did.cbSize = uint32(unsafe.Sizeof(did))
+	errno := setupDiCall(
+		setupDiEnumDeviceInterfaces,
+		5,
+		hdev, 0,
+		uintptr(unsafe.Pointer(&guidDeviceBattery)),
+		uintptr(idx),
+		uintptr(unsafe.Pointer(&did)),
+		0,
+	)
+	if errno == 259 { // ERROR_NO_MORE_ITEMS
+		return Battery{}, errNotFound
+	}
+	if errno != 0 {
+		return Battery{}, errno
+	}
+
+	var cbRequired uint32
+	errno = setupDiCall(
+		setupDiGetDeviceInterfaceDetailW,
+		6,
+		hdev,
+		uintptr(unsafe.Pointer(&did)),
+		0, 0,
+		uintptr(unsafe.Pointer(&cbRequired)),
+		0,
+	)
+	if errno != 0 && errno != 122 { // ERROR_INSUFFICIENT_BUFFER
+		return Battery{}, errno
+	}
+	didd := make([]uint16, cbRequired/2)
+	cbSize := (*uint32)(unsafe.Pointer(&didd[0]))
+	if unsafe.Sizeof(uint(0)) == 8 {
+		*cbSize = 8
+	} else {
+		*cbSize = 6
+	}
+	errno = setupDiCall(
+		setupDiGetDeviceInterfaceDetailW,
+		6,
+		hdev,
+		uintptr(unsafe.Pointer(&did)),
+		uintptr(unsafe.Pointer(&didd[0])),
+		uintptr(cbRequired),
+		uintptr(unsafe.Pointer(&cbRequired)),
+		0,
+	)
+	if errno != 0 {
+		return Battery{}, errno
+	}
+	devicePath := &didd[2:][0]
+
+	handle, err := windows.CreateFile(
+		devicePath,
+		windows.GENERIC_READ|windows.GENERIC_WRITE,
+		windows.FILE_SHARE_READ|windows.FILE_SHARE_WRITE,
+		nil,
+		windows.OPEN_EXISTING,
+		windows.FILE_ATTRIBUTE_NORMAL,
+		0,
+	)
+	if err != nil {
+		return Battery{}, err
+	}
+	defer windows.CloseHandle(handle)
+
+	var dwOut uint32
+	var dwWait uint32
+	var bqi batteryQueryInformation
+	err = windows.DeviceIoControl(
+		handle,
+		2703424, // IOCTL_BATTERY_QUERY_TAG
+		(*byte)(unsafe.Pointer(&dwWait)),
+		uint32(unsafe.Sizeof(dwWait)),
+		(*byte)(unsafe.Pointer(&bqi.BatteryTag)),
+		uint32(unsafe.Sizeof(bqi.BatteryTag)),
+		&dwOut, nil,
+	)
+	if err != nil || bqi.BatteryTag == 0 {
+		return Battery{}, errors.New("battery tag not returned")
+	}
+
+	var bi batteryInformation
+	if err = windows.DeviceIoControl(
+		handle,
+		2703428, // IOCTL_BATTERY_QUERY_INFORMATION
+		(*byte)(unsafe.Pointer(&bqi)),
+		uint32(unsafe.Sizeof(bqi)),
+		(*byte)(unsafe.Pointer(&bi)),
+		uint32(unsafe.Sizeof(bi)),
+		&dwOut, nil,
+	); err != nil {
+		return Battery{}, err
+	}
+
+	// BatteryDeviceName is optional, so retain the deterministic fallback on error.
+	name := ""
+	nameQuery := bqi
+	nameQuery.InformationLevel = 4 // BatteryDeviceName
+	nameBuffer := make([]uint16, 128)
+	if err := windows.DeviceIoControl(
+		handle, 2703428,
+		(*byte)(unsafe.Pointer(&nameQuery)), uint32(unsafe.Sizeof(nameQuery)),
+		(*byte)(unsafe.Pointer(&nameBuffer[0])), uint32(len(nameBuffer)*2),
+		&dwOut, nil,
+	); err == nil {
+		name = windows.UTF16ToString(nameBuffer)
+	}
+
+	bws := batteryWaitStatus{BatteryTag: bqi.BatteryTag}
+	var bs batteryStatus
+	if err = windows.DeviceIoControl(
+		handle,
+		2703436, // IOCTL_BATTERY_QUERY_STATUS
+		(*byte)(unsafe.Pointer(&bws)),
+		uint32(unsafe.Sizeof(bws)),
+		(*byte)(unsafe.Pointer(&bs)),
+		uint32(unsafe.Sizeof(bs)),
+		&dwOut, nil,
+	); err != nil {
+		return Battery{}, err
+	}
+
+	if bs.Capacity == 0xffffffff || bi.FullChargedCapacity == 0 || bi.FullChargedCapacity == 0xffffffff {
+		return Battery{}, errors.New("battery capacity unknown")
+	}
+	percent := min(float64(bs.Capacity)/float64(bi.FullChargedCapacity)*100, 100)
+	return Battery{Name: name, Percent: uint8(percent), State: readWinBatteryState(bs.PowerState),
+		FullChargeCapacity: uint64(bi.FullChargedCapacity), HasFullChargeCapacity: true, System: true}, nil
+}
+
+// HasReadableBattery checks if the system has a battery and returns true if it does.
+func HasReadableBattery() bool {
+	batteries, _ := GetBatteryStats()
+	return len(batteries) > 0
+}
+
+// GetBatteryStats returns every readable battery reported by Windows.
+func GetBatteryStats() ([]Battery, error) {
+	batteries := make([]Battery, 0, 2)
+	for i := 0; ; i++ {
+		battery, bErr := winBatteryGet(i)
+		if errors.Is(bErr, errNotFound) {
+			break
+		}
+		if bErr != nil {
+			continue
+		}
+		batteries = append(batteries, battery)
+	}
+	if len(batteries) == 0 {
+		return nil, errNoBatteries
+	}
+	return normalizeBatteries(batteries), nil
+}

agent/client.go

@@ -20,6 +20,7 @@ import (
 	"github.com/fxamacker/cbor/v2"
 	"github.com/lxzan/gws"
 	"golang.org/x/crypto/ssh"
+	"golang.org/x/net/proxy"
 )
 
 const (
@@ -52,8 +53,8 @@ func newWebSocketClient(agent *Agent) (client *WebSocketClient, err error) {
 	client = &WebSocketClient{}
 
 	client.hubURL, err = url.Parse(hubURLStr)
-	if err != nil {
-		return nil, errors.New("invalid hub URL")
+	if err != nil || client.hubURL.Host == "" {
+		return nil, fmt.Errorf("invalid HUB_URL %q: must include scheme and host (e.g. http://hub.example.com:8090)", hubURLStr)
 	}
 	// get registration token
 	client.token, err = getToken()
@@ -104,6 +105,11 @@ func (client *WebSocketClient) getOptions() *gws.ClientOption {
 	}
 	client.hubURL.Path = path.Join(client.hubURL.Path, "api/beszel/agent-connect")
 
+	// make sure BESZEL_AGENT_ALL_PROXY works (GWS only checks ALL_PROXY)
+	if val := os.Getenv("BESZEL_AGENT_ALL_PROXY"); val != "" {
+		os.Setenv("ALL_PROXY", val)
+	}
+
 	client.options = &gws.ClientOption{
 		Addr:      client.hubURL.String(),
 		TlsConfig: &tls.Config{InsecureSkipVerify: true},
@@ -112,6 +118,9 @@ func (client *WebSocketClient) getOptions() *gws.ClientOption {
 			"X-Token":    []string{client.token},
 			"X-Beszel":   []string{beszel.Version},
 		},
+		NewDialer: func() (gws.Dialer, error) {
+			return proxy.FromEnvironment(), nil
+		},
 	}
 	return client.options
 }

agent/client_test.go

@@ -51,11 +51,18 @@ func TestNewWebSocketClient(t *testing.T) {
 			errorMsg:    "HUB_URL environment variable not set",
 		},
 		{
-			name:        "invalid URL",
+			name:        "malformed URL",
 			hubURL:      "ht\ttp://invalid",
 			token:       "test-token",
 			expectError: true,
-			errorMsg:    "invalid hub URL",
+			errorMsg:    "invalid HUB_URL",
+		},
+		{
+			name:        "URL without host",
+			hubURL:      "http:/api",
+			token:       "test-token",
+			expectError: true,
+			errorMsg:    "invalid HUB_URL",
 		},
 		{
 			name:        "missing token",
@@ -70,19 +77,11 @@ func TestNewWebSocketClient(t *testing.T) {
 		t.Run(tc.name, func(t *testing.T) {
 			// Set up environment
 			if tc.hubURL != "" {
-				os.Setenv("BESZEL_AGENT_HUB_URL", tc.hubURL)
-			} else {
-				os.Unsetenv("BESZEL_AGENT_HUB_URL")
+				t.Setenv("BESZEL_AGENT_HUB_URL", tc.hubURL)
 			}
 			if tc.token != "" {
-				os.Setenv("BESZEL_AGENT_TOKEN", tc.token)
-			} else {
-				os.Unsetenv("BESZEL_AGENT_TOKEN")
+				t.Setenv("BESZEL_AGENT_TOKEN", tc.token)
 			}
-			defer func() {
-				os.Unsetenv("BESZEL_AGENT_HUB_URL")
-				os.Unsetenv("BESZEL_AGENT_TOKEN")
-			}()
 
 			client, err := newWebSocketClient(agent)
 
@@ -138,12 +137,8 @@ func TestWebSocketClient_GetOptions(t *testing.T) {
 	for _, tc := range testCases {
 		t.Run(tc.name, func(t *testing.T) {
 			// Set up environment
-			os.Setenv("BESZEL_AGENT_HUB_URL", tc.inputURL)
-			os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-			defer func() {
-				os.Unsetenv("BESZEL_AGENT_HUB_URL")
-				os.Unsetenv("BESZEL_AGENT_TOKEN")
-			}()
+			t.Setenv("BESZEL_AGENT_HUB_URL", tc.inputURL)
+			t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 			client, err := newWebSocketClient(agent)
 			require.NoError(t, err)
@@ -185,12 +180,8 @@ func TestWebSocketClient_VerifySignature(t *testing.T) {
 	require.NoError(t, err)
 
 	// Set up environment
-	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	client, err := newWebSocketClient(agent)
 	require.NoError(t, err)
@@ -258,12 +249,8 @@ func TestWebSocketClient_HandleHubRequest(t *testing.T) {
 	agent := createTestAgent(t)
 
 	// Set up environment
-	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	client, err := newWebSocketClient(agent)
 	require.NoError(t, err)
@@ -350,13 +337,8 @@ func TestGetUserAgent(t *testing.T) {
 func TestWebSocketClient_Close(t *testing.T) {
 	agent := createTestAgent(t)
 
-	// Set up environment
-	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	client, err := newWebSocketClient(agent)
 	require.NoError(t, err)
@@ -371,13 +353,8 @@ func TestWebSocketClient_Close(t *testing.T) {
 func TestWebSocketClient_ConnectRateLimit(t *testing.T) {
 	agent := createTestAgent(t)
 
-	// Set up environment
-	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	client, err := newWebSocketClient(agent)
 	require.NoError(t, err)
@@ -393,20 +370,10 @@ func TestWebSocketClient_ConnectRateLimit(t *testing.T) {
 
 // TestGetToken tests the getToken function with various scenarios
 func TestGetToken(t *testing.T) {
-	unsetEnvVars := func() {
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-		os.Unsetenv("TOKEN")
-		os.Unsetenv("BESZEL_AGENT_TOKEN_FILE")
-		os.Unsetenv("TOKEN_FILE")
-	}
-
 	t.Run("token from TOKEN environment variable", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Set TOKEN env var
 		expectedToken := "test-token-from-env"
-		os.Setenv("TOKEN", expectedToken)
-		defer os.Unsetenv("TOKEN")
+		t.Setenv("TOKEN", expectedToken)
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -414,12 +381,9 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("token from BESZEL_AGENT_TOKEN environment variable", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Set BESZEL_AGENT_TOKEN env var (should take precedence)
 		expectedToken := "test-token-from-beszel-env"
-		os.Setenv("BESZEL_AGENT_TOKEN", expectedToken)
-		defer os.Unsetenv("BESZEL_AGENT_TOKEN")
+		t.Setenv("BESZEL_AGENT_TOKEN", expectedToken)
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -427,8 +391,6 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("token from TOKEN_FILE", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Create a temporary token file
 		expectedToken := "test-token-from-file"
 		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
@@ -440,8 +402,7 @@ func TestGetToken(t *testing.T) {
 		tokenFile.Close()
 
 		// Set TOKEN_FILE env var
-		os.Setenv("TOKEN_FILE", tokenFile.Name())
-		defer os.Unsetenv("TOKEN_FILE")
+		t.Setenv("TOKEN_FILE", tokenFile.Name())
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -449,8 +410,6 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("token from BESZEL_AGENT_TOKEN_FILE", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Create a temporary token file
 		expectedToken := "test-token-from-beszel-file"
 		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
@@ -462,8 +421,7 @@ func TestGetToken(t *testing.T) {
 		tokenFile.Close()
 
 		// Set BESZEL_AGENT_TOKEN_FILE env var (should take precedence)
-		os.Setenv("BESZEL_AGENT_TOKEN_FILE", tokenFile.Name())
-		defer os.Unsetenv("BESZEL_AGENT_TOKEN_FILE")
+		t.Setenv("BESZEL_AGENT_TOKEN_FILE", tokenFile.Name())
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -471,8 +429,6 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("TOKEN takes precedence over TOKEN_FILE", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Create a temporary token file
 		fileToken := "token-from-file"
 		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
@@ -485,12 +441,8 @@ func TestGetToken(t *testing.T) {
 
 		// Set both TOKEN and TOKEN_FILE
 		envToken := "token-from-env"
-		os.Setenv("TOKEN", envToken)
-		os.Setenv("TOKEN_FILE", tokenFile.Name())
-		defer func() {
-			os.Unsetenv("TOKEN")
-			os.Unsetenv("TOKEN_FILE")
-		}()
+		t.Setenv("TOKEN", envToken)
+		t.Setenv("TOKEN_FILE", tokenFile.Name())
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -498,7 +450,10 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("error when neither TOKEN nor TOKEN_FILE is set", func(t *testing.T) {
-		unsetEnvVars()
+		t.Setenv("BESZEL_AGENT_TOKEN", "")
+		t.Setenv("TOKEN", "")
+		t.Setenv("BESZEL_AGENT_TOKEN_FILE", "")
+		t.Setenv("TOKEN_FILE", "")
 
 		token, err := getToken()
 		assert.Error(t, err)
@@ -507,11 +462,8 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("error when TOKEN_FILE points to non-existent file", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Set TOKEN_FILE to a non-existent file
-		os.Setenv("TOKEN_FILE", "/non/existent/file.txt")
-		defer os.Unsetenv("TOKEN_FILE")
+		t.Setenv("TOKEN_FILE", "/non/existent/file.txt")
 
 		token, err := getToken()
 		assert.Error(t, err)
@@ -520,8 +472,6 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("handles empty token file", func(t *testing.T) {
-		unsetEnvVars()
-
 		// Create an empty token file
 		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
 		require.NoError(t, err)
@@ -529,8 +479,7 @@ func TestGetToken(t *testing.T) {
 		tokenFile.Close()
 
 		// Set TOKEN_FILE env var
-		os.Setenv("TOKEN_FILE", tokenFile.Name())
-		defer os.Unsetenv("TOKEN_FILE")
+		t.Setenv("TOKEN_FILE", tokenFile.Name())
 
 		token, err := getToken()
 		assert.NoError(t, err)
@@ -538,8 +487,6 @@ func TestGetToken(t *testing.T) {
 	})
 
 	t.Run("strips whitespace from TOKEN_FILE", func(t *testing.T) {
-		unsetEnvVars()
-
 		tokenWithWhitespace := "  test-token-with-whitespace  \n\t"
 		expectedToken := "test-token-with-whitespace"
 		tokenFile, err := os.CreateTemp("", "token-test-*.txt")
@@ -550,8 +497,7 @@ func TestGetToken(t *testing.T) {
 		require.NoError(t, err)
 		tokenFile.Close()
 
-		os.Setenv("TOKEN_FILE", tokenFile.Name())
-		defer os.Unsetenv("TOKEN_FILE")
+		t.Setenv("TOKEN_FILE", tokenFile.Name())
 
 		token, err := getToken()
 		assert.NoError(t, err)

agent/connection_manager.go

@@ -4,11 +4,15 @@ import (
 	"context"
 	"errors"
 	"log/slog"
+	"net"
+	"os"
 	"os/signal"
+	"strings"
 	"syscall"
 	"time"
 
 	"github.com/henrygd/beszel/agent/health"
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
@@ -111,13 +115,36 @@ func (c *ConnectionManager) Start(serverOptions ServerOptions) error {
 			_ = health.Update()
 		case <-sigCtx.Done():
 			slog.Info("Shutting down", "cause", context.Cause(sigCtx))
-			_ = c.agent.StopServer()
-			c.closeWebSocket()
-			return health.CleanUp()
+			return c.stop()
 		}
 	}
 }
 
+// stop does not stop the connection manager itself, just any active connections. The manager will attempt to reconnect after stopping, so this should only be called immediately before shutting down the entire agent.
+//
+// If we need or want to expose a graceful Stop method in the future, do something like this to actually stop the manager:
+//
+//	func (c *ConnectionManager) Start(serverOptions ServerOptions) error {
+//		ctx, cancel := context.WithCancel(context.Background())
+//		c.cancel = cancel
+//
+//		for {
+//			select {
+//			case <-ctx.Done():
+//				return c.stop()
+//			}
+//		}
+//	}
+//
+//	func (c *ConnectionManager) Stop() {
+//		c.cancel()
+//	}
+func (c *ConnectionManager) stop() error {
+	_ = c.agent.StopServer()
+	c.closeWebSocket()
+	return health.CleanUp()
+}
+
 // handleEvent processes connection events and updates the connection state accordingly.
 func (c *ConnectionManager) handleEvent(event ConnectionEvent) {
 	switch event {
@@ -185,9 +212,16 @@ func (c *ConnectionManager) connect() {
 
 	// Try WebSocket first, if it fails, start SSH server
 	err := c.startWebSocketConnection()
-	if err != nil && c.State == Disconnected {
-		c.startSSHServer()
-		c.startWsTicker()
+	if err != nil {
+		if shouldExitOnErr(err) {
+			time.Sleep(2 * time.Second) // prevent tight restart loop
+			_ = c.stop()
+			os.Exit(1)
+		}
+		if c.State == Disconnected {
+			c.startSSHServer()
+			c.startWsTicker()
+		}
 	}
 }
 
@@ -224,3 +258,14 @@ func (c *ConnectionManager) closeWebSocket() {
 		c.wsClient.Close()
 	}
 }
+
+// shouldExitOnErr checks if the error is a DNS resolution failure and if the
+// EXIT_ON_DNS_ERROR env var is set. https://github.com/henrygd/beszel/issues/1924.
+func shouldExitOnErr(err error) bool {
+	if val, _ := utils.GetEnv("EXIT_ON_DNS_ERROR"); val == "true" {
+		if opErr, ok := errors.AsType[*net.OpError](err); ok {
+			return strings.Contains(opErr.Err.Error(), "lookup")
+		}
+	}
+	return false
+}

agent/connection_manager_test.go

@@ -4,10 +4,10 @@ package agent
 
 import (
 	"crypto/ed25519"
+	"errors"
 	"fmt"
 	"net"
 	"net/url"
-	"os"
 	"testing"
 	"time"
 
@@ -183,10 +183,6 @@ func TestConnectionManager_TickerManagement(t *testing.T) {
 
 // TestConnectionManager_WebSocketConnectionFlow tests WebSocket connection logic
 func TestConnectionManager_WebSocketConnectionFlow(t *testing.T) {
-	if testing.Short() {
-		t.Skip("Skipping WebSocket connection test in short mode")
-	}
-
 	agent := createTestAgent(t)
 	cm := agent.connectionManager
 
@@ -196,19 +192,18 @@ func TestConnectionManager_WebSocketConnectionFlow(t *testing.T) {
 	assert.Equal(t, Disconnected, cm.State, "State should remain Disconnected after failed connection")
 
 	// Test with invalid URL
-	os.Setenv("BESZEL_AGENT_HUB_URL", "invalid-url")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
-
-	// Test with missing token
-	os.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
-	os.Unsetenv("BESZEL_AGENT_TOKEN")
+	t.Setenv("BESZEL_AGENT_HUB_URL", "1,33%")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	_, err2 := newWebSocketClient(agent)
-	assert.Error(t, err2, "WebSocket client creation should fail without token")
+	assert.Error(t, err2, "WebSocket client creation should fail with invalid URL")
+
+	// Test with missing token
+	t.Setenv("BESZEL_AGENT_HUB_URL", "http://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "")
+
+	_, err3 := newWebSocketClient(agent)
+	assert.Error(t, err3, "WebSocket client creation should fail without token")
 }
 
 // TestConnectionManager_ReconnectionLogic tests reconnection prevention logic
@@ -234,12 +229,8 @@ func TestConnectionManager_ConnectWithRateLimit(t *testing.T) {
 	cm := agent.connectionManager
 
 	// Set up environment for WebSocket client creation
-	os.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	// Create WebSocket client
 	wsClient, err := newWebSocketClient(agent)
@@ -285,12 +276,8 @@ func TestConnectionManager_CloseWebSocket(t *testing.T) {
 	}, "Should not panic when closing nil WebSocket client")
 
 	// Set up environment and create WebSocket client
-	os.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
-	os.Setenv("BESZEL_AGENT_TOKEN", "test-token")
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", "ws://localhost:8080")
+	t.Setenv("BESZEL_AGENT_TOKEN", "test-token")
 
 	wsClient, err := newWebSocketClient(agent)
 	require.NoError(t, err)
@@ -312,3 +299,65 @@ func TestConnectionManager_ConnectFlow(t *testing.T) {
 		cm.connect()
 	}, "Connect should not panic without WebSocket client")
 }
+
+func TestShouldExitOnErr(t *testing.T) {
+	createDialErr := func(msg string) error {
+		return &net.OpError{
+			Op:  "dial",
+			Net: "tcp",
+			Err: errors.New(msg),
+		}
+	}
+
+	tests := []struct {
+		name     string
+		err      error
+		envValue string
+		expected bool
+	}{
+		{
+			name:     "no env var",
+			err:      createDialErr("lookup lkahsdfasdf: no such host"),
+			envValue: "",
+			expected: false,
+		},
+		{
+			name:     "env var false",
+			err:      createDialErr("lookup lkahsdfasdf: no such host"),
+			envValue: "false",
+			expected: false,
+		},
+		{
+			name:     "env var true, matching error",
+			err:      createDialErr("lookup lkahsdfasdf: no such host"),
+			envValue: "true",
+			expected: true,
+		},
+		{
+			name:     "env var true, matching error with extra context",
+			err:      createDialErr("lookup beszel.server.lan on [::1]:53: read udp [::1]:44557->[::1]:53: read: connection refused"),
+			envValue: "true",
+			expected: true,
+		},
+		{
+			name:     "env var true, non-matching error",
+			err:      errors.New("connection refused"),
+			envValue: "true",
+			expected: false,
+		},
+		{
+			name:     "env var true, dial but not lookup",
+			err:      createDialErr("connection timeout"),
+			envValue: "true",
+			expected: false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			t.Setenv("EXIT_ON_DNS_ERROR", tt.envValue)
+			result := shouldExitOnErr(tt.err)
+			assert.Equal(t, tt.expected, result)
+		})
+	}
+}

agent/data_dir_test.go

@@ -39,17 +39,7 @@ func TestGetDataDir(t *testing.T) {
 	t.Run("DATA_DIR environment variable", func(t *testing.T) {
 		tempDir := t.TempDir()
 
-		// Set environment variable
-		oldValue := os.Getenv("DATA_DIR")
-		defer func() {
-			if oldValue == "" {
-				os.Unsetenv("BESZEL_AGENT_DATA_DIR")
-			} else {
-				os.Setenv("BESZEL_AGENT_DATA_DIR", oldValue)
-			}
-		}()
-
-		os.Setenv("BESZEL_AGENT_DATA_DIR", tempDir)
+		t.Setenv("BESZEL_AGENT_DATA_DIR", tempDir)
 
 		result, err := GetDataDir()
 		require.NoError(t, err)
@@ -65,17 +55,6 @@ func TestGetDataDir(t *testing.T) {
 
 	// Test fallback behavior (empty dataDir, no env var)
 	t.Run("fallback to default directories", func(t *testing.T) {
-		// Clear DATA_DIR environment variable
-		oldValue := os.Getenv("DATA_DIR")
-		defer func() {
-			if oldValue == "" {
-				os.Unsetenv("DATA_DIR")
-			} else {
-				os.Setenv("DATA_DIR", oldValue)
-			}
-		}()
-		os.Unsetenv("DATA_DIR")
-
 		// This will try platform-specific defaults, which may or may not work
 		// We're mainly testing that it doesn't panic and returns some result
 		result, err := GetDataDir()

agent/disk.go

@@ -1,6 +1,7 @@
 package agent
 
 import (
+	"context"
 	"log/slog"
 	"os"
 	"path/filepath"
@@ -33,6 +34,34 @@ type diskDiscovery struct {
 	ctx            fsRegistrationContext
 }
 
+// prevDisk stores previous per-device disk counters for a given cache interval
+type prevDisk struct {
+	readBytes  uint64
+	writeBytes uint64
+	readTime   uint64 // cumulative ms spent on reads (from ReadTime)
+	writeTime  uint64 // cumulative ms spent on writes (from WriteTime)
+	ioTime     uint64 // cumulative ms spent doing I/O (from IoTime)
+	weightedIO uint64 // cumulative weighted ms (queue-depth × ms, from WeightedIO)
+	readCount  uint64 // cumulative read operation count
+	writeCount uint64 // cumulative write operation count
+	at         time.Time
+}
+
+// prevDiskFromCounter creates a prevDisk snapshot from a disk.IOCountersStat at time t.
+func prevDiskFromCounter(d disk.IOCountersStat, t time.Time) prevDisk {
+	return prevDisk{
+		readBytes:  d.ReadBytes,
+		writeBytes: d.WriteBytes,
+		readTime:   d.ReadTime,
+		writeTime:  d.WriteTime,
+		ioTime:     d.IoTime,
+		weightedIO: d.WeightedIO,
+		readCount:  d.ReadCount,
+		writeCount: d.WriteCount,
+		at:         t,
+	}
+}
+
 // parseFilesystemEntry parses a filesystem entry in the format "device__customname"
 // Returns the device/filesystem part and the custom name part
 func parseFilesystemEntry(entry string) (device, customName string) {
@@ -238,9 +267,11 @@ func (d *diskDiscovery) addConfiguredExtraFilesystems(extraFilesystems string) {
 
 // addPartitionExtraFs registers partitions mounted under /extra-filesystems so
 // their display names can come from the folder name while their I/O keys still
-// prefer the underlying partition device.
+// prefer the underlying partition device. Only direct children are matched to
+// avoid registering nested virtual mounts (e.g. /proc, /sys) that are returned by
+// disk.Partitions(true) when the host root is bind-mounted in /extra-filesystems.
 func (d *diskDiscovery) addPartitionExtraFs(p disk.PartitionStat) {
-	if !strings.HasPrefix(p.Mountpoint, d.ctx.efPath) {
+	if filepath.Dir(p.Mountpoint) != d.ctx.efPath {
 		return
 	}
 	device, customName := extraFilesystemPartitionInfo(p)
@@ -273,7 +304,7 @@ func (a *Agent) initializeDiskInfo() {
 	hasRoot := false
 	isWindows := runtime.GOOS == "windows"
 
-	partitions, err := disk.Partitions(false)
+	partitions, err := disk.PartitionsWithContext(context.Background(), true)
 	if err != nil {
 		slog.Error("Error getting disk partitions", "err", err)
 	}
@@ -578,16 +609,29 @@ func (a *Agent) updateDiskIo(cacheTimeMs uint16, systemStats *system.Stats) {
 			prev, hasPrev := a.diskPrev[cacheTimeMs][name]
 			if !hasPrev {
 				// Seed from agent-level fsStats if present, else seed from current
-				prev = prevDisk{readBytes: stats.TotalRead, writeBytes: stats.TotalWrite, at: stats.Time}
+				prev = prevDisk{
+					readBytes:  stats.TotalRead,
+					writeBytes: stats.TotalWrite,
+					readTime:   d.ReadTime,
+					writeTime:  d.WriteTime,
+					ioTime:     d.IoTime,
+					weightedIO: d.WeightedIO,
+					readCount:  d.ReadCount,
+					writeCount: d.WriteCount,
+					at:         stats.Time,
+				}
 				if prev.at.IsZero() {
-					prev = prevDisk{readBytes: d.ReadBytes, writeBytes: d.WriteBytes, at: now}
+					prev = prevDiskFromCounter(d, now)
 				}
 			}
 
 			msElapsed := uint64(now.Sub(prev.at).Milliseconds())
+
+			// Update per-interval snapshot
+			a.diskPrev[cacheTimeMs][name] = prevDiskFromCounter(d, now)
+
+			// Avoid division by zero or clock issues
 			if msElapsed < 100 {
-				// Avoid division by zero or clock issues; update snapshot and continue
-				a.diskPrev[cacheTimeMs][name] = prevDisk{readBytes: d.ReadBytes, writeBytes: d.WriteBytes, at: now}
 				continue
 			}
 
@@ -599,15 +643,38 @@ func (a *Agent) updateDiskIo(cacheTimeMs uint16, systemStats *system.Stats) {
 			// validate values
 			if readMbPerSecond > 50_000 || writeMbPerSecond > 50_000 {
 				slog.Warn("Invalid disk I/O. Resetting.", "name", d.Name, "read", readMbPerSecond, "write", writeMbPerSecond)
-				// Reset interval snapshot and seed from current
-				a.diskPrev[cacheTimeMs][name] = prevDisk{readBytes: d.ReadBytes, writeBytes: d.WriteBytes, at: now}
 				// also refresh agent baseline to avoid future negatives
 				a.initializeDiskIoStats(ioCounters)
 				continue
 			}
 
-			// Update per-interval snapshot
-			a.diskPrev[cacheTimeMs][name] = prevDisk{readBytes: d.ReadBytes, writeBytes: d.WriteBytes, at: now}
+			// These properties are calculated differently on different platforms,
+			// but generally represent cumulative time spent doing reads/writes on the device.
+			// This can surpass 100% if there are multiple concurrent I/O operations.
+			// Linux kernel docs:
+			// This is the total number of milliseconds spent by all reads (as
+			// measured from __make_request() to end_that_request_last()).
+			// https://www.kernel.org/doc/Documentation/iostats.txt (fields 4, 8)
+			diskReadTime := utils.TwoDecimals(float64(d.ReadTime-prev.readTime) / float64(msElapsed) * 100)
+			diskWriteTime := utils.TwoDecimals(float64(d.WriteTime-prev.writeTime) / float64(msElapsed) * 100)
+
+			// I/O utilization %: fraction of wall time the device had any I/O in progress (0-100).
+			diskIoUtilPct := utils.TwoDecimals(float64(d.IoTime-prev.ioTime) / float64(msElapsed) * 100)
+
+			// Weighted I/O: queue-depth weighted I/O time, normalized to interval (can exceed 100%).
+			// Linux kernel field 11: incremented by iops_in_progress × ms_since_last_update.
+			// Used to display queue depth. Multipled by 100 to increase accuracy of digit truncation (divided by 100 in UI).
+			diskWeightedIO := utils.TwoDecimals(float64(d.WeightedIO-prev.weightedIO) / float64(msElapsed) * 100)
+
+			// r_await / w_await: average time per read/write operation in milliseconds.
+			// Equivalent to r_await and w_await in iostat.
+			var rAwait, wAwait float64
+			if deltaReadCount := d.ReadCount - prev.readCount; deltaReadCount > 0 {
+				rAwait = utils.TwoDecimals(float64(d.ReadTime-prev.readTime) / float64(deltaReadCount))
+			}
+			if deltaWriteCount := d.WriteCount - prev.writeCount; deltaWriteCount > 0 {
+				wAwait = utils.TwoDecimals(float64(d.WriteTime-prev.writeTime) / float64(deltaWriteCount))
+			}
 
 			// Update global fsStats baseline for cross-interval correctness
 			stats.Time = now
@@ -617,20 +684,40 @@ func (a *Agent) updateDiskIo(cacheTimeMs uint16, systemStats *system.Stats) {
 			stats.DiskWritePs = writeMbPerSecond
 			stats.DiskReadBytes = diskIORead
 			stats.DiskWriteBytes = diskIOWrite
+			stats.DiskIoStats[0] = diskReadTime
+			stats.DiskIoStats[1] = diskWriteTime
+			stats.DiskIoStats[2] = diskIoUtilPct
+			stats.DiskIoStats[3] = rAwait
+			stats.DiskIoStats[4] = wAwait
+			stats.DiskIoStats[5] = diskWeightedIO
 
 			if stats.Root {
 				systemStats.DiskReadPs = stats.DiskReadPs
 				systemStats.DiskWritePs = stats.DiskWritePs
 				systemStats.DiskIO[0] = diskIORead
 				systemStats.DiskIO[1] = diskIOWrite
+				systemStats.DiskIoStats[0] = diskReadTime
+				systemStats.DiskIoStats[1] = diskWriteTime
+				systemStats.DiskIoStats[2] = diskIoUtilPct
+				systemStats.DiskIoStats[3] = rAwait
+				systemStats.DiskIoStats[4] = wAwait
+				systemStats.DiskIoStats[5] = diskWeightedIO
 			}
 		}
 	}
 }
 
-// getRootMountPoint returns the appropriate root mount point for the system
+// getRootMountPoint returns the appropriate root mount point for the system.
+// On Windows it returns the system drive (e.g. "C:").
 // For immutable systems like Fedora Silverblue, it returns /sysroot instead of /
 func (a *Agent) getRootMountPoint() string {
+	if runtime.GOOS == "windows" {
+		if sd := os.Getenv("SystemDrive"); sd != "" {
+			return sd
+		}
+		return "C:"
+	}
+
 	// 1. Check if /etc/os-release contains indicators of an immutable system
 	if osReleaseContent, err := os.ReadFile("/etc/os-release"); err == nil {
 		content := string(osReleaseContent)

agent/disk_test.go

@@ -530,6 +530,87 @@ func TestAddExtraFilesystemFolders(t *testing.T) {
 	})
 }
 
+func TestAddPartitionExtraFs(t *testing.T) {
+	makeDiscovery := func(agent *Agent) diskDiscovery {
+		return diskDiscovery{
+			agent: agent,
+			ctx: fsRegistrationContext{
+				isWindows: false,
+				efPath:    "/extra-filesystems",
+				diskIoCounters: map[string]disk.IOCountersStat{
+					"nvme0n1p1": {Name: "nvme0n1p1"},
+					"nvme1n1":   {Name: "nvme1n1"},
+				},
+			},
+		}
+	}
+
+	t.Run("registers direct child of extra-filesystems", func(t *testing.T) {
+		agent := &Agent{fsStats: make(map[string]*system.FsStats)}
+		d := makeDiscovery(agent)
+
+		d.addPartitionExtraFs(disk.PartitionStat{
+			Device:     "/dev/nvme0n1p1",
+			Mountpoint: "/extra-filesystems/nvme0n1p1__caddy1-root",
+		})
+
+		stats, exists := agent.fsStats["nvme0n1p1"]
+		assert.True(t, exists)
+		assert.Equal(t, "/extra-filesystems/nvme0n1p1__caddy1-root", stats.Mountpoint)
+		assert.Equal(t, "caddy1-root", stats.Name)
+	})
+
+	t.Run("skips nested mount under extra-filesystem bind mount", func(t *testing.T) {
+		agent := &Agent{fsStats: make(map[string]*system.FsStats)}
+		d := makeDiscovery(agent)
+
+		// These simulate the virtual mounts that appear when host / is bind-mounted
+		// with disk.Partitions(all=true) — e.g. /proc, /sys, /dev visible under the mount.
+		for _, nested := range []string{
+			"/extra-filesystems/nvme0n1p1__caddy1-root/proc",
+			"/extra-filesystems/nvme0n1p1__caddy1-root/sys",
+			"/extra-filesystems/nvme0n1p1__caddy1-root/dev",
+			"/extra-filesystems/nvme0n1p1__caddy1-root/run",
+		} {
+			d.addPartitionExtraFs(disk.PartitionStat{Device: "tmpfs", Mountpoint: nested})
+		}
+
+		assert.Empty(t, agent.fsStats)
+	})
+
+	t.Run("registers both direct children, skips their nested mounts", func(t *testing.T) {
+		agent := &Agent{fsStats: make(map[string]*system.FsStats)}
+		d := makeDiscovery(agent)
+
+		partitions := []disk.PartitionStat{
+			{Device: "/dev/nvme0n1p1", Mountpoint: "/extra-filesystems/nvme0n1p1__caddy1-root"},
+			{Device: "/dev/nvme1n1", Mountpoint: "/extra-filesystems/nvme1n1__caddy1-docker"},
+			{Device: "proc", Mountpoint: "/extra-filesystems/nvme0n1p1__caddy1-root/proc"},
+			{Device: "sysfs", Mountpoint: "/extra-filesystems/nvme0n1p1__caddy1-root/sys"},
+			{Device: "overlay", Mountpoint: "/extra-filesystems/nvme0n1p1__caddy1-root/var/lib/docker"},
+		}
+		for _, p := range partitions {
+			d.addPartitionExtraFs(p)
+		}
+
+		assert.Len(t, agent.fsStats, 2)
+		assert.Equal(t, "caddy1-root", agent.fsStats["nvme0n1p1"].Name)
+		assert.Equal(t, "caddy1-docker", agent.fsStats["nvme1n1"].Name)
+	})
+
+	t.Run("skips partition not under extra-filesystems", func(t *testing.T) {
+		agent := &Agent{fsStats: make(map[string]*system.FsStats)}
+		d := makeDiscovery(agent)
+
+		d.addPartitionExtraFs(disk.PartitionStat{
+			Device:     "/dev/nvme0n1p1",
+			Mountpoint: "/",
+		})
+
+		assert.Empty(t, agent.fsStats)
+	})
+}
+
 func TestFindIoDevice(t *testing.T) {
 	t.Run("matches by device name", func(t *testing.T) {
 		ioCounters := map[string]disk.IOCountersStat{
@@ -687,18 +768,8 @@ func TestIsDockerSpecialMountpoint(t *testing.T) {
 }
 
 func TestInitializeDiskInfoWithCustomNames(t *testing.T) {
-	// Set up environment variables
-	oldEnv := os.Getenv("EXTRA_FILESYSTEMS")
-	defer func() {
-		if oldEnv != "" {
-			os.Setenv("EXTRA_FILESYSTEMS", oldEnv)
-		} else {
-			os.Unsetenv("EXTRA_FILESYSTEMS")
-		}
-	}()
-
 	// Test with custom names
-	os.Setenv("EXTRA_FILESYSTEMS", "sda1__my-storage,/dev/sdb1__backup-drive,nvme0n1p2")
+	t.Setenv("EXTRA_FILESYSTEMS", "sda1__my-storage,/dev/sdb1__backup-drive,nvme0n1p2")
 
 	// Mock disk partitions (we'll just test the parsing logic)
 	// Since the actual disk operations are system-dependent, we'll focus on the parsing
@@ -726,7 +797,7 @@ func TestInitializeDiskInfoWithCustomNames(t *testing.T) {
 
 	for _, tc := range testCases {
 		t.Run("env_"+tc.envValue, func(t *testing.T) {
-			os.Setenv("EXTRA_FILESYSTEMS", tc.envValue)
+			t.Setenv("EXTRA_FILESYSTEMS", tc.envValue)
 
 			// Create mock partitions that would match our test cases
 			partitions := []disk.PartitionStat{}

agent/docker.go

@@ -16,6 +16,8 @@ import (
 	"os"
 	"path"
 	"regexp"
+	"sort"
+	"strconv"
 	"strings"
 	"sync"
 	"time"
@@ -23,6 +25,7 @@ import (
 	"github.com/henrygd/beszel/agent/deltatracker"
 	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/container"
+	"github.com/henrygd/beszel/internal/entities/system"
 
 	"github.com/blang/semver"
 )
@@ -50,20 +53,21 @@ const (
 )
 
 type dockerManager struct {
-	client              *http.Client                // Client to query Docker API
-	wg                  sync.WaitGroup              // WaitGroup to wait for all goroutines to finish
-	sem                 chan struct{}               // Semaphore to limit concurrent container requests
-	containerStatsMutex sync.RWMutex                // Mutex to prevent concurrent access to containerStatsMap
-	apiContainerList    []*container.ApiInfo        // List of containers from Docker API
-	containerStatsMap   map[string]*container.Stats // Keeps track of container stats
-	validIds            map[string]struct{}         // Map of valid container ids, used to prune invalid containers from containerStatsMap
-	goodDockerVersion   bool                        // Whether docker version is at least 25.0.0 (one-shot works correctly)
-	isWindows           bool                        // Whether the Docker Engine API is running on Windows
-	buf                 *bytes.Buffer               // Buffer to store and read response bodies
-	decoder             *json.Decoder               // Reusable JSON decoder that reads from buf
-	apiStats            *container.ApiStats         // Reusable API stats object
-	excludeContainers   []string                    // Patterns to exclude containers by name
-	usingPodman         bool                        // Whether the Docker Engine API is running on Podman
+	agent                *Agent                      // Used to propagate system detail changes back to the agent
+	client               *http.Client                // Client to query Docker API
+	wg                   sync.WaitGroup              // WaitGroup to wait for all goroutines to finish
+	sem                  chan struct{}               // Semaphore to limit concurrent container requests
+	containerStatsMutex  sync.RWMutex                // Mutex to prevent concurrent access to containerStatsMap
+	apiContainerList     []*container.ApiInfo        // List of containers from Docker API
+	containerStatsMap    map[string]*container.Stats // Keeps track of container stats
+	validIds             map[string]struct{}         // Map of valid container ids, used to prune invalid containers from containerStatsMap
+	goodDockerVersion    bool                        // Whether docker version is at least 25.0.0 (one-shot works correctly)
+	dockerVersionChecked bool                        // Whether a version probe has completed successfully
+	isWindows            bool                        // Whether the Docker Engine API is running on Windows
+	buf                  *bytes.Buffer               // Buffer to store and read response bodies
+	apiStats             *container.ApiStats         // Reusable API stats object
+	excludeContainers    []string                    // Patterns to exclude containers by name
+	usingPodman          bool                        // Whether the Docker Engine API is running on Podman
 
 	// Cache-time-aware tracking for CPU stats (similar to cpu.go)
 	// Maps cache time intervals to container-specific CPU usage tracking
@@ -75,7 +79,7 @@ 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)
+	lastNetworkReadTime map[uint16]map[string]time.Time // cacheTimeMs -> containerId -> last network read time
 }
 
 // userAgentRoundTripper is a custom http.RoundTripper that adds a User-Agent header to all requests
@@ -84,6 +88,14 @@ type userAgentRoundTripper struct {
 	userAgent string
 }
 
+// dockerVersionResponse contains the /version fields used for engine checks.
+type dockerVersionResponse struct {
+	Version    string `json:"Version"`
+	Components []struct {
+		Name string `json:"Name"`
+	} `json:"Components"`
+}
+
 // RoundTrip implements the http.RoundTripper interface
 func (u *userAgentRoundTripper) RoundTrip(req *http.Request) (*http.Response, error) {
 	req.Header.Set("User-Agent", u.userAgent)
@@ -131,7 +143,14 @@ func (dm *dockerManager) getDockerStats(cacheTimeMs uint16) ([]*container.Stats,
 		return nil, err
 	}
 
-	dm.isWindows = strings.Contains(resp.Header.Get("Server"), "windows")
+	// Detect Podman and Windows from Server header
+	serverHeader := resp.Header.Get("Server")
+	if !dm.usingPodman && detectPodmanFromHeader(serverHeader) {
+		dm.setIsPodman()
+	}
+	dm.isWindows = strings.Contains(serverHeader, "windows")
+
+	dm.ensureDockerVersionChecked()
 
 	containersLength := len(dm.apiContainerList)
 
@@ -283,7 +302,7 @@ func (dm *dockerManager) cycleNetworkDeltasForCacheTime(cacheTimeMs uint16) {
 }
 
 // calculateNetworkStats calculates network sent/receive deltas using DeltaTracker
-func (dm *dockerManager) calculateNetworkStats(ctr *container.ApiInfo, apiStats *container.ApiStats, stats *container.Stats, initialized bool, name string, cacheTimeMs uint16) (uint64, uint64) {
+func (dm *dockerManager) calculateNetworkStats(ctr *container.ApiInfo, apiStats *container.ApiStats, name string, cacheTimeMs uint16) (uint64, uint64) {
 	var total_sent, total_recv uint64
 	for _, v := range apiStats.Networks {
 		total_sent += v.TxBytes
@@ -302,10 +321,11 @@ func (dm *dockerManager) calculateNetworkStats(ctr *container.ApiInfo, apiStats
 	sent_delta_raw := sentTracker.Delta(ctr.IdShort)
 	recv_delta_raw := recvTracker.Delta(ctr.IdShort)
 
-	// Calculate bytes per second independently for Tx and Rx if we have previous data
+	// Calculate bytes per second using per-cache-time read time to avoid
+	// interference between different cache intervals (e.g. 1000ms vs 60000ms)
 	var sent_delta, recv_delta uint64
-	if initialized {
-		millisecondsElapsed := uint64(time.Since(stats.PrevReadTime).Milliseconds())
+	if prevReadTime, ok := dm.lastNetworkReadTime[cacheTimeMs][ctr.IdShort]; ok {
+		millisecondsElapsed := uint64(time.Since(prevReadTime).Milliseconds())
 		if millisecondsElapsed > 0 {
 			if sent_delta_raw > 0 {
 				sent_delta = sent_delta_raw * 1000 / millisecondsElapsed
@@ -346,6 +366,49 @@ func updateContainerStatsValues(stats *container.Stats, cpuPct float64, usedMemo
 	stats.PrevReadTime = readTime
 }
 
+// convertContainerPortsToString formats the ports of a container into a sorted, deduplicated string.
+// ctr.Ports is nilled out after processing so the slice is not accidentally reused.
+func convertContainerPortsToString(ctr *container.ApiInfo) string {
+	if len(ctr.Ports) == 0 {
+		return ""
+	}
+	sort.Slice(ctr.Ports, func(i, j int) bool {
+		if ctr.Ports[i].PublicPort != ctr.Ports[j].PublicPort {
+			return ctr.Ports[i].PublicPort < ctr.Ports[j].PublicPort
+		}
+		return ctr.Ports[i].IP < ctr.Ports[j].IP
+	})
+	var builder strings.Builder
+	seen := make(map[string]struct{})
+	for _, p := range ctr.Ports {
+		if p.PublicPort == 0 {
+			continue
+		}
+		keyIP := p.IP
+		if keyIP == "0.0.0.0" || keyIP == "::" {
+			keyIP = ""
+		}
+		key := keyIP + ":" + strconv.Itoa(int(p.PublicPort))
+		if _, ok := seen[key]; ok {
+			continue
+		}
+		seen[key] = struct{}{}
+		if builder.Len() > 0 {
+			builder.WriteString(", ")
+		}
+		switch p.IP {
+		case "0.0.0.0", "::":
+		default:
+			builder.WriteString(p.IP)
+			builder.WriteByte(':')
+		}
+		builder.WriteString(strconv.Itoa(int(p.PublicPort)))
+	}
+	// clear ports slice so it doesn't get reused and blend into next response
+	ctr.Ports = nil
+	return builder.String()
+}
+
 func parseDockerStatus(status string) (string, container.DockerHealth) {
 	trimmed := strings.TrimSpace(status)
 	if trimmed == "" {
@@ -365,22 +428,60 @@ func parseDockerStatus(status string) (string, container.DockerHealth) {
 		statusText = trimmed
 	}
 
-	healthText := strings.ToLower(strings.TrimSpace(strings.TrimSuffix(trimmed[openIdx+1:], ")")))
+	healthText := strings.TrimSpace(strings.TrimSuffix(trimmed[openIdx+1:], ")"))
 	// Some Docker statuses include a "health:" prefix inside the parentheses.
 	// Strip it so it maps correctly to the known health states.
 	if colonIdx := strings.IndexRune(healthText, ':'); colonIdx != -1 {
-		prefix := strings.TrimSpace(healthText[:colonIdx])
+		prefix := strings.ToLower(strings.TrimSpace(healthText[:colonIdx]))
 		if prefix == "health" || prefix == "health status" {
 			healthText = strings.TrimSpace(healthText[colonIdx+1:])
 		}
 	}
-	if health, ok := container.DockerHealthStrings[healthText]; ok {
+	if health, ok := parseDockerHealthStatus(healthText); ok {
 		return statusText, health
 	}
 
 	return trimmed, container.DockerHealthNone
 }
 
+// parseDockerHealthStatus maps Docker health status strings to container.DockerHealth values
+func parseDockerHealthStatus(status string) (container.DockerHealth, bool) {
+	health, ok := container.DockerHealthStrings[strings.ToLower(strings.TrimSpace(status))]
+	return health, ok
+}
+
+// getPodmanContainerHealth fetches container health status from the container inspect endpoint.
+// Used for Podman which doesn't provide health status in the /containers/json endpoint as of March 2026.
+// https://github.com/containers/podman/issues/27786
+func (dm *dockerManager) getPodmanContainerHealth(containerID string) (container.DockerHealth, error) {
+	resp, err := dm.client.Get(fmt.Sprintf("http://localhost/containers/%s/json", url.PathEscape(containerID)))
+	if err != nil {
+		return container.DockerHealthNone, err
+	}
+	defer resp.Body.Close()
+
+	if resp.StatusCode != http.StatusOK {
+		return container.DockerHealthNone, fmt.Errorf("container inspect request failed: %s", resp.Status)
+	}
+
+	var inspectInfo struct {
+		State struct {
+			Health struct {
+				Status string
+			}
+		}
+	}
+	if err := json.NewDecoder(resp.Body).Decode(&inspectInfo); err != nil {
+		return container.DockerHealthNone, err
+	}
+
+	if health, ok := parseDockerHealthStatus(inspectInfo.State.Health.Status); ok {
+		return health, nil
+	}
+
+	return container.DockerHealthNone, nil
+}
+
 // Updates stats for individual container with cache-time-aware delta tracking
 func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo, cacheTimeMs uint16) error {
 	name := ctr.Names[0][1:]
@@ -390,6 +491,21 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo, cacheTimeM
 		return err
 	}
 
+	statusText, health := parseDockerStatus(ctr.Status)
+
+	// Docker exposes Health.Status on /containers/json in API 1.52+.
+	// Podman currently requires falling back to the inspect endpoint as of March 2026.
+	// https://github.com/containers/podman/issues/27786
+	if ctr.Health.Status != "" {
+		if h, ok := parseDockerHealthStatus(ctr.Health.Status); ok {
+			health = h
+		}
+	} else if dm.usingPodman {
+		if podmanHealth, err := dm.getPodmanContainerHealth(ctr.IdShort); err == nil {
+			health = podmanHealth
+		}
+	}
+
 	dm.containerStatsMutex.Lock()
 	defer dm.containerStatsMutex.Unlock()
 
@@ -401,11 +517,13 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo, cacheTimeM
 	}
 
 	stats.Id = ctr.IdShort
-
-	statusText, health := parseDockerStatus(ctr.Status)
 	stats.Status = statusText
 	stats.Health = health
 
+	if len(ctr.Ports) > 0 {
+		stats.Ports = convertContainerPortsToString(ctr)
+	}
+
 	// reset current stats
 	stats.Cpu = 0
 	stats.Mem = 0
@@ -452,7 +570,13 @@ func (dm *dockerManager) updateContainerStats(ctr *container.ApiInfo, cacheTimeM
 	}
 
 	// Calculate network stats using DeltaTracker
-	sent_delta, recv_delta := dm.calculateNetworkStats(ctr, res, stats, initialized, name, cacheTimeMs)
+	sent_delta, recv_delta := dm.calculateNetworkStats(ctr, res, name, cacheTimeMs)
+
+	// Store per-cache-time network read time for next rate calculation
+	if dm.lastNetworkReadTime[cacheTimeMs] == nil {
+		dm.lastNetworkReadTime[cacheTimeMs] = make(map[string]time.Time)
+	}
+	dm.lastNetworkReadTime[cacheTimeMs][ctr.IdShort] = time.Now()
 
 	// Store current network values for legacy compatibility
 	var total_sent, total_recv uint64
@@ -484,10 +608,13 @@ func (dm *dockerManager) deleteContainerStatsSync(id string) {
 	for ct := range dm.lastCpuReadTime {
 		delete(dm.lastCpuReadTime[ct], id)
 	}
+	for ct := range dm.lastNetworkReadTime {
+		delete(dm.lastNetworkReadTime[ct], id)
+	}
 }
 
 // Creates a new http client for Docker or Podman API
-func newDockerManager() *dockerManager {
+func newDockerManager(agent *Agent) *dockerManager {
 	dockerHost, exists := utils.GetEnv("DOCKER_HOST")
 	if exists {
 		// return nil if set to empty string
@@ -553,6 +680,7 @@ func newDockerManager() *dockerManager {
 	}
 
 	manager := &dockerManager{
+		agent: agent,
 		client: &http.Client{
 			Timeout:   timeout,
 			Transport: userAgentTransport,
@@ -569,51 +697,55 @@ 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,
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 	}
 
-	// If using podman, return client
-	if strings.Contains(dockerHost, "podman") {
-		manager.usingPodman = true
-		manager.goodDockerVersion = true
-		return manager
-	}
-
-	// 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
-	time.Sleep(50 * time.Millisecond)
+	// Best-effort startup probe. If the engine is not ready yet, getDockerStats will
+	// retry after the first successful /containers/json request.
+	_, _ = manager.checkDockerVersion()
 
 	return manager
 }
 
 // checkDockerVersion checks Docker version and sets goodDockerVersion if at least 25.0.0.
 // Versions before 25.0.0 have a bug with one-shot which requires all requests to be made in one batch.
-func (dm *dockerManager) checkDockerVersion() {
-	var err error
-	var resp *http.Response
-	var versionInfo struct {
-		Version string `json:"Version"`
+func (dm *dockerManager) checkDockerVersion() (bool, error) {
+	resp, err := dm.client.Get("http://localhost/version")
+	if err != nil {
+		return false, err
 	}
-	const versionMaxTries = 2
-	for i := 1; i <= versionMaxTries; i++ {
-		resp, err = dm.client.Get("http://localhost/version")
-		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, "err", err, "response", resp)
-			dm.retrySleep(5 * time.Second)
-		}
+	if resp.StatusCode != http.StatusOK {
+		status := resp.Status
+		resp.Body.Close()
+		return false, fmt.Errorf("docker version request failed: %s", status)
 	}
-	if err != nil || resp.StatusCode != http.StatusOK {
+
+	var versionInfo dockerVersionResponse
+	serverHeader := resp.Header.Get("Server")
+	if err := dm.decode(resp, &versionInfo); err != nil {
+		return false, err
+	}
+
+	dm.applyDockerVersionInfo(serverHeader, &versionInfo)
+	dm.dockerVersionChecked = true
+	return true, nil
+}
+
+// ensureDockerVersionChecked retries the version probe after a successful
+// container list request.
+func (dm *dockerManager) ensureDockerVersionChecked() {
+	if dm.dockerVersionChecked {
 		return
 	}
-	if err := dm.decode(resp, &versionInfo); err != nil {
+	if _, err := dm.checkDockerVersion(); err != nil {
+		slog.Debug("Failed to get Docker version", "err", err)
+	}
+}
+
+// applyDockerVersionInfo updates version-dependent behavior from engine metadata.
+func (dm *dockerManager) applyDockerVersionInfo(serverHeader string, versionInfo *dockerVersionResponse) {
+	if detectPodmanEngine(serverHeader, versionInfo) {
+		dm.setIsPodman()
 		return
 	}
 	// if version > 24, one-shot works correctly and we can limit concurrent operations
@@ -624,20 +756,18 @@ func (dm *dockerManager) checkDockerVersion() {
 	}
 }
 
-// Decodes Docker API JSON response using a reusable buffer and decoder. Not thread safe.
+// Decodes a Docker API JSON response using a reusable buffer. Not thread safe.
 func (dm *dockerManager) decode(resp *http.Response, d any) error {
 	if dm.buf == nil {
 		// initialize buffer with 256kb starting size
 		dm.buf = bytes.NewBuffer(make([]byte, 0, 1024*256))
-		dm.decoder = json.NewDecoder(dm.buf)
 	}
 	defer resp.Body.Close()
 	defer dm.buf.Reset()
-	_, err := dm.buf.ReadFrom(resp.Body)
-	if err != nil {
+	if _, err := dm.buf.ReadFrom(resp.Body); err != nil {
 		return err
 	}
-	return dm.decoder.Decode(d)
+	return json.Unmarshal(dm.buf.Bytes(), d)
 }
 
 // Test docker / podman sockets and return if one exists
@@ -839,3 +969,46 @@ func (dm *dockerManager) GetHostInfo() (info container.HostInfo, err error) {
 func (dm *dockerManager) IsPodman() bool {
 	return dm.usingPodman
 }
+
+// setIsPodman sets the manager to Podman mode and updates system details accordingly.
+func (dm *dockerManager) setIsPodman() {
+	if dm.usingPodman {
+		return
+	}
+	dm.usingPodman = true
+	dm.goodDockerVersion = true
+	dm.dockerVersionChecked = true
+	// keep system details updated - this may be detected late if server isn't ready when
+	// agent starts, so make sure we notify the hub if this happens later.
+	if dm.agent != nil {
+		dm.agent.updateSystemDetails(func(details *system.Details) {
+			details.Podman = true
+		})
+	}
+}
+
+// detectPodmanFromHeader identifies Podman from the Docker API server header.
+func detectPodmanFromHeader(server string) bool {
+	return strings.HasPrefix(server, "Libpod")
+}
+
+// detectPodmanFromVersion identifies Podman from the version payload.
+func detectPodmanFromVersion(versionInfo *dockerVersionResponse) bool {
+	if versionInfo == nil {
+		return false
+	}
+	for _, component := range versionInfo.Components {
+		if strings.HasPrefix(component.Name, "Podman") {
+			return true
+		}
+	}
+	return false
+}
+
+// detectPodmanEngine checks both header and version metadata for Podman.
+func detectPodmanEngine(serverHeader string, versionInfo *dockerVersionResponse) bool {
+	if detectPodmanFromHeader(serverHeader) {
+		return true
+	}
+	return detectPodmanFromVersion(versionInfo)
+}

agent/docker_test.go

@@ -35,6 +35,12 @@ type recordingRoundTripper struct {
 	lastQuery   map[string]string
 }
 
+type roundTripFunc func(*http.Request) (*http.Response, error)
+
+func (fn roundTripFunc) RoundTrip(req *http.Request) (*http.Response, error) {
+	return fn(req)
+}
+
 func (rt *recordingRoundTripper) RoundTrip(req *http.Request) (*http.Response, error) {
 	rt.called = true
 	rt.lastPath = req.URL.EscapedPath()
@@ -214,6 +220,28 @@ func TestContainerDetailsRequestsUseExpectedDockerPaths(t *testing.T) {
 	})
 }
 
+func TestGetPodmanContainerHealth(t *testing.T) {
+	called := false
+	dm := &dockerManager{
+		client: &http.Client{Transport: roundTripFunc(func(req *http.Request) (*http.Response, error) {
+			called = true
+			assert.Equal(t, "/containers/0123456789ab/json", req.URL.EscapedPath())
+			return &http.Response{
+				StatusCode: http.StatusOK,
+				Status:     "200 OK",
+				Header:     make(http.Header),
+				Body:       io.NopCloser(strings.NewReader(`{"State":{"Health":{"Status":"healthy"}}}`)),
+				Request:    req,
+			}, nil
+		})},
+	}
+
+	health, err := dm.getPodmanContainerHealth("0123456789ab")
+	require.NoError(t, err)
+	assert.True(t, called)
+	assert.Equal(t, container.DockerHealthHealthy, health)
+}
+
 func TestValidateCpuPercentage(t *testing.T) {
 	tests := []struct {
 		name          string
@@ -380,6 +408,7 @@ func TestCalculateNetworkStats(t *testing.T) {
 	dm := &dockerManager{
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 	}
 
 	cacheTimeMs := uint16(30000)
@@ -395,6 +424,11 @@ func TestCalculateNetworkStats(t *testing.T) {
 	dm.networkSentTrackers[cacheTimeMs] = sentTracker
 	dm.networkRecvTrackers[cacheTimeMs] = recvTracker
 
+	// Set per-cache-time network read time (1 second ago)
+	dm.lastNetworkReadTime[cacheTimeMs] = map[string]time.Time{
+		"container1": time.Now().Add(-time.Second),
+	}
+
 	ctr := &container.ApiInfo{
 		IdShort: "container1",
 	}
@@ -405,12 +439,8 @@ func TestCalculateNetworkStats(t *testing.T) {
 		},
 	}
 
-	stats := &container.Stats{
-		PrevReadTime: time.Now().Add(-time.Second), // 1 second ago
-	}
-
 	// Test with initialized container
-	sent, recv := dm.calculateNetworkStats(ctr, apiStats, stats, true, "test-container", cacheTimeMs)
+	sent, recv := dm.calculateNetworkStats(ctr, apiStats, "test-container", cacheTimeMs)
 
 	// Should return calculated byte rates per second
 	assert.GreaterOrEqual(t, sent, uint64(0))
@@ -418,12 +448,76 @@ func TestCalculateNetworkStats(t *testing.T) {
 
 	// Cycle and test one-direction change (Tx only) is reflected independently
 	dm.cycleNetworkDeltasForCacheTime(cacheTimeMs)
+	dm.lastNetworkReadTime[cacheTimeMs]["container1"] = time.Now().Add(-time.Second)
 	apiStats.Networks["eth0"] = container.NetworkStats{TxBytes: 2500, RxBytes: 1800} // +500 Tx only
-	sent, recv = dm.calculateNetworkStats(ctr, apiStats, stats, true, "test-container", cacheTimeMs)
+	sent, recv = dm.calculateNetworkStats(ctr, apiStats, "test-container", cacheTimeMs)
 	assert.Greater(t, sent, uint64(0))
 	assert.Equal(t, uint64(0), recv)
 }
 
+// TestNetworkStatsCacheTimeIsolation verifies that frequent collections at one cache time
+// (e.g. 1000ms) don't cause inflated rates at another cache time (e.g. 60000ms).
+// This was a bug where PrevReadTime was shared, so the 60000ms tracker would see a
+// large byte delta divided by a tiny elapsed time (set by the 1000ms path).
+func TestNetworkStatsCacheTimeIsolation(t *testing.T) {
+	dm := &dockerManager{
+		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
+	}
+
+	ctr := &container.ApiInfo{IdShort: "container1"}
+	fastCache := uint16(1000)
+	slowCache := uint16(60000)
+
+	// Baseline for both cache times at T=0 with 100 bytes total
+	baseline := &container.ApiStats{
+		Networks: map[string]container.NetworkStats{
+			"eth0": {TxBytes: 100, RxBytes: 100},
+		},
+	}
+	dm.calculateNetworkStats(ctr, baseline, "test", fastCache)
+	dm.calculateNetworkStats(ctr, baseline, "test", slowCache)
+
+	// Record read times and cycle both
+	now := time.Now()
+	dm.lastNetworkReadTime[fastCache] = map[string]time.Time{"container1": now}
+	dm.lastNetworkReadTime[slowCache] = map[string]time.Time{"container1": now}
+	dm.cycleNetworkDeltasForCacheTime(fastCache)
+	dm.cycleNetworkDeltasForCacheTime(slowCache)
+
+	// Simulate many fast (1000ms) collections over ~5 seconds, each adding 10 bytes
+	totalBytes := uint64(100)
+	for i := 0; i < 5; i++ {
+		totalBytes += 10
+		stats := &container.ApiStats{
+			Networks: map[string]container.NetworkStats{
+				"eth0": {TxBytes: totalBytes, RxBytes: totalBytes},
+			},
+		}
+		// Set fast cache read time to 1 second ago
+		dm.lastNetworkReadTime[fastCache]["container1"] = time.Now().Add(-time.Second)
+		sent, _ := dm.calculateNetworkStats(ctr, stats, "test", fastCache)
+		// Fast cache should see ~10 bytes/sec per interval
+		assert.LessOrEqual(t, sent, uint64(100), "fast cache rate should be reasonable")
+		dm.cycleNetworkDeltasForCacheTime(fastCache)
+	}
+
+	// Now do slow cache collection — total delta is 50 bytes over ~5 seconds
+	// Set slow cache read time to 5 seconds ago (the actual elapsed time)
+	dm.lastNetworkReadTime[slowCache]["container1"] = time.Now().Add(-5 * time.Second)
+	finalStats := &container.ApiStats{
+		Networks: map[string]container.NetworkStats{
+			"eth0": {TxBytes: totalBytes, RxBytes: totalBytes},
+		},
+	}
+	sent, _ := dm.calculateNetworkStats(ctr, finalStats, "test", slowCache)
+
+	// Slow cache rate should be ~10 bytes/sec (50 bytes / 5 seconds), NOT 100x inflated
+	assert.LessOrEqual(t, sent, uint64(100), "slow cache rate should NOT be inflated by fast cache collections")
+	assert.GreaterOrEqual(t, sent, uint64(1), "slow cache should still report some traffic")
+}
+
 func TestDockerManagerCreation(t *testing.T) {
 	// Test that dockerManager can be created without panicking
 	dm := &dockerManager{
@@ -432,6 +526,7 @@ func TestDockerManagerCreation(t *testing.T) {
 		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 	}
 
 	assert.NotNil(t, dm)
@@ -439,63 +534,58 @@ func TestDockerManagerCreation(t *testing.T) {
 	assert.NotNil(t, dm.lastCpuSystem)
 	assert.NotNil(t, dm.networkSentTrackers)
 	assert.NotNil(t, dm.networkRecvTrackers)
+	assert.NotNil(t, dm.lastNetworkReadTime)
 }
 
 func TestCheckDockerVersion(t *testing.T) {
 	tests := []struct {
-		name      string
-		responses []struct {
-			statusCode int
-			body       string
-		}
-		expectedGood     bool
-		expectedRequests int
+		name            string
+		statusCode      int
+		body            string
+		server          string
+		expectSuccess   bool
+		expectedGood    bool
+		expectedPodman  bool
+		expectError     bool
+		expectedRequest string
 	}{
 		{
-			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:            "good docker version",
+			statusCode:      http.StatusOK,
+			body:            `{"Version":"25.0.1"}`,
+			expectSuccess:   true,
+			expectedGood:    true,
+			expectedPodman:  false,
+			expectedRequest: "/version",
 		},
 		{
-			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:            "old docker version",
+			statusCode:      http.StatusOK,
+			body:            `{"Version":"24.0.7"}`,
+			expectSuccess:   true,
+			expectedGood:    false,
+			expectedPodman:  false,
+			expectedRequest: "/version",
 		},
 		{
-			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:            "podman from server header",
+			statusCode:      http.StatusOK,
+			body:            `{"Version":"5.5.0"}`,
+			server:          "Libpod/5.5.0",
+			expectSuccess:   true,
+			expectedGood:    true,
+			expectedPodman:  true,
+			expectedRequest: "/version",
 		},
 		{
-			name: "non-200 on all retries",
-			responses: []struct {
-				statusCode int
-				body       string
-			}{
-				{http.StatusInternalServerError, `"error"`},
-				{http.StatusUnauthorized, `"error"`},
-			},
-			expectedGood:     false,
-			expectedRequests: 2,
+			name:            "non-200 response",
+			statusCode:      http.StatusServiceUnavailable,
+			body:            `"not ready"`,
+			expectSuccess:   false,
+			expectedGood:    false,
+			expectedPodman:  false,
+			expectError:     true,
+			expectedRequest: "/version",
 		},
 	}
 
@@ -503,13 +593,13 @@ func TestCheckDockerVersion(t *testing.T) {
 		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
+				assert.Equal(t, tt.expectedRequest, r.URL.EscapedPath())
+				if tt.server != "" {
+					w.Header().Set("Server", tt.server)
 				}
-				w.WriteHeader(tt.responses[idx].statusCode)
-				fmt.Fprint(w, tt.responses[idx].body)
+				w.WriteHeader(tt.statusCode)
+				fmt.Fprint(w, tt.body)
 			}))
 			defer server.Close()
 
@@ -521,17 +611,24 @@ func TestCheckDockerVersion(t *testing.T) {
 						},
 					},
 				},
-				retrySleep: func(time.Duration) {},
 			}
 
-			dm.checkDockerVersion()
+			success, err := dm.checkDockerVersion()
 
+			assert.Equal(t, tt.expectSuccess, success)
+			assert.Equal(t, tt.expectSuccess, dm.dockerVersionChecked)
 			assert.Equal(t, tt.expectedGood, dm.goodDockerVersion)
-			assert.Equal(t, tt.expectedRequests, requestCount)
+			assert.Equal(t, tt.expectedPodman, dm.usingPodman)
+			assert.Equal(t, 1, requestCount)
+			if tt.expectError {
+				require.Error(t, err)
+			} else {
+				require.NoError(t, err)
+			}
 		})
 	}
 
-	t.Run("request error on all retries", func(t *testing.T) {
+	t.Run("request error", func(t *testing.T) {
 		requestCount := 0
 		dm := &dockerManager{
 			client: &http.Client{
@@ -542,16 +639,189 @@ func TestCheckDockerVersion(t *testing.T) {
 					},
 				},
 			},
-			retrySleep: func(time.Duration) {},
 		}
 
-		dm.checkDockerVersion()
+		success, err := dm.checkDockerVersion()
 
+		assert.False(t, success)
+		require.Error(t, err)
+		assert.False(t, dm.dockerVersionChecked)
 		assert.False(t, dm.goodDockerVersion)
-		assert.Equal(t, 2, requestCount)
+		assert.False(t, dm.usingPodman)
+		assert.Equal(t, 1, requestCount)
 	})
 }
 
+// newDockerManagerForVersionTest creates a dockerManager wired to a test server.
+func newDockerManagerForVersionTest(server *httptest.Server) *dockerManager {
+	return &dockerManager{
+		client: &http.Client{
+			Transport: &http.Transport{
+				DialContext: func(_ context.Context, network, _ string) (net.Conn, error) {
+					return net.Dial(network, server.Listener.Addr().String())
+				},
+			},
+		},
+		containerStatsMap:   make(map[string]*container.Stats),
+		lastCpuContainer:    make(map[uint16]map[string]uint64),
+		lastCpuSystem:       make(map[uint16]map[string]uint64),
+		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
+		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
+	}
+}
+
+func TestGetDockerStatsChecksDockerVersionAfterContainerList(t *testing.T) {
+	tests := []struct {
+		name            string
+		containerServer string
+		versionServer   string
+		versionBody     string
+		expectedGood    bool
+		expectedPodman  bool
+	}{
+		{
+			name:           "200 with good version on first try",
+			versionBody:    `{"Version":"25.0.1"}`,
+			expectedGood:   true,
+			expectedPodman: false,
+		},
+		{
+			name:           "200 with old version on first try",
+			versionBody:    `{"Version":"24.0.7"}`,
+			expectedGood:   false,
+			expectedPodman: false,
+		},
+		{
+			name:            "podman detected from server header",
+			containerServer: "Libpod/5.5.0",
+			expectedGood:    true,
+			expectedPodman:  true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			requestCounts := map[string]int{}
+			server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+				requestCounts[r.URL.EscapedPath()]++
+				switch r.URL.EscapedPath() {
+				case "/containers/json":
+					if tt.containerServer != "" {
+						w.Header().Set("Server", tt.containerServer)
+					}
+					w.WriteHeader(http.StatusOK)
+					fmt.Fprint(w, `[]`)
+				case "/version":
+					if tt.versionServer != "" {
+						w.Header().Set("Server", tt.versionServer)
+					}
+					w.WriteHeader(http.StatusOK)
+					fmt.Fprint(w, tt.versionBody)
+				default:
+					t.Fatalf("unexpected path: %s", r.URL.EscapedPath())
+				}
+			}))
+			defer server.Close()
+
+			dm := newDockerManagerForVersionTest(server)
+
+			stats, err := dm.getDockerStats(defaultCacheTimeMs)
+			require.NoError(t, err)
+			assert.Empty(t, stats)
+			assert.True(t, dm.dockerVersionChecked)
+			assert.Equal(t, tt.expectedGood, dm.goodDockerVersion)
+			assert.Equal(t, tt.expectedPodman, dm.usingPodman)
+			assert.Equal(t, 1, requestCounts["/containers/json"])
+			if tt.expectedPodman {
+				assert.Equal(t, 0, requestCounts["/version"])
+			} else {
+				assert.Equal(t, 1, requestCounts["/version"])
+			}
+
+			stats, err = dm.getDockerStats(defaultCacheTimeMs)
+			require.NoError(t, err)
+			assert.Empty(t, stats)
+			assert.Equal(t, tt.expectedGood, dm.goodDockerVersion)
+			assert.Equal(t, tt.expectedPodman, dm.usingPodman)
+			assert.Equal(t, 2, requestCounts["/containers/json"])
+			if tt.expectedPodman {
+				assert.Equal(t, 0, requestCounts["/version"])
+			} else {
+				assert.Equal(t, 1, requestCounts["/version"])
+			}
+		})
+	}
+
+}
+
+func TestGetDockerStatsRetriesVersionCheckUntilSuccess(t *testing.T) {
+	requestCounts := map[string]int{}
+	versionStatuses := []int{http.StatusServiceUnavailable, http.StatusOK}
+	versionBodies := []string{`"not ready"`, `{"Version":"25.1.0"}`}
+
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		requestCounts[r.URL.EscapedPath()]++
+		switch r.URL.EscapedPath() {
+		case "/containers/json":
+			w.WriteHeader(http.StatusOK)
+			fmt.Fprint(w, `[]`)
+		case "/version":
+			idx := requestCounts["/version"] - 1
+			if idx >= len(versionStatuses) {
+				idx = len(versionStatuses) - 1
+			}
+			w.WriteHeader(versionStatuses[idx])
+			fmt.Fprint(w, versionBodies[idx])
+		default:
+			t.Fatalf("unexpected path: %s", r.URL.EscapedPath())
+		}
+	}))
+	defer server.Close()
+
+	dm := newDockerManagerForVersionTest(server)
+
+	stats, err := dm.getDockerStats(defaultCacheTimeMs)
+	require.NoError(t, err)
+	assert.Empty(t, stats)
+	assert.False(t, dm.dockerVersionChecked)
+	assert.False(t, dm.goodDockerVersion)
+	assert.Equal(t, 1, requestCounts["/version"])
+
+	stats, err = dm.getDockerStats(defaultCacheTimeMs)
+	require.NoError(t, err)
+	assert.Empty(t, stats)
+	assert.True(t, dm.dockerVersionChecked)
+	assert.True(t, dm.goodDockerVersion)
+	assert.Equal(t, 2, requestCounts["/containers/json"])
+	assert.Equal(t, 2, requestCounts["/version"])
+
+	stats, err = dm.getDockerStats(defaultCacheTimeMs)
+	require.NoError(t, err)
+	assert.Empty(t, stats)
+	assert.Equal(t, 3, requestCounts["/containers/json"])
+	assert.Equal(t, 2, requestCounts["/version"])
+}
+
+// A failed decode must not break later decodes. Previously the reused json.Decoder
+// stayed desynced after one truncated response, breaking decode until restart.
+func TestDecodeRecoversFromError(t *testing.T) {
+	dm := &dockerManager{}
+
+	// truncated JSON: body reads fine, decode fails
+	var bad []container.ApiInfo
+	err := dm.decode(&http.Response{Body: io.NopCloser(strings.NewReader(`[{"Id":"abc`))}, &bad)
+	require.Error(t, err)
+
+	// the next decode must still succeed
+	var good []container.ApiInfo
+	err = dm.decode(&http.Response{Body: io.NopCloser(strings.NewReader(`[{"Id":"abcdef012345","Names":["/ok"]}]`))}, &good)
+	require.NoError(t, err)
+	require.Len(t, good, 1)
+	assert.Equal(t, "abcdef012345", good[0].Id)
+}
+
 func TestCycleCpuDeltas(t *testing.T) {
 	dm := &dockerManager{
 		lastCpuContainer: map[uint16]map[string]uint64{
@@ -623,6 +893,7 @@ func TestDockerStatsWithMockData(t *testing.T) {
 		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 		containerStatsMap:   make(map[string]*container.Stats),
 	}
 
@@ -750,6 +1021,44 @@ func TestCpuPercentageCalculationWithRealData(t *testing.T) {
 	assert.InDelta(t, expectedPct, actualPct, 0.01)
 }
 
+func TestCpuPercentageHandlesCounterRollback(t *testing.T) {
+	// If a stats response is processed after a newer one for the same container,
+	// or an accounting counter resets, the current total can be lower than the
+	// stored previous value. Unsigned subtraction wraps to ~2^64 instead of
+	// going negative, so the percentage explodes, validateCpuPercentage rejects
+	// the sample, and the whole collection is discarded - network stats too.
+	stats := &container.ApiStats{
+		CPUStats: container.CPUStats{
+			CPUUsage:    container.CPUUsage{TotalUsage: 1_000_000},
+			SystemUsage: 20_000_000,
+		},
+	}
+
+	// Container counter went backwards.
+	assert.Equal(t, 0.0, stats.CalculateCpuPercentLinux(2_000_000, 10_000_000))
+	// System counter went backwards.
+	assert.Equal(t, 0.0, stats.CalculateCpuPercentLinux(500_000, 30_000_000))
+	// A normal forward sample is unaffected: 500000 / 10000000 * 100 = 5%.
+	assert.InDelta(t, 5.0, stats.CalculateCpuPercentLinux(500_000, 10_000_000), 0.001)
+}
+
+func TestCpuPercentageWindowsHandlesCounterRollback(t *testing.T) {
+	now := time.Now()
+	stats := &container.ApiStats{
+		Read:     now,
+		NumProcs: 4,
+		CPUStats: container.CPUStats{
+			CPUUsage: container.CPUUsage{TotalUsage: 1_000_000},
+		},
+	}
+	prevRead := now.Add(-time.Second)
+
+	// Container counter went backwards.
+	assert.Equal(t, 0.0, stats.CalculateCpuPercentWindows(2_000_000, prevRead))
+	// A normal forward sample is unaffected.
+	assert.Greater(t, stats.CalculateCpuPercentWindows(500_000, prevRead), 0.0)
+}
+
 func TestNetworkStatsCalculationWithRealData(t *testing.T) {
 	// Create synthetic test data to avoid timing issues
 	apiStats1 := &container.ApiStats{
@@ -768,23 +1077,22 @@ func TestNetworkStatsCalculationWithRealData(t *testing.T) {
 	dm := &dockerManager{
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 	}
 
 	ctr := &container.ApiInfo{IdShort: "test-container"}
 	cacheTimeMs := uint16(30000) // Test with 30 second cache
 
-	// Use exact timing for deterministic results
-	exactly1000msAgo := time.Now().Add(-1000 * time.Millisecond)
-	stats := &container.Stats{
-		PrevReadTime: exactly1000msAgo,
-	}
-
-	// First call sets baseline
-	sent1, recv1 := dm.calculateNetworkStats(ctr, apiStats1, stats, true, "test", cacheTimeMs)
+	// First call sets baseline (no previous read time, so rates should be 0)
+	sent1, recv1 := dm.calculateNetworkStats(ctr, apiStats1, "test", cacheTimeMs)
 	assert.Equal(t, uint64(0), sent1)
 	assert.Equal(t, uint64(0), recv1)
 
-	// Cycle to establish baseline for this cache time
+	// Record read time and cycle to establish baseline for this cache time
+	exactly1000msAgo := time.Now().Add(-1000 * time.Millisecond)
+	dm.lastNetworkReadTime[cacheTimeMs] = map[string]time.Time{
+		"test-container": exactly1000msAgo,
+	}
 	dm.cycleNetworkDeltasForCacheTime(cacheTimeMs)
 
 	// Calculate expected results precisely
@@ -795,7 +1103,7 @@ func TestNetworkStatsCalculationWithRealData(t *testing.T) {
 	expectedRecvRate := deltaRecv * 1000 / expectedElapsedMs // Should be exactly 1000000
 
 	// Second call with changed data
-	sent2, recv2 := dm.calculateNetworkStats(ctr, apiStats2, stats, true, "test", cacheTimeMs)
+	sent2, recv2 := dm.calculateNetworkStats(ctr, apiStats2, "test", cacheTimeMs)
 
 	// Should be exactly the expected rates (no tolerance needed)
 	assert.Equal(t, expectedSentRate, sent2)
@@ -803,12 +1111,13 @@ func TestNetworkStatsCalculationWithRealData(t *testing.T) {
 
 	// Bad speed cap: set absurd delta over 1ms and expect 0 due to cap
 	dm.cycleNetworkDeltasForCacheTime(cacheTimeMs)
-	stats.PrevReadTime = time.Now().Add(-1 * time.Millisecond)
+	dm.lastNetworkReadTime[cacheTimeMs]["test-container"] = time.Now().Add(-1 * time.Millisecond)
 	apiStats1.Networks["eth0"] = container.NetworkStats{TxBytes: 0, RxBytes: 0}
 	apiStats2.Networks["eth0"] = container.NetworkStats{TxBytes: 10 * 1024 * 1024 * 1024, RxBytes: 0} // 10GB delta
-	_, _ = dm.calculateNetworkStats(ctr, apiStats1, stats, true, "test", cacheTimeMs)                 // baseline
+	_, _ = dm.calculateNetworkStats(ctr, apiStats1, "test", cacheTimeMs)                              // baseline
 	dm.cycleNetworkDeltasForCacheTime(cacheTimeMs)
-	sent3, recv3 := dm.calculateNetworkStats(ctr, apiStats2, stats, true, "test", cacheTimeMs)
+	dm.lastNetworkReadTime[cacheTimeMs]["test-container"] = time.Now().Add(-1 * time.Millisecond)
+	sent3, recv3 := dm.calculateNetworkStats(ctr, apiStats2, "test", cacheTimeMs)
 	assert.Equal(t, uint64(0), sent3)
 	assert.Equal(t, uint64(0), recv3)
 }
@@ -829,6 +1138,7 @@ func TestContainerStatsEndToEndWithRealData(t *testing.T) {
 		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 		containerStatsMap:   make(map[string]*container.Stats),
 	}
 
@@ -950,6 +1260,7 @@ func TestDockerStatsWorkflow(t *testing.T) {
 		lastCpuSystem:       make(map[uint16]map[string]uint64),
 		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
 		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
 		containerStatsMap:   make(map[string]*container.Stats),
 	}
 
@@ -1129,6 +1440,18 @@ func TestParseDockerStatus(t *testing.T) {
 			expectedStatus: "",
 			expectedHealth: container.DockerHealthNone,
 		},
+		{
+			name:           "status health with health: prefix",
+			input:          "Up 5 minutes (health: starting)",
+			expectedStatus: "Up 5 minutes",
+			expectedHealth: container.DockerHealthStarting,
+		},
+		{
+			name:           "status health with health status: prefix",
+			input:          "Up 10 minutes (health status: unhealthy)",
+			expectedStatus: "Up 10 minutes",
+			expectedHealth: container.DockerHealthUnhealthy,
+		},
 	}
 
 	for _, tt := range tests {
@@ -1140,6 +1463,85 @@ func TestParseDockerStatus(t *testing.T) {
 	}
 }
 
+func TestParseDockerHealthStatus(t *testing.T) {
+	tests := []struct {
+		input          string
+		expectedHealth container.DockerHealth
+		expectedOk     bool
+	}{
+		{"healthy", container.DockerHealthHealthy, true},
+		{"unhealthy", container.DockerHealthUnhealthy, true},
+		{"starting", container.DockerHealthStarting, true},
+		{"none", container.DockerHealthNone, true},
+		{" Healthy ", container.DockerHealthHealthy, true},
+		{"unknown", container.DockerHealthNone, false},
+		{"", container.DockerHealthNone, false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.input, func(t *testing.T) {
+			health, ok := parseDockerHealthStatus(tt.input)
+			assert.Equal(t, tt.expectedHealth, health)
+			assert.Equal(t, tt.expectedOk, ok)
+		})
+	}
+}
+
+func TestUpdateContainerStatsUsesPodmanInspectHealthFallback(t *testing.T) {
+	var requestedPaths []string
+	dm := &dockerManager{
+		client: &http.Client{Transport: roundTripFunc(func(req *http.Request) (*http.Response, error) {
+			requestedPaths = append(requestedPaths, req.URL.EscapedPath())
+			switch req.URL.EscapedPath() {
+			case "/containers/0123456789ab/stats":
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Status:     "200 OK",
+					Header:     make(http.Header),
+					Body: io.NopCloser(strings.NewReader(`{
+						"read":"2026-03-15T21:26:59Z",
+						"cpu_stats":{"cpu_usage":{"total_usage":1000},"system_cpu_usage":2000},
+						"memory_stats":{"usage":1048576,"stats":{"inactive_file":262144}},
+						"networks":{"eth0":{"rx_bytes":0,"tx_bytes":0}}
+					}`)),
+					Request: req,
+				}, nil
+			case "/containers/0123456789ab/json":
+				return &http.Response{
+					StatusCode: http.StatusOK,
+					Status:     "200 OK",
+					Header:     make(http.Header),
+					Body:       io.NopCloser(strings.NewReader(`{"State":{"Health":{"Status":"healthy"}}}`)),
+					Request:    req,
+				}, nil
+			default:
+				return nil, fmt.Errorf("unexpected path: %s", req.URL.EscapedPath())
+			}
+		})},
+		containerStatsMap:   make(map[string]*container.Stats),
+		apiStats:            &container.ApiStats{},
+		usingPodman:         true,
+		lastCpuContainer:    make(map[uint16]map[string]uint64),
+		lastCpuSystem:       make(map[uint16]map[string]uint64),
+		lastCpuReadTime:     make(map[uint16]map[string]time.Time),
+		networkSentTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		networkRecvTrackers: make(map[uint16]*deltatracker.DeltaTracker[string, uint64]),
+		lastNetworkReadTime: make(map[uint16]map[string]time.Time),
+	}
+
+	ctr := &container.ApiInfo{
+		IdShort: "0123456789ab",
+		Names:   []string{"/beszel"},
+		Status:  "Up 2 minutes",
+		Image:   "beszel:latest",
+	}
+
+	err := dm.updateContainerStats(ctr, defaultCacheTimeMs)
+	require.NoError(t, err)
+	assert.Equal(t, []string{"/containers/0123456789ab/stats", "/containers/0123456789ab/json"}, requestedPaths)
+	assert.Equal(t, container.DockerHealthHealthy, dm.containerStatsMap[ctr.IdShort].Health)
+	assert.Equal(t, "Up 2 minutes", dm.containerStatsMap[ctr.IdShort].Status)
+}
+
 func TestConstantsAndUtilityFunctions(t *testing.T) {
 	// Test constants are properly defined
 	assert.Equal(t, uint16(60000), defaultCacheTimeMs)
@@ -1455,3 +1857,123 @@ func TestAnsiEscapePattern(t *testing.T) {
 		})
 	}
 }
+
+func TestConvertContainerPortsToString(t *testing.T) {
+	type port = struct {
+		PublicPort uint16
+		IP         string
+	}
+	tests := []struct {
+		name     string
+		ports    []port
+		expected string
+	}{
+		{
+			name:     "empty ports",
+			ports:    nil,
+			expected: "",
+		},
+		{
+			name: "single port",
+			ports: []port{
+				{PublicPort: 80, IP: "0.0.0.0"},
+			},
+			expected: "80",
+		},
+		{
+			name: "single port with non-default IP",
+			ports: []port{
+				{PublicPort: 80, IP: "1.2.3.4"},
+			},
+			expected: "1.2.3.4:80",
+		},
+		{
+			name: "ipv6 default ip",
+			ports: []port{
+				{PublicPort: 80, IP: "::"},
+			},
+			expected: "80",
+		},
+		{
+			name: "zero PublicPort is skipped",
+			ports: []port{
+				{PublicPort: 0, IP: "0.0.0.0"},
+				{PublicPort: 80, IP: "0.0.0.0"},
+			},
+			expected: "80",
+		},
+		{
+			name: "ports sorted ascending by PublicPort",
+			ports: []port{
+				{PublicPort: 443, IP: "0.0.0.0"},
+				{PublicPort: 80, IP: "0.0.0.0"},
+				{PublicPort: 8080, IP: "0.0.0.0"},
+			},
+			expected: "80, 443, 8080",
+		},
+		{
+			name: "duplicates are deduplicated",
+			ports: []port{
+				{PublicPort: 80, IP: "0.0.0.0"},
+				{PublicPort: 80, IP: "0.0.0.0"},
+				{PublicPort: 443, IP: "0.0.0.0"},
+			},
+			expected: "80, 443",
+		},
+		{
+			name: "ipv4 and ipv6 wildcard bindings are deduplicated",
+			ports: []port{
+				{PublicPort: 80, IP: "0.0.0.0"},
+				{PublicPort: 80, IP: "::"},
+			},
+			expected: "80",
+		},
+		{
+			name: "multiple ports with different IPs",
+			ports: []port{
+				{PublicPort: 80, IP: "0.0.0.0"},
+				{PublicPort: 443, IP: "1.2.3.4"},
+			},
+			expected: "80, 1.2.3.4:443",
+		},
+		{
+			name: "same port bound to multiple IPs shows all entries",
+			ports: []port{
+				{PublicPort: 65533, IP: "172.16.151.72"},
+				{PublicPort: 65533, IP: "172.16.156.25"},
+			},
+			expected: "172.16.151.72:65533, 172.16.156.25:65533",
+		},
+		{
+			name: "same port bound to IPv4 and IPv6",
+			ports: []port{
+				{PublicPort: 65534, IP: "172.16.151.72"},
+				{PublicPort: 65534, IP: "fd04:38e2:98c6:3fd::72"},
+			},
+			expected: "172.16.151.72:65534, fd04:38e2:98c6:3fd::72:65534",
+		},
+		{
+			name: "ports slice is nilled after call",
+			ports: []port{
+				{PublicPort: 8080, IP: "0.0.0.0"},
+			},
+			expected: "8080",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ctr := &container.ApiInfo{}
+			for _, p := range tt.ports {
+				ctr.Ports = append(ctr.Ports, struct {
+					PublicPort uint16
+					IP         string
+				}{PublicPort: p.PublicPort, IP: p.IP})
+			}
+			result := convertContainerPortsToString(ctr)
+			assert.Equal(t, tt.expected, result)
+			// Ports slice must be cleared to prevent bleed-over into the next response
+			assert.Nil(t, ctr.Ports, "ctr.Ports should be nil after formatContainerPorts")
+		})
+	}
+}

agent/fans.go

@@ -0,0 +1,101 @@
+package agent
+
+import (
+	"log/slog"
+	"os"
+	"path/filepath"
+	"strings"
+	"sync"
+
+	"github.com/henrygd/beszel/agent/utils"
+	"github.com/henrygd/beszel/internal/entities/system"
+)
+
+type fanSensor struct {
+	key, path string
+}
+
+var getFanSensors = newFanSensorCache(hwmonRoot)
+
+func newFanSensorCache(root string) func() ([]fanSensor, error) {
+	return sync.OnceValues(func() ([]fanSensor, error) {
+		return discoverHwmonFans(root)
+	})
+}
+
+// updateFans populates systemStats.Fans from the host's hwmon sysfs tree.
+// No-op on platforms where hwmon isn't available (see fans_other.go).
+func (a *Agent) updateFans(systemStats *system.Stats) {
+	if hwmonRoot == "" {
+		return
+	}
+	sensors, err := getFanSensors()
+	if err != nil {
+		slog.Debug("Error reading fans", "err", err)
+		return
+	}
+	fans := readFanSensors(sensors)
+	if len(fans) == 0 {
+		return
+	}
+	systemStats.Fans = fans
+	// Note: Commented out because we don't currently use this value in the UI.
+	// Compute the single "dashboard" value used by the FanSpeed alert.
+	// Per-sensor RPMs live in Stats.Fans and drive the multi-line FanChart
+	// in the UI; the alert path only needs one number to compare against
+	// the user's threshold, so we use the highest RPM across all fans
+	// a.systemInfo.DashboardFan = 0
+	// for _, rpm := range fans {
+	// 	if rpm > a.systemInfo.DashboardFan {
+	// 		a.systemInfo.DashboardFan = rpm
+	// 	}
+	// }
+}
+
+// readHwmonFans walks the given hwmon root (typically /sys/class/hwmon) and
+// returns a map of "<chip>_<label-or-fan-idx>" → RPM for every fan*_input
+// file it finds. Zero RPM is retained because it can represent a real fan that
+// has stopped; negative and malformed readings are ignored.
+func readHwmonFans(root string) (map[string]uint16, error) {
+	sensors, err := discoverHwmonFans(root)
+	if err != nil {
+		return nil, err
+	}
+	return readFanSensors(sensors), nil
+}
+
+func discoverHwmonFans(root string) ([]fanSensor, error) {
+	entries, err := os.ReadDir(root)
+	if err != nil {
+		return nil, err
+	}
+	var sensors []fanSensor
+	for _, entry := range entries {
+		chipDir := filepath.Join(root, entry.Name())
+		chipName := utils.ReadStringFile(filepath.Join(chipDir, "name"))
+		if chipName == "" {
+			chipName = entry.Name()
+		}
+		inputs, _ := filepath.Glob(filepath.Join(chipDir, "fan*_input"))
+		for _, inputPath := range inputs {
+			base := strings.TrimSuffix(filepath.Base(inputPath), "_input")
+			label := utils.ReadStringFile(filepath.Join(chipDir, base+"_label"))
+			key := chipName + "_" + base
+			if label != "" {
+				key = chipName + "_" + label
+			}
+			sensors = append(sensors, fanSensor{key, inputPath})
+		}
+	}
+	return sensors, nil
+}
+
+func readFanSensors(sensors []fanSensor) map[string]uint16 {
+	fans := make(map[string]uint16, len(sensors))
+	for _, sensor := range sensors {
+		if rpm, ok := utils.ReadUintFile(sensor.path); ok {
+			fans[sensor.key] = uint16(rpm)
+		}
+	}
+	return fans
+}

agent/fans_linux.go

@@ -0,0 +1,8 @@
+//go:build linux
+
+package agent
+
+// hwmonRoot is the sysfs entry point for hardware monitor chips. Each
+// subdirectory (hwmon0, hwmon1, …) is one chip; fan*_input files inside it
+// expose RPM readings.
+const hwmonRoot = "/sys/class/hwmon"

agent/fans_other.go

@@ -0,0 +1,7 @@
+//go:build !linux
+
+package agent
+
+// hwmonRoot is empty on non-Linux platforms — fan RPM reporting via sysfs
+// hwmon is Linux-specific. updateFans() short-circuits when this is empty.
+const hwmonRoot = ""

agent/fans_test.go

@@ -0,0 +1,87 @@
+//go:build testing
+
+package agent
+
+import (
+	"os"
+	"path/filepath"
+	"testing"
+
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+// writeFile creates path with parents and writes contents.
+func writeFile(t *testing.T, path, contents string) {
+	t.Helper()
+	require.NoError(t, os.MkdirAll(filepath.Dir(path), 0o755))
+	require.NoError(t, os.WriteFile(path, []byte(contents), 0o644))
+}
+
+// TestReadHwmonFans verifies the /sys/class/hwmon walker:
+//   - picks up fan*_input from every chip,
+//   - keys entries by chip name + sensor label (or fan idx if no label),
+//   - retains 0 RPM for stopped fans,
+//   - tolerates chips with no fan files at all.
+func TestReadHwmonFans(t *testing.T) {
+	root := t.TempDir()
+
+	// hwmon0: Raspberry Pi 5 active cooler — one fan, no label.
+	writeFile(t, filepath.Join(root, "hwmon0", "name"), "pwmfan\n")
+	writeFile(t, filepath.Join(root, "hwmon0", "fan1_input"), "6500\n")
+
+	// hwmon1: a thermal-only chip, no fan files. Must not error.
+	writeFile(t, filepath.Join(root, "hwmon1", "name"), "cpu_thermal\n")
+	writeFile(t, filepath.Join(root, "hwmon1", "temp1_input"), "55000\n")
+
+	// hwmon2: two fans — one stopped (0 RPM) and one labeled "chassis".
+	writeFile(t, filepath.Join(root, "hwmon2", "name"), "nct6798\n")
+	writeFile(t, filepath.Join(root, "hwmon2", "fan1_input"), "0\n")
+	writeFile(t, filepath.Join(root, "hwmon2", "fan2_input"), "1200\n")
+	writeFile(t, filepath.Join(root, "hwmon2", "fan2_label"), "chassis\n")
+
+	fans, err := readHwmonFans(root)
+	require.NoError(t, err)
+
+	assert.Equal(t, map[string]uint16{
+		"pwmfan_fan1":     6500,
+		"nct6798_fan1":    0,
+		"nct6798_chassis": 1200,
+	}, fans)
+}
+
+// TestReadHwmonFansMissingRoot returns an error rather than panicking when the
+// hwmon root doesn't exist (e.g. running on a kernel without hwmon support).
+func TestReadHwmonFansMissingRoot(t *testing.T) {
+	_, err := readHwmonFans(filepath.Join(t.TempDir(), "does-not-exist"))
+	assert.Error(t, err)
+}
+
+// TestReadHwmonFansEmpty returns an empty map (not nil error) when the root
+// exists but contains no chips at all.
+func TestReadHwmonFansEmpty(t *testing.T) {
+	root := t.TempDir()
+	fans, err := readHwmonFans(root)
+	require.NoError(t, err)
+	assert.Empty(t, fans)
+}
+
+func TestFanDiscoveryCache(t *testing.T) {
+	root := t.TempDir()
+	input := filepath.Join(root, "hwmon0", "fan1_input")
+	writeFile(t, filepath.Join(root, "hwmon0", "name"), "chip\n")
+	writeFile(t, input, "1000\n")
+
+	getSensors := newFanSensorCache(root)
+	sensors, err := getSensors()
+	require.NoError(t, err)
+	fans := readFanSensors(sensors)
+	assert.Equal(t, uint16(1000), fans["chip_fan1"])
+
+	writeFile(t, input, "1200\n")
+	writeFile(t, filepath.Join(root, "hwmon0", "fan1_label"), "case\n")
+	sensors, err = getSensors()
+	require.NoError(t, err)
+	fans = readFanSensors(sensors)
+	assert.Equal(t, map[string]uint16{"chip_fan1": 1200}, fans)
+}

agent/gpu.go

@@ -461,7 +461,7 @@ func (gm *GPUManager) discoverGpuCapabilities() gpuCapabilities {
 		caps.hasNvtop = true
 	}
 	if runtime.GOOS == "darwin" {
-		if _, err := exec.LookPath(macmonCmd); err == nil {
+		if _, err := utils.LookPathHomebrew(macmonCmd); err == nil {
 			caps.hasMacmon = true
 		}
 		if _, err := exec.LookPath(powermetricsCmd); err == nil {
@@ -542,7 +542,7 @@ func (gm *GPUManager) collectorDefinitions(caps gpuCapabilities) map[collectorSo
 	return map[collectorSource]collectorDefinition{
 		collectorSourceNVML: {
 			group:     collectorGroupNvidia,
-			available: caps.hasNvidiaSmi,
+			available: true,
 			start: func(_ func()) bool {
 				return gm.startNvmlCollector()
 			},
@@ -734,9 +734,6 @@ func NewGPUManager() (*GPUManager, error) {
 	}
 	var gm GPUManager
 	caps := gm.discoverGpuCapabilities()
-	if !hasAnyGpuCollector(caps) {
-		return nil, fmt.Errorf(noGPUFoundMsg)
-	}
 	gm.GpuDataMap = make(map[string]*system.GPUData)
 
 	// Jetson devices should always use tegrastats (ignore GPU_COLLECTOR).
@@ -745,7 +742,7 @@ func NewGPUManager() (*GPUManager, error) {
 		return &gm, nil
 	}
 
-	// if GPU_COLLECTOR is set, start user-defined collectors.
+	// Respect explicit collector selection before capability auto-detection.
 	if collectorConfig, ok := utils.GetEnv("GPU_COLLECTOR"); ok && strings.TrimSpace(collectorConfig) != "" {
 		priorities := parseCollectorPriority(collectorConfig)
 		if gm.startCollectorsByPriority(priorities, caps) == 0 {
@@ -754,6 +751,10 @@ func NewGPUManager() (*GPUManager, error) {
 		return &gm, nil
 	}
 
+	if !hasAnyGpuCollector(caps) {
+		return nil, fmt.Errorf(noGPUFoundMsg)
+	}
+
 	// auto-detect and start collectors when GPU_COLLECTOR is unset.
 	if gm.startCollectorsByPriority(gm.resolveLegacyCollectorPriority(caps), caps) == 0 {
 		return nil, fmt.Errorf(noGPUFoundMsg)

agent/gpu_amd_linux.go

@@ -156,6 +156,7 @@ func (gm *GPUManager) updateAmdGpuData(cardPath string) bool {
 func readSysfsFloat(path string) (float64, error) {
 	val, err := utils.ReadStringFileLimited(path, 64)
 	if err != nil {
+		slog.Debug("Failed to read sysfs value", "path", path, "error", err)
 		return 0, err
 	}
 	return strconv.ParseFloat(val, 64)

agent/gpu_darwin.go

@@ -13,6 +13,7 @@ import (
 	"strings"
 	"time"
 
+	"github.com/henrygd/beszel/agent/utils"
 	"github.com/henrygd/beszel/internal/entities/system"
 )
 
@@ -171,7 +172,11 @@ type macmonSample struct {
 }
 
 func (gm *GPUManager) collectMacmonPipe() (err error) {
-	cmd := exec.Command(macmonCmd, "pipe", "-i", strconv.Itoa(macmonIntervalMs))
+	macmonPath, err := utils.LookPathHomebrew(macmonCmd)
+	if err != nil {
+		return err
+	}
+	cmd := exec.Command(macmonPath, "pipe", "-i", strconv.Itoa(macmonIntervalMs))
 	// Avoid blocking if macmon writes to stderr.
 	cmd.Stderr = io.Discard
 	stdout, err := cmd.StdoutPipe()

agent/gpu_test.go

@@ -1083,8 +1083,6 @@ func TestCalculateGPUAverage(t *testing.T) {
 
 func TestGPUCapabilitiesAndLegacyPriority(t *testing.T) {
 	// Save original PATH
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
 	hasAmdSysfs := (&GPUManager{}).hasAmdSysfs()
 
 	tests := []struct {
@@ -1178,7 +1176,7 @@ echo "[]"`
 		{
 			name: "no gpu tools available",
 			setupCommands: func(_ string) error {
-				os.Setenv("PATH", "")
+				t.Setenv("PATH", "")
 				return nil
 			},
 			wantErr: true,
@@ -1188,7 +1186,7 @@ echo "[]"`
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
 			tempDir := t.TempDir()
-			os.Setenv("PATH", tempDir)
+			t.Setenv("PATH", tempDir)
 			if err := tt.setupCommands(tempDir); err != nil {
 				t.Fatal(err)
 			}
@@ -1234,13 +1232,9 @@ echo "[]"`
 }
 
 func TestCollectorStartHelpers(t *testing.T) {
-	// Save original PATH
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	// Set up temp dir with the commands
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 
 	tests := []struct {
 		name     string
@@ -1370,11 +1364,8 @@ echo '[{"device_name":"NVIDIA Test GPU","temp":"52C","power_draw":"31W","gpu_uti
 }
 
 func TestNewGPUManagerPriorityNvtopFallback(t *testing.T) {
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 	t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "nvtop,nvidia-smi")
 
 	nvtopPath := filepath.Join(dir, "nvtop")
@@ -1399,11 +1390,8 @@ echo "0, NVIDIA Priority GPU, 45, 512, 2048, 12, 25"`
 }
 
 func TestNewGPUManagerPriorityMixedCollectors(t *testing.T) {
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 	t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "intel_gpu_top,rocm-smi")
 
 	intelPath := filepath.Join(dir, "intel_gpu_top")
@@ -1433,11 +1421,8 @@ echo '{"card0": {"Temperature (Sensor edge) (C)": "49.0", "Current Socket Graphi
 }
 
 func TestNewGPUManagerPriorityNvmlFallbackToNvidiaSmi(t *testing.T) {
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 	t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "nvml,nvidia-smi")
 
 	nvidiaPath := filepath.Join(dir, "nvidia-smi")
@@ -1456,11 +1441,8 @@ echo "0, NVIDIA Fallback GPU, 41, 256, 1024, 8, 14"`
 }
 
 func TestNewGPUManagerConfiguredCollectorsMustStart(t *testing.T) {
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 
 	t.Run("configured valid collector unavailable", func(t *testing.T) {
 		t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "nvidia-smi")
@@ -1479,12 +1461,28 @@ func TestNewGPUManagerConfiguredCollectorsMustStart(t *testing.T) {
 	})
 }
 
-func TestNewGPUManagerJetsonIgnoresCollectorConfig(t *testing.T) {
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
+func TestCollectorDefinitionsNvmlDoesNotRequireNvidiaSmi(t *testing.T) {
+	gm := &GPUManager{}
+	definitions := gm.collectorDefinitions(gpuCapabilities{})
+	require.Contains(t, definitions, collectorSourceNVML)
+	assert.True(t, definitions[collectorSourceNVML].available)
+}
 
+func TestNewGPUManagerConfiguredNvmlBypassesCapabilityGate(t *testing.T) {
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
+	t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "nvml")
+
+	gm, err := NewGPUManager()
+	require.Nil(t, gm)
+	require.Error(t, err)
+	assert.Contains(t, err.Error(), "no configured GPU collectors are available")
+	assert.NotContains(t, err.Error(), noGPUFoundMsg)
+}
+
+func TestNewGPUManagerJetsonIgnoresCollectorConfig(t *testing.T) {
+	dir := t.TempDir()
+	t.Setenv("PATH", dir)
 	t.Setenv("BESZEL_AGENT_GPU_COLLECTOR", "nvidia-smi")
 
 	tegraPath := filepath.Join(dir, "tegrastats")
@@ -1719,12 +1717,8 @@ func TestIntelUpdateFromStats(t *testing.T) {
 }
 
 func TestIntelCollectorStreaming(t *testing.T) {
-	// Save and override PATH
-	origPath := os.Getenv("PATH")
-	defer os.Setenv("PATH", origPath)
-
 	dir := t.TempDir()
-	os.Setenv("PATH", dir)
+	t.Setenv("PATH", dir)
 
 	// Create a fake intel_gpu_top that prints -l format with four samples (first will be skipped) and exits
 	scriptPath := filepath.Join(dir, "intel_gpu_top")

agent/handlers.go

@@ -166,14 +166,16 @@ type GetSmartDataHandler struct{}
 
 func (h *GetSmartDataHandler) Handle(hctx *HandlerContext) error {
 	if hctx.Agent.smartManager == nil {
-		// return empty map to indicate no data
-		return hctx.SendResponse(map[string]smart.SmartData{}, hctx.RequestID)
+		return hctx.SendResponse(smart.SmartDataResponse{Data: map[string]smart.SmartData{}}, hctx.RequestID)
 	}
-	if err := hctx.Agent.smartManager.Refresh(false); err != nil {
+	complete, err := hctx.Agent.smartManager.Refresh(false)
+	if err != nil {
 		slog.Debug("smart refresh failed", "err", err)
 	}
-	data := hctx.Agent.smartManager.GetCurrentData()
-	return hctx.SendResponse(data, hctx.RequestID)
+	return hctx.SendResponse(smart.SmartDataResponse{
+		Data:     hctx.Agent.smartManager.GetCurrentData(),
+		Complete: complete,
+	}, hctx.RequestID)
 }
 
 ////////////////////////////////////////////////////////////////////////////

agent/handlers_test.go

@@ -7,6 +7,7 @@ import (
 
 	"github.com/fxamacker/cbor/v2"
 	"github.com/henrygd/beszel/internal/common"
+	"github.com/henrygd/beszel/internal/entities/smart"
 	"github.com/stretchr/testify/assert"
 )
 
@@ -17,6 +18,18 @@ type MockHandler struct {
 	handleFunc           func(ctx *HandlerContext) error
 }
 
+func TestNewAgentResponseSmartData(t *testing.T) {
+	response := newAgentResponse(smart.SmartDataResponse{
+		Data: map[string]smart.SmartData{
+			"AAA": {SerialNumber: "AAA"},
+		},
+		Complete: true,
+	}, nil)
+
+	assert.Equal(t, "AAA", response.SmartData["AAA"].SerialNumber)
+	assert.True(t, response.SmartComplete)
+}
+
 func (m *MockHandler) Handle(ctx *HandlerContext) error {
 	if m.handleFunc != nil {
 		return m.handleFunc(ctx)

agent/lhm/beszel_lhm.csproj

@@ -8,6 +8,6 @@
   </PropertyGroup>
 
   <ItemGroup>
-    <PackageReference Include="LibreHardwareMonitorLib" Version="0.9.5" />
+    <PackageReference Include="LibreHardwareMonitorLib" Version="0.9.6" />
   </ItemGroup>
 </Project>

agent/mdraid_linux.go

@@ -180,8 +180,13 @@ func mdraidSmartStatus(health mdraidHealth) string {
 	if health.degraded > 0 {
 		return "FAILED"
 	}
+	if health.mismatchCnt > 0 {
+		return "WARNING"
+	}
+	// "check" scans for consistency problems without repairing mismatches.
+	// With no mismatches, keep it green while reporting progress attributes.
 	switch syncAction {
-	case "check", "repair":
+	case "repair":
 		return "WARNING"
 	}
 	switch state {

agent/mdraid_linux_test.go

@@ -94,6 +94,18 @@ func TestMdraidSmartStatus(t *testing.T) {
 	if got := mdraidSmartStatus(mdraidHealth{arrayState: "active", syncAction: "recover"}); got != "WARNING" {
 		t.Fatalf("mdraidSmartStatus(recover) = %q, want WARNING", got)
 	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean", syncAction: "check"}); got != "PASSED" {
+		t.Fatalf("mdraidSmartStatus(clean+check) = %q, want PASSED", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean", syncAction: "check", mismatchCnt: 1}); got != "WARNING" {
+		t.Fatalf("mdraidSmartStatus(clean+check+mismatch) = %q, want WARNING", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean", mismatchCnt: 1}); got != "WARNING" {
+		t.Fatalf("mdraidSmartStatus(clean+mismatch) = %q, want WARNING", got)
+	}
+	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean", syncAction: "repair"}); got != "WARNING" {
+		t.Fatalf("mdraidSmartStatus(repair) = %q, want WARNING", got)
+	}
 	if got := mdraidSmartStatus(mdraidHealth{arrayState: "clean"}); got != "PASSED" {
 		t.Fatalf("mdraidSmartStatus(clean) = %q, want PASSED", got)
 	}

agent/response.go

@@ -21,6 +21,9 @@ func newAgentResponse(data any, requestID *uint32) common.AgentResponse {
 		response.String = &v
 	case map[string]smart.SmartData:
 		response.SmartData = v
+	case smart.SmartDataResponse:
+		response.SmartData = v.Data
+		response.SmartComplete = v.Complete
 	case systemd.ServiceDetails:
 		response.ServiceInfo = v
 	default:

agent/sensors.go

@@ -2,12 +2,14 @@ package agent
 
 import (
 	"context"
+	"errors"
 	"fmt"
 	"log/slog"
 	"path"
 	"runtime"
 	"strconv"
 	"strings"
+	"time"
 	"unicode/utf8"
 
 	"github.com/henrygd/beszel/agent/utils"
@@ -17,13 +19,20 @@ import (
 	"github.com/shirou/gopsutil/v4/sensors"
 )
 
+var errTemperatureFetchTimeout = errors.New("temperature collection timed out")
+
+// Matches sensors.TemperaturesWithContext to allow for panic recovery (gopsutil/issues/1832)
+type getTempsFn func(ctx context.Context) ([]sensors.TemperatureStat, error)
+
 type SensorConfig struct {
 	context        context.Context
 	sensors        map[string]struct{}
 	primarySensor  string
+	timeout        time.Duration
 	isBlacklist    bool
 	hasWildcards   bool
 	skipCollection bool
+	firstRun       bool
 }
 
 func (a *Agent) newSensorConfig() *SensorConfig {
@@ -31,20 +40,29 @@ func (a *Agent) newSensorConfig() *SensorConfig {
 	sysSensors, _ := utils.GetEnv("SYS_SENSORS")
 	sensorsEnvVal, sensorsSet := utils.GetEnv("SENSORS")
 	skipCollection := sensorsSet && sensorsEnvVal == ""
+	sensorsTimeout, _ := utils.GetEnv("SENSORS_TIMEOUT")
 
-	return a.newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal, skipCollection)
+	return a.newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal, sensorsTimeout, skipCollection)
 }
 
-// Matches sensors.TemperaturesWithContext to allow for panic recovery (gopsutil/issues/1832)
-type getTempsFn func(ctx context.Context) ([]sensors.TemperatureStat, error)
-
 // newSensorConfigWithEnv creates a SensorConfig with the provided environment variables
 // sensorsSet indicates if the SENSORS environment variable was explicitly set (even to empty string)
-func (a *Agent) newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal string, skipCollection bool) *SensorConfig {
+func (a *Agent) newSensorConfigWithEnv(primarySensor, sysSensors, sensorsEnvVal, sensorsTimeout string, skipCollection bool) *SensorConfig {
+	timeout := 2 * time.Second
+	if sensorsTimeout != "" {
+		if d, err := time.ParseDuration(sensorsTimeout); err == nil {
+			timeout = d
+		} else {
+			slog.Warn("Invalid SENSORS_TIMEOUT", "value", sensorsTimeout)
+		}
+	}
+
 	config := &SensorConfig{
 		context:        context.Background(),
 		primarySensor:  primarySensor,
+		timeout:        timeout,
 		skipCollection: skipCollection,
+		firstRun:       true,
 		sensors:        make(map[string]struct{}),
 	}
 
@@ -86,10 +104,12 @@ func (a *Agent) updateTemperatures(systemStats *system.Stats) {
 	// reset high temp
 	a.systemInfo.DashboardTemp = 0
 
-	temps, err := a.getTempsWithPanicRecovery(getSensorTemps)
+	temps, err := a.getTempsWithTimeout(getSensorTemps)
 	if err != nil {
 		// retry once on panic (gopsutil/issues/1832)
-		temps, err = a.getTempsWithPanicRecovery(getSensorTemps)
+		if !errors.Is(err, errTemperatureFetchTimeout) {
+			temps, err = a.getTempsWithTimeout(getSensorTemps)
+		}
 		if err != nil {
 			slog.Warn("Error updating temperatures", "err", err)
 			if len(systemStats.Temperatures) > 0 {
@@ -152,6 +172,34 @@ func (a *Agent) getTempsWithPanicRecovery(getTemps getTempsFn) (temps []sensors.
 	return
 }
 
+func (a *Agent) getTempsWithTimeout(getTemps getTempsFn) ([]sensors.TemperatureStat, error) {
+	type result struct {
+		temps []sensors.TemperatureStat
+		err   error
+	}
+
+	// Use a longer timeout on the first run to allow for initialization
+	// (e.g. Windows LHM subprocess startup)
+	timeout := a.sensorConfig.timeout
+	if a.sensorConfig.firstRun {
+		a.sensorConfig.firstRun = false
+		timeout = 10 * time.Second
+	}
+
+	resultCh := make(chan result, 1)
+	go func() {
+		temps, err := a.getTempsWithPanicRecovery(getTemps)
+		resultCh <- result{temps: temps, err: err}
+	}()
+
+	select {
+	case res := <-resultCh:
+		return res.temps, res.err
+	case <-time.After(timeout):
+		return nil, errTemperatureFetchTimeout
+	}
+}
+
 // isValidSensor checks if a sensor is valid based on the sensor name and the sensor config
 func isValidSensor(sensorName string, config *SensorConfig) bool {
 	// if no sensors configured, everything is valid

agent/sensors_test.go

@@ -5,8 +5,8 @@ package agent
 import (
 	"context"
 	"fmt"
-	"os"
 	"testing"
+	"time"
 
 	"github.com/henrygd/beszel/internal/entities/system"
 
@@ -168,6 +168,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 		primarySensor  string
 		sysSensors     string
 		sensors        string
+		sensorsTimeout string
 		skipCollection bool
 		expectedConfig *SensorConfig
 	}{
@@ -179,12 +180,37 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:        context.Background(),
 				primarySensor:  "",
+				timeout:        2 * time.Second,
 				sensors:        map[string]struct{}{},
 				isBlacklist:    false,
 				hasWildcards:   false,
 				skipCollection: false,
 			},
 		},
+		{
+			name:           "Custom timeout",
+			primarySensor:  "",
+			sysSensors:     "",
+			sensors:        "",
+			sensorsTimeout: "5s",
+			expectedConfig: &SensorConfig{
+				context: context.Background(),
+				timeout: 5 * time.Second,
+				sensors: map[string]struct{}{},
+			},
+		},
+		{
+			name:           "Invalid timeout falls back to default",
+			primarySensor:  "",
+			sysSensors:     "",
+			sensors:        "",
+			sensorsTimeout: "notaduration",
+			expectedConfig: &SensorConfig{
+				context: context.Background(),
+				timeout: 2 * time.Second,
+				sensors: map[string]struct{}{},
+			},
+		},
 		{
 			name:           "Explicitly set to empty string",
 			primarySensor:  "",
@@ -194,6 +220,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:        context.Background(),
 				primarySensor:  "",
+				timeout:        2 * time.Second,
 				sensors:        map[string]struct{}{},
 				isBlacklist:    false,
 				hasWildcards:   false,
@@ -208,6 +235,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:       context.Background(),
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors:       map[string]struct{}{},
 				isBlacklist:   false,
 				hasWildcards:  false,
@@ -221,6 +249,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:       context.Background(),
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors: map[string]struct{}{
 					"cpu_temp": {},
 					"gpu_temp": {},
@@ -237,6 +266,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:       context.Background(),
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors: map[string]struct{}{
 					"cpu_temp": {},
 					"gpu_temp": {},
@@ -253,6 +283,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:       context.Background(),
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors: map[string]struct{}{
 					"cpu_*":    {},
 					"gpu_temp": {},
@@ -269,6 +300,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			expectedConfig: &SensorConfig{
 				context:       context.Background(),
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors: map[string]struct{}{
 					"cpu_*":    {},
 					"gpu_temp": {},
@@ -284,6 +316,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			sensors:       "cpu_temp",
 			expectedConfig: &SensorConfig{
 				primarySensor: "cpu_temp",
+				timeout:       2 * time.Second,
 				sensors: map[string]struct{}{
 					"cpu_temp": {},
 				},
@@ -295,7 +328,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			result := agent.newSensorConfigWithEnv(tt.primarySensor, tt.sysSensors, tt.sensors, tt.skipCollection)
+			result := agent.newSensorConfigWithEnv(tt.primarySensor, tt.sysSensors, tt.sensors, tt.sensorsTimeout, tt.skipCollection)
 
 			// Check primary sensor
 			assert.Equal(t, tt.expectedConfig.primarySensor, result.primarySensor)
@@ -314,6 +347,7 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 			// Check flags
 			assert.Equal(t, tt.expectedConfig.isBlacklist, result.isBlacklist)
 			assert.Equal(t, tt.expectedConfig.hasWildcards, result.hasWildcards)
+			assert.Equal(t, tt.expectedConfig.timeout, result.timeout)
 
 			// Check context
 			if tt.sysSensors != "" {
@@ -329,40 +363,18 @@ func TestNewSensorConfigWithEnv(t *testing.T) {
 }
 
 func TestNewSensorConfig(t *testing.T) {
-	// Save original environment variables
-	originalPrimary, hasPrimary := os.LookupEnv("BESZEL_AGENT_PRIMARY_SENSOR")
-	originalSys, hasSys := os.LookupEnv("BESZEL_AGENT_SYS_SENSORS")
-	originalSensors, hasSensors := os.LookupEnv("BESZEL_AGENT_SENSORS")
-
-	// Restore environment variables after the test
-	defer func() {
-		// Clean up test environment variables
-		os.Unsetenv("BESZEL_AGENT_PRIMARY_SENSOR")
-		os.Unsetenv("BESZEL_AGENT_SYS_SENSORS")
-		os.Unsetenv("BESZEL_AGENT_SENSORS")
-
-		// Restore original values if they existed
-		if hasPrimary {
-			os.Setenv("BESZEL_AGENT_PRIMARY_SENSOR", originalPrimary)
-		}
-		if hasSys {
-			os.Setenv("BESZEL_AGENT_SYS_SENSORS", originalSys)
-		}
-		if hasSensors {
-			os.Setenv("BESZEL_AGENT_SENSORS", originalSensors)
-		}
-	}()
-
 	// Set test environment variables
-	os.Setenv("BESZEL_AGENT_PRIMARY_SENSOR", "test_primary")
-	os.Setenv("BESZEL_AGENT_SYS_SENSORS", "/test/path")
-	os.Setenv("BESZEL_AGENT_SENSORS", "test_sensor1,test_*,test_sensor3")
+	t.Setenv("BESZEL_AGENT_PRIMARY_SENSOR", "test_primary")
+	t.Setenv("BESZEL_AGENT_SYS_SENSORS", "/test/path")
+	t.Setenv("BESZEL_AGENT_SENSORS", "test_sensor1,test_*,test_sensor3")
+	t.Setenv("BESZEL_AGENT_SENSORS_TIMEOUT", "7s")
 
 	agent := &Agent{}
 	result := agent.newSensorConfig()
 
 	// Verify results
 	assert.Equal(t, "test_primary", result.primarySensor)
+	assert.Equal(t, 7*time.Second, result.timeout)
 	assert.NotNil(t, result.sensors)
 	assert.Equal(t, 3, len(result.sensors))
 	assert.True(t, result.hasWildcards)
@@ -551,3 +563,59 @@ func TestGetTempsWithPanicRecovery(t *testing.T) {
 		})
 	}
 }
+
+func TestGetTempsWithTimeout(t *testing.T) {
+	agent := &Agent{
+		sensorConfig: &SensorConfig{
+			context: context.Background(),
+			timeout: 10 * time.Millisecond,
+		},
+	}
+
+	t.Run("returns temperatures before timeout", func(t *testing.T) {
+		temps, err := agent.getTempsWithTimeout(func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+			return []sensors.TemperatureStat{{SensorKey: "cpu_temp", Temperature: 42}}, nil
+		})
+
+		require.NoError(t, err)
+		require.Len(t, temps, 1)
+		assert.Equal(t, "cpu_temp", temps[0].SensorKey)
+	})
+
+	t.Run("returns timeout error when collector hangs", func(t *testing.T) {
+		temps, err := agent.getTempsWithTimeout(func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+			time.Sleep(50 * time.Millisecond)
+			return []sensors.TemperatureStat{{SensorKey: "cpu_temp", Temperature: 42}}, nil
+		})
+
+		assert.Nil(t, temps)
+		assert.ErrorIs(t, err, errTemperatureFetchTimeout)
+	})
+}
+
+func TestUpdateTemperaturesSkipsOnTimeout(t *testing.T) {
+	agent := &Agent{
+		systemInfo: system.Info{DashboardTemp: 99},
+		sensorConfig: &SensorConfig{
+			context: context.Background(),
+			timeout: 10 * time.Millisecond,
+		},
+	}
+
+	t.Cleanup(func() {
+		getSensorTemps = sensors.TemperaturesWithContext
+	})
+	getSensorTemps = func(ctx context.Context) ([]sensors.TemperatureStat, error) {
+		time.Sleep(50 * time.Millisecond)
+		return nil, nil
+	}
+
+	stats := &system.Stats{
+		Temperatures: map[string]float64{"stale": 50},
+	}
+
+	agent.updateTemperatures(stats)
+
+	assert.Equal(t, 0.0, agent.systemInfo.DashboardTemp)
+	assert.Equal(t, map[string]float64{}, stats.Temperatures)
+}

agent/server.go

@@ -29,9 +29,6 @@ type ServerOptions struct {
 	Keys    []gossh.PublicKey // SSH public keys for authentication
 }
 
-// hubVersions caches hub versions by session ID to avoid repeated parsing.
-var hubVersions map[string]semver.Version
-
 // StartServer starts the SSH server with the provided options.
 // It configures the server with secure defaults, sets up authentication,
 // and begins listening for connections. Returns an error if the server
@@ -99,24 +96,15 @@ func (a *Agent) StartServer(opts ServerOptions) error {
 	return a.server.Serve(ln)
 }
 
-// getHubVersion retrieves and caches the hub version for a given session.
-// It extracts the version from the SSH client version string and caches
-// it to avoid repeated parsing. Returns a zero version if parsing fails.
-func (a *Agent) getHubVersion(sessionId string, sessionCtx ssh.Context) semver.Version {
-	if hubVersions == nil {
-		hubVersions = make(map[string]semver.Version, 1)
-	}
-	hubVersion, ok := hubVersions[sessionId]
-	if ok {
-		return hubVersion
-	}
-	// Extract hub version from SSH client version
+// getHubVersion extracts the hub version from the SSH client version string
+// for a given session. Returns a zero version if parsing fails.
+func (a *Agent) getHubVersion(sessionCtx ssh.Context) semver.Version {
 	clientVersion := sessionCtx.Value(ssh.ContextKeyClientVersion)
 	if versionStr, ok := clientVersion.(string); ok {
-		hubVersion, _ = extractHubVersion(versionStr)
+		hubVersion, _ := extractHubVersion(versionStr)
+		return hubVersion
 	}
-	hubVersions[sessionId] = hubVersion
-	return hubVersion
+	return semver.Version{}
 }
 
 // handleSession handles an incoming SSH session by gathering system statistics
@@ -127,9 +115,8 @@ func (a *Agent) handleSession(s ssh.Session) {
 	a.connectionManager.eventChan <- SSHConnect
 
 	sessionCtx := s.Context()
-	sessionID := sessionCtx.SessionID()
 
-	hubVersion := a.getHubVersion(sessionID, sessionCtx)
+	hubVersion := a.getHubVersion(sessionCtx)
 
 	// Legacy one-shot behavior for older hubs
 	if hubVersion.LT(beszel.MinVersionAgentResponse) {
@@ -193,7 +180,7 @@ func (a *Agent) handleSSHRequest(w io.Writer, req *common.HubRequest[cbor.RawMes
 
 // handleLegacyStats serves the legacy one-shot stats payload for older hubs
 func (a *Agent) handleLegacyStats(w io.Writer, hubVersion semver.Version) error {
-	stats := a.gatherStats(common.DataRequestOptions{CacheTimeMs: 60_000})
+	stats := a.gatherStats(common.DataRequestOptions{CacheTimeMs: defaultDataCacheTimeMs})
 	return a.writeToSession(w, stats, hubVersion)
 }
 

agent/server_test.go

@@ -183,8 +183,7 @@ func TestStartServer(t *testing.T) {
 }
 
 func TestStartServerDisableSSH(t *testing.T) {
-	os.Setenv("BESZEL_AGENT_DISABLE_SSH", "true")
-	defer os.Unsetenv("BESZEL_AGENT_DISABLE_SSH")
+	t.Setenv("BESZEL_AGENT_DISABLE_SSH", "true")
 
 	agent, err := NewAgent("")
 	require.NoError(t, err)
@@ -405,27 +404,23 @@ func TestGetHubVersion(t *testing.T) {
 		clientVersion: "SSH-2.0-beszel_0.12.0",
 	}
 
-	// Test first call - should extract and cache version
-	version := agent.getHubVersion("test-session-123", mockCtx)
+	// Test first call - should extract version
+	version := agent.getHubVersion(mockCtx)
 	assert.Equal(t, "0.12.0", version.String())
 
-	// Test second call - should return cached version
-	mockCtx.clientVersion = "SSH-2.0-beszel_0.11.0" // Change version but should still return cached
-	version = agent.getHubVersion("test-session-123", mockCtx)
-	assert.Equal(t, "0.12.0", version.String()) // Should still be cached version
-
-	// Test different session - should extract new version
-	version = agent.getHubVersion("different-session", mockCtx)
+	// Test that version reflects the current client version (no stale caching)
+	mockCtx.clientVersion = "SSH-2.0-beszel_0.11.0"
+	version = agent.getHubVersion(mockCtx)
 	assert.Equal(t, "0.11.0", version.String())
 
 	// Test with invalid version string (non-beszel client)
 	mockCtx.clientVersion = "SSH-2.0-OpenSSH_8.0"
-	version = agent.getHubVersion("invalid-session", mockCtx)
+	version = agent.getHubVersion(mockCtx)
 	assert.Equal(t, "0.0.0", version.String()) // Should be empty version for non-beszel clients
 
 	// Test with no client version
 	mockCtx.clientVersion = ""
-	version = agent.getHubVersion("no-version-session", mockCtx)
+	version = agent.getHubVersion(mockCtx)
 	assert.True(t, version.EQ(semver.Version{})) // Should be empty version
 }
 
@@ -502,9 +497,6 @@ func TestWriteToSessionEncoding(t *testing.T) {
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			// Reset the global hubVersions map to ensure clean state for each test
-			hubVersions = nil
-
 			agent, err := NewAgent("")
 			require.NoError(t, err)
 
@@ -586,39 +578,28 @@ func createTestCombinedData() *system.CombinedData {
 	}
 }
 
-func TestHubVersionCaching(t *testing.T) {
-	// Reset the global hubVersions map to ensure clean state
-	hubVersions = nil
-
+// TestGetHubVersionConcurrent guards against a regression of the
+// "concurrent map writes" panic previously caused by a shared, unsynchronized
+// hubVersions cache (see https://github.com/henrygd/beszel/issues/2128).
+// getHubVersion no longer shares mutable state between sessions, so calling
+// it concurrently from many goroutines must be safe under `go test -race`.
+func TestGetHubVersionConcurrent(t *testing.T) {
 	agent, err := NewAgent("")
 	require.NoError(t, err)
 
-	ctx1 := &mockSSHContext{
-		sessionID:     "session1",
-		clientVersion: "SSH-2.0-beszel_0.12.0",
+	const goroutines = 50
+	var wg sync.WaitGroup
+	wg.Add(goroutines)
+	for i := 0; i < goroutines; i++ {
+		go func(i int) {
+			defer wg.Done()
+			ctx := &mockSSHContext{
+				sessionID:     fmt.Sprintf("session-%d", i),
+				clientVersion: "SSH-2.0-beszel_0.12.0",
+			}
+			version := agent.getHubVersion(ctx)
+			assert.Equal(t, "0.12.0", version.String())
+		}(i)
 	}
-	ctx2 := &mockSSHContext{
-		sessionID:     "session2",
-		clientVersion: "SSH-2.0-beszel_0.11.0",
-	}
-
-	// First calls should cache the versions
-	v1 := agent.getHubVersion("session1", ctx1)
-	v2 := agent.getHubVersion("session2", ctx2)
-
-	assert.Equal(t, "0.12.0", v1.String())
-	assert.Equal(t, "0.11.0", v2.String())
-
-	// Verify caching by changing context but keeping same session ID
-	ctx1.clientVersion = "SSH-2.0-beszel_0.10.0"
-	v1Cached := agent.getHubVersion("session1", ctx1)
-	assert.Equal(t, "0.12.0", v1Cached.String()) // Should still be cached version
-
-	// New session should get new version
-	ctx3 := &mockSSHContext{
-		sessionID:     "session3",
-		clientVersion: "SSH-2.0-beszel_0.13.0",
-	}
-	v3 := agent.getHubVersion("session3", ctx3)
-	assert.Equal(t, "0.13.0", v3.String())
+	wg.Wait()
 }

agent/smart.go

@@ -25,12 +25,15 @@ import (
 // SmartManager manages data collection for SMART devices
 type SmartManager struct {
 	sync.Mutex
-	SmartDataMap    map[string]*smart.SmartData
-	SmartDevices    []*DeviceInfo
-	refreshMutex    sync.Mutex
-	lastScanTime    time.Time
-	smartctlPath    string
-	excludedDevices map[string]struct{}
+	SmartDataMap       map[string]*smart.SmartData
+	SmartDevices       []*DeviceInfo
+	refreshMutex       sync.Mutex
+	lastScanTime       time.Time
+	smartctlPath       string
+	excludedDevices    map[string]struct{}
+	darwinNvmeOnce     sync.Once
+	darwinNvmeCapacity map[string]uint64      // serial → bytes cache, written once via darwinNvmeOnce
+	darwinNvmeProvider func() ([]byte, error) // overridable for testing
 }
 
 type scanOutput struct {
@@ -52,6 +55,11 @@ type DeviceInfo struct {
 	typeVerified bool
 	// parserType holds the parser type (nvme, sat, scsi) that last succeeded.
 	parserType string
+	// explicitType reports whether Type came from an explicit ":type" hint in
+	// SMART_DEVICES. Such a type is a deliberate user override and must always be
+	// passed to smartctl via -d, even for scsi/ata where a scan-detected type is
+	// otherwise left off (see smartctlArgs and issue #1345).
+	explicitType bool
 }
 
 // deviceKey is a composite key for a device, used to identify a device uniquely.
@@ -62,8 +70,9 @@ type deviceKey struct {
 
 var errNoValidSmartData = fmt.Errorf("no valid SMART data found") // Error for missing data
 
-// Refresh updates SMART data for all known devices
-func (sm *SmartManager) Refresh(forceScan bool) error {
+// Refresh updates SMART data for all known devices and reports whether every
+// discovered device was collected successfully.
+func (sm *SmartManager) Refresh(forceScan bool) (bool, error) {
 	sm.refreshMutex.Lock()
 	defer sm.refreshMutex.Unlock()
 
@@ -84,7 +93,7 @@ func (sm *SmartManager) Refresh(forceScan bool) error {
 		}
 	}
 
-	return sm.resolveRefreshError(scanErr, collectErr)
+	return scanErr == nil && collectErr == nil, sm.resolveRefreshError(scanErr, collectErr)
 }
 
 // devicesSnapshot returns a copy of the current device slice to avoid iterating
@@ -248,8 +257,9 @@ func (sm *SmartManager) parseConfiguredDevices(config string) ([]*DeviceInfo, er
 		}
 
 		devices = append(devices, &DeviceInfo{
-			Name: name,
-			Type: devType,
+			Name:         name,
+			Type:         devType,
+			explicitType: devType != "",
 		})
 	}
 
@@ -555,7 +565,9 @@ func (sm *SmartManager) smartctlArgs(deviceInfo *DeviceInfo, includeStandby bool
 		deviceType = strings.ToLower(deviceInfo.Type)
 		parserType = strings.ToLower(deviceInfo.parserType)
 		// types sometimes misidentified in scan; see github.com/henrygd/beszel/issues/1345
-		if deviceType != "" && deviceType != "scsi" && deviceType != "ata" {
+		// An explicit SMART_DEVICES ":type" hint is a deliberate override, so always
+		// pass it through; otherwise scsi/ata are left off so smartctl can auto-detect.
+		if deviceType != "" && (deviceInfo.explicitType || (deviceType != "scsi" && deviceType != "ata")) {
 			args = append(args, "-d", deviceInfo.Type)
 		}
 	}
@@ -660,6 +672,9 @@ func mergeDeviceLists(existing, scanned, configured []*DeviceInfo) []*DeviceInfo
 		target.Type = prev.Type
 		target.typeVerified = true
 		target.parserType = prev.parserType
+		if prev.explicitType {
+			target.explicitType = true
+		}
 	}
 
 	// applyConfiguredMetadata updates a matched device with any configured
@@ -673,6 +688,9 @@ func mergeDeviceLists(existing, scanned, configured []*DeviceInfo) []*DeviceInfo
 			existingDev.typeVerified = false
 			existingDev.parserType = normalizeParserType(newType)
 		}
+		if configuredDev.explicitType {
+			existingDev.explicitType = true
+		}
 		if configuredDev.InfoName != "" {
 			existingDev.InfoName = configuredDev.InfoName
 		}
@@ -1033,6 +1051,52 @@ func parseScsiGigabytesProcessed(value string) int64 {
 	return parsed
 }
 
+// lookupDarwinNvmeCapacity returns the capacity in bytes for a given NVMe serial number on Darwin.
+// It uses system_profiler SPNVMeDataType to get capacity since Apple SSDs don't report user_capacity
+// via smartctl. Results are cached after the first call via sync.Once.
+func (sm *SmartManager) lookupDarwinNvmeCapacity(serial string) uint64 {
+	sm.darwinNvmeOnce.Do(func() {
+		sm.darwinNvmeCapacity = make(map[string]uint64)
+
+		provider := sm.darwinNvmeProvider
+		if provider == nil {
+			provider = func() ([]byte, error) {
+				ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+				defer cancel()
+				return exec.CommandContext(ctx, "system_profiler", "SPNVMeDataType", "-json").Output()
+			}
+		}
+
+		out, err := provider()
+		if err != nil {
+			slog.Debug("system_profiler NVMe lookup failed", "err", err)
+			return
+		}
+
+		var result struct {
+			SPNVMeDataType []struct {
+				Items []struct {
+					DeviceSerial string `json:"device_serial"`
+					SizeInBytes  uint64 `json:"size_in_bytes"`
+				} `json:"_items"`
+			} `json:"SPNVMeDataType"`
+		}
+		if err := json.Unmarshal(out, &result); err != nil {
+			slog.Debug("system_profiler NVMe parse failed", "err", err)
+			return
+		}
+
+		for _, controller := range result.SPNVMeDataType {
+			for _, item := range controller.Items {
+				if item.DeviceSerial != "" && item.SizeInBytes > 0 {
+					sm.darwinNvmeCapacity[item.DeviceSerial] = item.SizeInBytes
+				}
+			}
+		}
+	})
+	return sm.darwinNvmeCapacity[serial]
+}
+
 // parseSmartForNvme parses the output of smartctl --all -j /dev/nvmeX and updates the SmartDataMap
 // Returns hasValidData and exitStatus
 func (sm *SmartManager) parseSmartForNvme(output []byte) (bool, int) {
@@ -1069,6 +1133,12 @@ func (sm *SmartManager) parseSmartForNvme(output []byte) (bool, int) {
 	smartData.SerialNumber = data.SerialNumber
 	smartData.FirmwareVersion = data.FirmwareVersion
 	smartData.Capacity = data.UserCapacity.Bytes
+	if smartData.Capacity == 0 {
+		smartData.Capacity = data.NVMeTotalCapacity
+	}
+	if smartData.Capacity == 0 && (runtime.GOOS == "darwin" || sm.darwinNvmeProvider != nil) {
+		smartData.Capacity = sm.lookupDarwinNvmeCapacity(data.SerialNumber)
+	}
 	smartData.Temperature = data.NVMeSmartHealthInformationLog.Temperature
 	smartData.SmartStatus = getSmartStatus(smartData.Temperature, data.SmartStatus.Passed)
 	smartData.DiskName = data.Device.Name
@@ -1104,32 +1174,21 @@ func (sm *SmartManager) parseSmartForNvme(output []byte) (bool, int) {
 
 // detectSmartctl checks if smartctl is installed, returns an error if not
 func (sm *SmartManager) detectSmartctl() (string, error) {
-	isWindows := runtime.GOOS == "windows"
-
-	// Load embedded smartctl.exe for Windows amd64 builds.
-	if isWindows && runtime.GOARCH == "amd64" {
-		if path, err := ensureEmbeddedSmartctl(); err == nil {
-			return path, nil
+	if runtime.GOOS == "windows" {
+		// Load embedded smartctl.exe for Windows amd64 builds.
+		if runtime.GOARCH == "amd64" {
+			if path, err := ensureEmbeddedSmartctl(); err == nil {
+				return path, nil
+			}
 		}
-	}
-
-	if path, err := exec.LookPath("smartctl"); err == nil {
-		return path, nil
-	}
-	locations := []string{}
-	if isWindows {
-		locations = append(locations,
-			"C:\\Program Files\\smartmontools\\bin\\smartctl.exe",
-		)
-	} else {
-		locations = append(locations, "/opt/homebrew/bin/smartctl")
-	}
-	for _, location := range locations {
+		// Try to find smartctl in the default installation location
+		const location = "C:\\Program Files\\smartmontools\\bin\\smartctl.exe"
 		if _, err := os.Stat(location); err == nil {
 			return location, nil
 		}
 	}
-	return "", errors.New("smartctl not found")
+
+	return utils.LookPathHomebrew("smartctl")
 }
 
 // isNvmeControllerPath checks if the path matches an NVMe controller pattern

agent/smart_test.go

@@ -392,6 +392,81 @@ func TestSmartctlArgs(t *testing.T) {
 	)
 }
 
+// TestSmartctlArgsExplicitType verifies that an explicit SMART_DEVICES type hint
+// is always passed to smartctl via -d, while a scan-detected scsi/ata type is
+// still left off so smartctl can auto-detect it (see issue #1345).
+func TestSmartctlArgsExplicitType(t *testing.T) {
+	sm := &SmartManager{}
+
+	// Scan-detected scsi: -d is intentionally omitted.
+	scanScsi := &DeviceInfo{Name: "/dev/sda", Type: "scsi"}
+	assert.Equal(t,
+		[]string{"-a", "--json=c", "/dev/sda"},
+		sm.smartctlArgs(scanScsi, false),
+	)
+
+	// Explicit scsi from SMART_DEVICES: -d scsi must be passed.
+	explicitScsi := &DeviceInfo{Name: "/dev/sda", Type: "scsi", explicitType: true}
+	assert.Equal(t,
+		[]string{"-d", "scsi", "-a", "--json=c", "/dev/sda"},
+		sm.smartctlArgs(explicitScsi, false),
+	)
+
+	// Explicit ata from SMART_DEVICES: -d ata must be passed (devstat still added).
+	explicitAta := &DeviceInfo{Name: "/dev/sdb", Type: "ata", explicitType: true}
+	assert.Equal(t,
+		[]string{"-d", "ata", "-a", "--json=c", "-l", "devstat", "/dev/sdb"},
+		sm.smartctlArgs(explicitAta, false),
+	)
+}
+
+// TestSmartDevicesExplicitTypeFlowsToSmartctlArgs is a regression test for
+// issue #2072: an explicit SMART_DEVICES type (e.g. /dev/sda:scsi) must win over
+// a wrong scan-detected type (sat) and be handed to smartctl as -d scsi.
+func TestSmartDevicesExplicitTypeFlowsToSmartctlArgs(t *testing.T) {
+	sm := &SmartManager{}
+
+	configured, err := sm.parseConfiguredDevices("/dev/sda:scsi")
+	require.NoError(t, err)
+	require.Len(t, configured, 1)
+	assert.True(t, configured[0].explicitType)
+
+	// smartctl --scan misreports this USB drive as sat, which fails on it.
+	scanned := []*DeviceInfo{
+		{Name: "/dev/sda", Type: "sat", Protocol: "ATA"},
+	}
+
+	merged := mergeDeviceLists(nil, scanned, configured)
+	require.Len(t, merged, 1)
+
+	device := merged[0]
+	assert.Equal(t, "scsi", device.Type, "configured type should win over scan-detected sat")
+	assert.True(t, device.explicitType, "explicit hint must survive the merge")
+
+	assert.Equal(t,
+		[]string{"-d", "scsi", "-a", "--json=c", "/dev/sda"},
+		sm.smartctlArgs(device, false),
+		"explicit scsi type must be passed to smartctl, not dropped",
+	)
+}
+
+// TestMergeDeviceListsPreservesExplicitTypeAcrossRescan ensures a verified,
+// explicitly-typed device keeps its explicit flag when a later scan re-reports
+// it with a different auto-detected type.
+func TestMergeDeviceListsPreservesExplicitTypeAcrossRescan(t *testing.T) {
+	existing := []*DeviceInfo{
+		{Name: "/dev/sda", Type: "scsi", parserType: "scsi", typeVerified: true, explicitType: true},
+	}
+	scanned := []*DeviceInfo{
+		{Name: "/dev/sda", Type: "sat"},
+	}
+
+	merged := mergeDeviceLists(existing, scanned, nil)
+	require.Len(t, merged, 1)
+	assert.Equal(t, "scsi", merged[0].Type)
+	assert.True(t, merged[0].explicitType, "explicit type flag should survive a rescan")
+}
+
 func TestResolveRefreshError(t *testing.T) {
 	scanErr := errors.New("scan failed")
 	collectErr := errors.New("collect failed")
@@ -1035,7 +1110,7 @@ func TestRefreshExcludedDevices(t *testing.T) {
 				t.Setenv("EXCLUDE_SMART", tt.envValue)
 			} else {
 				// Ensure env var is not set for empty test
-				os.Unsetenv("EXCLUDE_SMART")
+				t.Setenv("EXCLUDE_SMART", "")
 			}
 
 			sm := &SmartManager{}
@@ -1199,3 +1274,81 @@ func TestIsNvmeControllerPath(t *testing.T) {
 		})
 	}
 }
+
+func TestParseSmartForNvmeAppleSSD(t *testing.T) {
+	// Apple SSDs don't report user_capacity via smartctl; capacity should be fetched
+	// from system_profiler via the darwinNvmeProvider fallback.
+	fixturePath := filepath.Join("test-data", "smart", "apple_nvme.json")
+	data, err := os.ReadFile(fixturePath)
+	require.NoError(t, err)
+
+	providerCalls := 0
+	fakeProvider := func() ([]byte, error) {
+		providerCalls++
+		return []byte(`{
+			"SPNVMeDataType": [{
+				"_items": [{
+					"device_serial": "0ba0147940253c15",
+					"size_in_bytes": 251000193024
+				}]
+			}]
+		}`), nil
+	}
+
+	sm := &SmartManager{
+		SmartDataMap:       make(map[string]*smart.SmartData),
+		darwinNvmeProvider: fakeProvider,
+	}
+
+	hasData, _ := sm.parseSmartForNvme(data)
+	require.True(t, hasData)
+
+	deviceData, ok := sm.SmartDataMap["0ba0147940253c15"]
+	require.True(t, ok)
+	assert.Equal(t, "APPLE SSD AP0256Q", deviceData.ModelName)
+	assert.Equal(t, uint64(251000193024), deviceData.Capacity)
+	assert.Equal(t, uint8(42), deviceData.Temperature)
+	assert.Equal(t, "PASSED", deviceData.SmartStatus)
+	assert.Equal(t, 1, providerCalls, "system_profiler should be called once")
+
+	// Second parse: provider should NOT be called again (cache hit)
+	_, _ = sm.parseSmartForNvme(data)
+	assert.Equal(t, 1, providerCalls, "system_profiler should not be called again after caching")
+}
+
+func TestLookupDarwinNvmeCapacityMultipleDisks(t *testing.T) {
+	fakeProvider := func() ([]byte, error) {
+		return []byte(`{
+			"SPNVMeDataType": [
+				{
+					"_items": [
+						{"device_serial": "serial-disk0", "size_in_bytes": 251000193024},
+						{"device_serial": "serial-disk1", "size_in_bytes": 1000204886016}
+					]
+				},
+				{
+					"_items": [
+						{"device_serial": "serial-disk2", "size_in_bytes": 512110190592}
+					]
+				}
+			]
+		}`), nil
+	}
+
+	sm := &SmartManager{darwinNvmeProvider: fakeProvider}
+	assert.Equal(t, uint64(251000193024), sm.lookupDarwinNvmeCapacity("serial-disk0"))
+	assert.Equal(t, uint64(1000204886016), sm.lookupDarwinNvmeCapacity("serial-disk1"))
+	assert.Equal(t, uint64(512110190592), sm.lookupDarwinNvmeCapacity("serial-disk2"))
+	assert.Equal(t, uint64(0), sm.lookupDarwinNvmeCapacity("unknown-serial"))
+}
+
+func TestLookupDarwinNvmeCapacityProviderError(t *testing.T) {
+	fakeProvider := func() ([]byte, error) {
+		return nil, errors.New("system_profiler not found")
+	}
+
+	sm := &SmartManager{darwinNvmeProvider: fakeProvider}
+	assert.Equal(t, uint64(0), sm.lookupDarwinNvmeCapacity("any-serial"))
+	// Cache should be initialized even on error so we don't retry (Once already fired)
+	assert.NotNil(t, sm.darwinNvmeCapacity)
+}

agent/system.go

@@ -8,7 +8,6 @@ import (
 	"os"
 	"runtime"
 	"strings"
-	"time"
 
 	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/agent/battery"
@@ -23,13 +22,6 @@ import (
 	"github.com/shirou/gopsutil/v4/mem"
 )
 
-// prevDisk stores previous per-device disk counters for a given cache interval
-type prevDisk struct {
-	readBytes  uint64
-	writeBytes uint64
-	at         time.Time
-}
-
 // Sets initial / non-changing values about the host system
 func (a *Agent) refreshSystemDetails() {
 	a.systemInfo.AgentVersion = beszel.Version
@@ -115,14 +107,39 @@ func (a *Agent) refreshSystemDetails() {
 	}
 }
 
+// attachSystemDetails returns details only for fresh default-interval responses.
+func (a *Agent) attachSystemDetails(data *system.CombinedData, cacheTimeMs uint16, includeRequested bool) *system.CombinedData {
+	if cacheTimeMs != defaultDataCacheTimeMs || (!includeRequested && !a.detailsDirty) {
+		return data
+	}
+
+	// copy data to avoid adding details to the original cached struct
+	response := *data
+	response.Details = &a.systemDetails
+	a.detailsDirty = false
+	return &response
+}
+
+// updateSystemDetails applies a mutation to the static details payload and marks
+// it for inclusion on the next fresh default-interval response.
+func (a *Agent) updateSystemDetails(updateFunc func(details *system.Details)) {
+	updateFunc(&a.systemDetails)
+	a.detailsDirty = true
+}
+
 // Returns current info, stats about the host system
 func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	var systemStats system.Stats
 
 	// battery
-	if batteryPercent, batteryState, err := battery.GetBatteryStats(); err == nil {
-		systemStats.Battery[0] = batteryPercent
-		systemStats.Battery[1] = batteryState
+	if batteries, err := battery.GetBatteryStats(); err == nil {
+		systemStats.Batteries = make(map[string]uint8, len(batteries))
+		for _, device := range batteries {
+			systemStats.Batteries[device.Name] = device.Percent
+		}
+		if primary, ok := battery.Primary(batteries); ok {
+			systemStats.Battery = [2]uint8{primary.Percent, primary.State}
+		}
 	}
 
 	// cpu metrics
@@ -157,21 +174,11 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 
 	// memory
 	if v, err := mem.VirtualMemory(); err == nil {
+		used, cacheBuff, swapUsed := calculateHostMemoryUsage(v, a.memCalc == "htop")
 		// swap
 		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
-		if cacheBuff <= 0 {
-			cacheBuff = max(v.Total-v.Free-v.Used, 0)
-		}
-		// htop memory calculation overrides (likely outdated as of mid 2025)
-		if a.memCalc == "htop" {
-			// cacheBuff = v.Cached + v.Buffers - v.Shared
-			v.Used = v.Total - (v.Free + cacheBuff)
-			v.UsedPercent = float64(v.Used) / float64(v.Total) * 100.0
-		}
+		systemStats.SwapUsed = utils.BytesToGigabytes(swapUsed)
+		v.Used = used
 		// if a.memCalc == "legacy" {
 		// 	v.Used = v.Total - v.Free - v.Buffers - v.Cached
 		// 	cacheBuff = v.Total - v.Free - v.Used
@@ -181,10 +188,14 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 		if a.zfs {
 			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 = utils.BytesToGigabytes(arcSize)
 			}
 		}
+		if v.Total > 0 {
+			v.UsedPercent = float64(v.Used) / float64(v.Total) * 100.0
+		} else {
+			v.UsedPercent = 0
+		}
 		systemStats.Mem = utils.BytesToGigabytes(v.Total)
 		systemStats.MemBuffCache = utils.BytesToGigabytes(cacheBuff)
 		systemStats.MemUsed = utils.BytesToGigabytes(v.Used)
@@ -204,6 +215,9 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	// TODO: maybe refactor to methods on systemStats
 	a.updateTemperatures(&systemStats)
 
+	// fan speeds (Linux-only; sysfs hwmon)
+	a.updateFans(&systemStats)
+
 	// GPU data
 	if a.gpuManager != nil {
 		// reset high gpu percent
@@ -251,6 +265,38 @@ func (a *Agent) getSystemStats(cacheTimeMs uint16) system.Stats {
 	return systemStats
 }
 
+// calculateHostMemoryUsage derives counters defensively because /proc/meminfo may
+// change while gopsutil reads it. Invalid unsigned subtractions saturate at zero.
+func calculateHostMemoryUsage(v *mem.VirtualMemoryStat, htop bool) (used, cacheBuff, swapUsed uint64) {
+	used = v.Used
+	if used > v.Total {
+		used = saturatingSub(v.Total, v.Available)
+	}
+
+	// gopsutil automatically adds SReclaimable to Cached.
+	cacheBuff = min(v.Cached, v.Total)
+	cacheBuff += min(v.Buffers, v.Total-cacheBuff)
+	cacheBuff = saturatingSub(cacheBuff, min(v.Shared, v.Total))
+	if v.Cached == 0 && v.Buffers == 0 {
+		cacheBuff = saturatingSub(v.Total, v.Free, used)
+	}
+	if htop {
+		used = saturatingSub(v.Total, v.Free, cacheBuff)
+	}
+	return used, cacheBuff, saturatingSub(v.SwapTotal, v.SwapFree, v.SwapCached)
+}
+
+// saturatingSub subtracts each value, returning zero on underflow.
+func saturatingSub(value uint64, subtrahends ...uint64) uint64 {
+	for _, subtrahend := range subtrahends {
+		if subtrahend > value {
+			return 0
+		}
+		value -= subtrahend
+	}
+	return value
+}
+
 // 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/system_test.go

@@ -0,0 +1,115 @@
+package agent
+
+import (
+	"testing"
+
+	"github.com/henrygd/beszel/internal/common"
+	"github.com/henrygd/beszel/internal/entities/system"
+	"github.com/shirou/gopsutil/v4/mem"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestGatherStatsDoesNotAttachDetailsToCachedRequests(t *testing.T) {
+	agent := &Agent{
+		cache:         NewSystemDataCache(),
+		systemDetails: system.Details{Hostname: "updated-host", Podman: true},
+		detailsDirty:  true,
+	}
+	cached := &system.CombinedData{
+		Info: system.Info{Hostname: "cached-host"},
+	}
+	agent.cache.Set(cached, defaultDataCacheTimeMs)
+
+	response := agent.gatherStats(common.DataRequestOptions{CacheTimeMs: defaultDataCacheTimeMs})
+
+	assert.Same(t, cached, response)
+	assert.Nil(t, response.Details)
+	assert.True(t, agent.detailsDirty)
+	assert.Equal(t, "cached-host", response.Info.Hostname)
+	assert.Nil(t, cached.Details)
+
+	secondResponse := agent.gatherStats(common.DataRequestOptions{CacheTimeMs: defaultDataCacheTimeMs})
+	assert.Same(t, cached, secondResponse)
+	assert.Nil(t, secondResponse.Details)
+}
+
+func TestCalculateHostMemoryUsage(t *testing.T) {
+	tests := []struct {
+		name                      string
+		memory                    mem.VirtualMemoryStat
+		htop                      bool
+		used, cacheBuff, swapUsed uint64
+	}{
+		{
+			name:      "normal",
+			memory:    mem.VirtualMemoryStat{Total: 100, Available: 40, Used: 60, Free: 20, Cached: 25, Buffers: 10, Shared: 5, SwapTotal: 20, SwapFree: 8, SwapCached: 2},
+			used:      60,
+			cacheBuff: 30,
+			swapUsed:  10,
+		},
+		{
+			name:      "inconsistent counters saturate",
+			memory:    mem.VirtualMemoryStat{Total: 100, Available: 110, Used: ^uint64(0) - 9, Free: 90, Cached: 5, Buffers: 10, Shared: 20, SwapTotal: 10, SwapFree: 9, SwapCached: 2},
+			used:      0,
+			cacheBuff: 0,
+			swapUsed:  0,
+		},
+		{
+			name:      "htop subtraction saturates",
+			memory:    mem.VirtualMemoryStat{Total: 100, Available: 20, Used: 80, Free: 90, Cached: 20, Buffers: 5, SwapTotal: 30, SwapFree: 10, SwapCached: 5},
+			htop:      true,
+			used:      0,
+			cacheBuff: 25,
+			swapUsed:  15,
+		},
+		{
+			name:      "zero cache from shared cancellation does not fall back",
+			memory:    mem.VirtualMemoryStat{Total: 100, Used: 60, Free: 10, Cached: 20, Buffers: 10, Shared: 30},
+			used:      60,
+			cacheBuff: 0,
+		},
+		{
+			name:      "absent cache counters use fallback",
+			memory:    mem.VirtualMemoryStat{Total: 100, Used: 60, Free: 10},
+			used:      60,
+			cacheBuff: 30,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			used, cacheBuff, swapUsed := calculateHostMemoryUsage(&tt.memory, tt.htop)
+			assert.Equal(t, tt.used, used)
+			assert.Equal(t, tt.cacheBuff, cacheBuff)
+			assert.Equal(t, tt.swapUsed, swapUsed)
+		})
+	}
+}
+
+func TestUpdateSystemDetailsMarksDetailsDirty(t *testing.T) {
+	agent := &Agent{}
+
+	agent.updateSystemDetails(func(details *system.Details) {
+		details.Hostname = "updated-host"
+		details.Podman = true
+	})
+
+	assert.True(t, agent.detailsDirty)
+	assert.Equal(t, "updated-host", agent.systemDetails.Hostname)
+	assert.True(t, agent.systemDetails.Podman)
+
+	original := &system.CombinedData{}
+	realTimeResponse := agent.attachSystemDetails(original, 1000, true)
+	assert.Same(t, original, realTimeResponse)
+	assert.Nil(t, realTimeResponse.Details)
+	assert.True(t, agent.detailsDirty)
+
+	response := agent.attachSystemDetails(original, defaultDataCacheTimeMs, false)
+	require.NotNil(t, response.Details)
+	assert.NotSame(t, original, response)
+	assert.Equal(t, "updated-host", response.Details.Hostname)
+	assert.True(t, response.Details.Podman)
+	assert.False(t, agent.detailsDirty)
+	assert.Nil(t, original.Details)
+}

agent/systemd.go

@@ -301,7 +301,7 @@ func getServicePatterns() []string {
 			if pattern == "" {
 				continue
 			}
-			if !strings.HasSuffix(pattern, ".service") {
+			if !strings.HasSuffix(pattern, "timer") && !strings.HasSuffix(pattern, ".service") {
 				pattern += ".service"
 			}
 			patterns = append(patterns, pattern)

agent/systemd_test.go

@@ -156,20 +156,23 @@ func TestGetServicePatterns(t *testing.T) {
 			expected:       []string{"*nginx*.service", "*apache*.service"},
 			cleanupEnvVars: true,
 		},
+		{
+			name:           "opt into timer monitoring",
+			prefixedEnv:    "nginx.service,docker,apache.timer",
+			unprefixedEnv:  "",
+			expected:       []string{"nginx.service", "docker.service", "apache.timer"},
+			cleanupEnvVars: true,
+		},
 	}
 
 	for _, tt := range tests {
 		t.Run(tt.name, func(t *testing.T) {
-			// Clean up any existing env vars
-			os.Unsetenv("BESZEL_AGENT_SERVICE_PATTERNS")
-			os.Unsetenv("SERVICE_PATTERNS")
-
 			// Set up environment variables
 			if tt.prefixedEnv != "" {
-				os.Setenv("BESZEL_AGENT_SERVICE_PATTERNS", tt.prefixedEnv)
+				t.Setenv("BESZEL_AGENT_SERVICE_PATTERNS", tt.prefixedEnv)
 			}
 			if tt.unprefixedEnv != "" {
-				os.Setenv("SERVICE_PATTERNS", tt.unprefixedEnv)
+				t.Setenv("SERVICE_PATTERNS", tt.unprefixedEnv)
 			}
 
 			// Run the function
@@ -177,12 +180,6 @@ func TestGetServicePatterns(t *testing.T) {
 
 			// Verify results
 			assert.Equal(t, tt.expected, result, "Patterns should match expected values")
-
-			// Cleanup
-			if tt.cleanupEnvVars {
-				os.Unsetenv("BESZEL_AGENT_SERVICE_PATTERNS")
-				os.Unsetenv("SERVICE_PATTERNS")
-			}
 		})
 	}
 }

agent/test-data/smart/apple_nvme.json

@@ -0,0 +1,51 @@
+{
+  "json_format_version": [1, 0],
+  "smartctl": {
+    "version": [7, 4],
+    "argv": ["smartctl", "-aix", "-j", "IOService:/AppleARMPE/arm-io@10F00000/AppleT810xIO/ans@77400000/AppleASCWrapV4/iop-ans-nub/RTBuddy(ANS2)/RTBuddyService/AppleANS3NVMeController/NS_01@1"],
+    "exit_status": 4
+  },
+  "device": {
+    "name": "IOService:/AppleARMPE/arm-io@10F00000/AppleT810xIO/ans@77400000/AppleASCWrapV4/iop-ans-nub/RTBuddy(ANS2)/RTBuddyService/AppleANS3NVMeController/NS_01@1",
+    "info_name": "IOService:/AppleARMPE/arm-io@10F00000/AppleT810xIO/ans@77400000/AppleASCWrapV4/iop-ans-nub/RTBuddy(ANS2)/RTBuddyService/AppleANS3NVMeController/NS_01@1",
+    "type": "nvme",
+    "protocol": "NVMe"
+  },
+  "model_name": "APPLE SSD AP0256Q",
+  "serial_number": "0ba0147940253c15",
+  "firmware_version": "555",
+  "smart_support": {
+    "available": true,
+    "enabled": true
+  },
+  "smart_status": {
+    "passed": true,
+    "nvme": {
+      "value": 0
+    }
+  },
+  "nvme_smart_health_information_log": {
+    "critical_warning": 0,
+    "temperature": 42,
+    "available_spare": 100,
+    "available_spare_threshold": 99,
+    "percentage_used": 1,
+    "data_units_read": 270189386,
+    "data_units_written": 166753862,
+    "host_reads": 7543766995,
+    "host_writes": 3761621926,
+    "controller_busy_time": 0,
+    "power_cycles": 366,
+    "power_on_hours": 2850,
+    "unsafe_shutdowns": 195,
+    "media_errors": 0,
+    "num_err_log_entries": 0
+  },
+  "temperature": {
+    "current": 42
+  },
+  "power_cycle_count": 366,
+  "power_on_time": {
+    "hours": 2850
+  }
+}

agent/utils/utils.go

@@ -1,9 +1,14 @@
+// Package utils provides utility functions for the agent.
 package utils
 
 import (
+	"fmt"
 	"io"
 	"math"
 	"os"
+	"os/exec"
+	"path/filepath"
+	"runtime"
 	"strconv"
 	"strings"
 )
@@ -65,6 +70,9 @@ func ReadStringFileLimited(path string, maxSize int) (string, error) {
 	if err != nil && err != io.EOF {
 		return "", err
 	}
+	if n < 0 {
+		return "", fmt.Errorf("%s returned negative bytes: %d", path, n)
+	}
 	return strings.TrimSpace(string(buf[:n])), nil
 }
 
@@ -86,3 +94,24 @@ func ReadUintFile(path string) (uint64, bool) {
 	}
 	return parsed, true
 }
+
+// LookPathHomebrew is like exec.LookPath but also checks Homebrew paths.
+func LookPathHomebrew(file string) (string, error) {
+	foundPath, lookPathErr := exec.LookPath(file)
+	if lookPathErr == nil {
+		return foundPath, nil
+	}
+	var homebrewPath string
+	switch runtime.GOOS {
+	case "darwin":
+		homebrewPath = filepath.Join("/opt", "homebrew", "bin", file)
+	case "linux":
+		homebrewPath = filepath.Join("/home", "linuxbrew", ".linuxbrew", "bin", file)
+	}
+	if homebrewPath != "" {
+		if _, err := os.Stat(homebrewPath); err == nil {
+			return homebrewPath, nil
+		}
+	}
+	return "", lookPathErr
+}

agent/utils/utils_test.go

@@ -134,10 +134,8 @@ func TestGetEnv(t *testing.T) {
 	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)
+		t.Setenv(prefixedKey, "prefixed_val")
+		t.Setenv(key, "unprefixed_val")
 
 		val, exists := GetEnv(key)
 		assert.True(t, exists)
@@ -145,9 +143,7 @@ func TestGetEnv(t *testing.T) {
 	})
 
 	t.Run("only unprefixed variable exists", func(t *testing.T) {
-		os.Unsetenv(prefixedKey)
-		os.Setenv(key, "unprefixed_val")
-		defer os.Unsetenv(key)
+		t.Setenv(key, "unprefixed_val")
 
 		val, exists := GetEnv(key)
 		assert.True(t, exists)
@@ -155,9 +151,6 @@ func TestGetEnv(t *testing.T) {
 	})
 
 	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)

beszel.go

@@ -6,7 +6,7 @@ import "github.com/blang/semver"
 
 const (
 	// Version is the current version of the application.
-	Version = "0.18.4"
+	Version = "0.18.7"
 	// AppName is the name of the application.
 	AppName = "beszel"
 )

go.mod

@@ -1,28 +1,29 @@
 module github.com/henrygd/beszel
 
-go 1.26.1
+go 1.26.6
 
 require (
 	github.com/blang/semver v3.5.1+incompatible
 	github.com/coreos/go-systemd/v22 v22.7.0
-	github.com/distatus/battery v0.11.0
-	github.com/ebitengine/purego v0.9.1
-	github.com/fxamacker/cbor/v2 v2.9.0
+	github.com/ebitengine/purego v0.10.2
+	github.com/fxamacker/cbor/v2 v2.9.2
 	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 v0.13.2
+	github.com/lxzan/gws v1.10.1
+	github.com/nicholas-fedor/shoutrrr v0.17.0
 	github.com/pocketbase/dbx v1.12.0
-	github.com/pocketbase/pocketbase v0.36.4
-	github.com/shirou/gopsutil/v4 v4.26.1
+	github.com/pocketbase/pocketbase v0.39.10
+	github.com/shirou/gopsutil/v4 v4.26.7
 	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 v0.48.0
-	golang.org/x/exp v0.0.0-20260218203240-3dfff04db8fa
-	golang.org/x/sys v0.41.0
+	golang.org/x/crypto v0.55.0
+	golang.org/x/exp v0.0.0-20260811152304-ee035b5b010f
+	golang.org/x/net v0.57.0
+	golang.org/x/sys v0.47.0
 	gopkg.in/yaml.v3 v3.0.1
+	howett.net/plist v1.0.1
 )
 
 require (
@@ -30,39 +31,39 @@ require (
 	github.com/asaskevich/govalidator v0.0.0-20230301143203-a9d515a09cc2 // indirect
 	github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc // indirect
 	github.com/disintegration/imaging v1.6.2 // indirect
-	github.com/dolthub/maphash v0.1.0 // indirect
 	github.com/domodwyer/mailyak/v3 v3.6.2 // indirect
 	github.com/dustin/go-humanize v1.0.1 // indirect
-	github.com/fatih/color v1.18.0 // indirect
-	github.com/gabriel-vasile/mimetype v1.4.13 // indirect
-	github.com/ganigeorgiev/fexpr v0.5.0 // indirect
+	github.com/eclipse/paho.golang v0.23.0 // indirect
+	github.com/fatih/color v1.19.0 // indirect
+	github.com/fsnotify/fsnotify v1.10.1 // indirect
+	github.com/gabriel-vasile/mimetype v1.4.15 // indirect
+	github.com/ganigeorgiev/fexpr v0.6.0 // indirect
 	github.com/go-ole/go-ole v1.3.0 // indirect
-	github.com/go-ozzo/ozzo-validation/v4 v4.3.0 // indirect
 	github.com/go-sql-driver/mysql v1.9.1 // indirect
 	github.com/godbus/dbus/v5 v5.2.2 // indirect
 	github.com/golang-jwt/jwt/v5 v5.3.1 // indirect
+	github.com/gorilla/websocket v1.5.3 // indirect
 	github.com/inconshreveable/mousetrap v1.1.0 // 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/klauspost/compress v1.19.2 // indirect
+	github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5 // indirect
+	github.com/mattn/go-colorable v0.1.15 // indirect
+	github.com/mattn/go-isatty v0.0.24 // indirect
 	github.com/ncruces/go-strftime v1.0.0 // indirect
 	github.com/pmezard/go-difflib v1.0.1-0.20181226105442-5d4384ee4fb2 // indirect
-	github.com/power-devops/perfstat v0.0.0-20240221224432-82ca36839d55 // indirect
+	github.com/pocketbase/ozzo-validation/v4 v4.3.0 // indirect
+	github.com/power-devops/perfstat v0.0.0-20260805114148-88456608a4f6 // indirect
 	github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec // indirect
-	github.com/tklauser/go-sysconf v0.3.16 // indirect
-	github.com/tklauser/numcpus v0.11.0 // indirect
+	github.com/tklauser/go-sysconf v0.4.0 // indirect
+	github.com/tklauser/numcpus v0.12.0 // indirect
 	github.com/x448/float16 v0.8.4 // indirect
 	github.com/yusufpapurcu/wmi v1.2.4 // 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 v0.40.0 // indirect
-	golang.org/x/text v0.34.0 // indirect
-	howett.net/plist v1.0.1 // indirect
-	modernc.org/libc v1.67.6 // indirect
+	golang.org/x/image v0.45.0 // indirect
+	golang.org/x/oauth2 v0.36.0 // indirect
+	golang.org/x/sync v0.22.0 // indirect
+	golang.org/x/term v0.45.0 // indirect
+	golang.org/x/text v0.41.0 // indirect
+	modernc.org/libc v1.74.1 // indirect
 	modernc.org/mathutil v1.7.1 // indirect
-	modernc.org/memory v1.11.0 // indirect
-	modernc.org/sqlite v1.45.0 // indirect
+	modernc.org/memory v1.12.0 // indirect
+	modernc.org/sqlite v1.55.0 // indirect
 )

go.sum

@@ -1,7 +1,7 @@
 filippo.io/edwards25519 v1.1.0 h1:FNf4tywRC1HmFuKW5xopWpigGjJKiJSV0Cqo0cJWDaA=
 filippo.io/edwards25519 v1.1.0/go.mod h1:BxyFTGdWcka3PhytdK4V28tE5sGfRvvvRV7EaN4VDT4=
-github.com/Masterminds/semver/v3 v3.4.0 h1:Zog+i5UMtVoCU8oKka5P7i9q9HgrJeGzI9SA1Xbatp0=
-github.com/Masterminds/semver/v3 v3.4.0/go.mod h1:4V+yj/TJE1HU9XfppCwVMZq3I84lprf4nC11bSS5beM=
+github.com/Masterminds/semver/v3 v3.5.0 h1:kQceYJfbupGfZOKZQg0kou0DgAKhzDg2NZPAwZ/2OOE=
+github.com/Masterminds/semver/v3 v3.5.0/go.mod h1:4V+yj/TJE1HU9XfppCwVMZq3I84lprf4nC11bSS5beM=
 github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be h1:9AeTilPcZAjCFIImctFaOjnTIavg87rW78vTPkQqLI8=
 github.com/anmitsu/go-shlex v0.0.0-20200514113438-38f4b401e2be/go.mod h1:ySMOLuWl6zY27l47sB3qLNK6tF2fkHG55UZxx8oIVo4=
 github.com/asaskevich/govalidator v0.0.0-20200108200545-475eaeb16496/go.mod h1:oGkLhpf+kjZl6xBf758TQhh5XrAeiJv/7FRz/2spLIg=
@@ -17,35 +17,33 @@ github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc h1:U9qPSI2PIWSS1
 github.com/davecgh/go-spew v1.1.2-0.20180830191138-d8f796af33cc/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/disintegration/imaging v1.6.2 h1:w1LecBlG2Lnp8B3jk5zSuNqd7b4DXhcjwek1ei82L+c=
 github.com/disintegration/imaging v1.6.2/go.mod h1:44/5580QXChDfwIclfc/PCwrr44amcmDAg8hxG0Ewe4=
-github.com/distatus/battery v0.11.0 h1:KJk89gz90Iq/wJtbjjM9yUzBXV+ASV/EG2WOOL7N8lc=
-github.com/distatus/battery v0.11.0/go.mod h1:KmVkE8A8hpIX4T78QRdMktYpEp35QfOL8A8dwZBxq2k=
-github.com/dolthub/maphash v0.1.0 h1:bsQ7JsF4FkkWyrP3oCnFJgrCUAFbFf3kOl4L/QxPDyQ=
-github.com/dolthub/maphash v0.1.0/go.mod h1:gkg4Ch4CdCDu5h6PMriVLawB7koZ+5ijb9puGMV50a4=
 github.com/domodwyer/mailyak/v3 v3.6.2 h1:x3tGMsyFhTCaxp6ycgR0FE/bu5QiNp+hetUuCOBXMn8=
 github.com/domodwyer/mailyak/v3 v3.6.2/go.mod h1:lOm/u9CyCVWHeaAmHIdF4RiKVxKUT/H5XX10lIKAL6c=
 github.com/dustin/go-humanize v1.0.1 h1:GzkhY7T5VNhEkwH0PVJgjz+fX1rhBrR7pRT3mDkpeCY=
 github.com/dustin/go-humanize v1.0.1/go.mod h1:Mu1zIs6XwVuF/gI1OepvI0qD18qycQx+mFykh5fBlto=
-github.com/ebitengine/purego v0.9.1 h1:a/k2f2HQU3Pi399RPW1MOaZyhKJL9w/xFpKAg4q1s0A=
-github.com/ebitengine/purego v0.9.1/go.mod h1:iIjxzd6CiRiOG0UyXP+V1+jWqUXVjPKLAI0mRfJZTmQ=
-github.com/fatih/color v1.18.0 h1:S8gINlzdQ840/4pfAwic/ZE0djQEH3wM94VfqLTZcOM=
-github.com/fatih/color v1.18.0/go.mod h1:4FelSpRwEGDpQ12mAdzqdOukCy4u8WUtOY6lkT/6HfU=
+github.com/ebitengine/purego v0.10.2 h1:W809HbnvzAxgdm+aOvlSekrM16wGCdT/e76+9tS7gzE=
+github.com/ebitengine/purego v0.10.2/go.mod h1:iIjxzd6CiRiOG0UyXP+V1+jWqUXVjPKLAI0mRfJZTmQ=
+github.com/eclipse/paho.golang v0.23.0 h1:KHgl2wz6EJo7cMBmkuhpt7C576vP+kpPv7jjvSyR6Mk=
+github.com/eclipse/paho.golang v0.23.0/go.mod h1:nQRhTkoZv8EAiNs5UU0/WdQIx2NrnWUpL9nsGJTQN04=
+github.com/fatih/color v1.19.0 h1:Zp3PiM21/9Ld6FzSKyL5c/BULoe/ONr9KlbYVOfG8+w=
+github.com/fatih/color v1.19.0/go.mod h1:zNk67I0ZUT1bEGsSGyCZYZNrHuTkJJB+r6Q9VuMi0LE=
 github.com/frankban/quicktest v1.14.6 h1:7Xjx+VpznH+oBnejlPUj8oUpdxnVs4f8XU8WnHkI4W8=
 github.com/frankban/quicktest v1.14.6/go.mod h1:4ptaffx2x8+WTWXmUCuVU6aPUX1/Mz7zb5vbUoiM6w0=
-github.com/fxamacker/cbor/v2 v2.9.0 h1:NpKPmjDBgUfBms6tr6JZkTHtfFGcMKsw3eGcmD/sapM=
-github.com/fxamacker/cbor/v2 v2.9.0/go.mod h1:vM4b+DJCtHn+zz7h3FFp/hDAI9WNWCsZj23V5ytsSxQ=
-github.com/gabriel-vasile/mimetype v1.4.13 h1:46nXokslUBsAJE/wMsp5gtO500a4F3Nkz9Ufpk2AcUM=
-github.com/gabriel-vasile/mimetype v1.4.13/go.mod h1:d+9Oxyo1wTzWdyVUPMmXFvp4F9tea18J8ufA774AB3s=
-github.com/ganigeorgiev/fexpr v0.5.0 h1:XA9JxtTE/Xm+g/JFI6RfZEHSiQlk+1glLvRK1Lpv/Tk=
-github.com/ganigeorgiev/fexpr v0.5.0/go.mod h1:RyGiGqmeXhEQ6+mlGdnUleLHgtzzu/VGO2WtJkF5drE=
+github.com/fsnotify/fsnotify v1.10.1 h1:b0/UzAf9yR5rhf3RPm9gf3ehBPpf0oZKIjtpKrx59Ho=
+github.com/fsnotify/fsnotify v1.10.1/go.mod h1:TLheqan6HD6GBK6PrDWyDPBaEV8LspOxvPSjC+bVfgo=
+github.com/fxamacker/cbor/v2 v2.9.2 h1:X4Ksno9+x3cz0TZv69ec1hxP/+tymuR8PXQJyDwfh78=
+github.com/fxamacker/cbor/v2 v2.9.2/go.mod h1:vM4b+DJCtHn+zz7h3FFp/hDAI9WNWCsZj23V5ytsSxQ=
+github.com/gabriel-vasile/mimetype v1.4.15 h1:05iP/CYtZ/w455R/KZM6rZ5ieAdh99UPtd+d3YzLmaI=
+github.com/gabriel-vasile/mimetype v1.4.15/go.mod h1:azpTcoLcDZRNgFou5j+APrqQx9HqVPWa6ijYQIIVswQ=
+github.com/ganigeorgiev/fexpr v0.6.0 h1:Fza3O/QMBKEudUvxV862qe6GjxM60GJjjKytdp+VQus=
+github.com/ganigeorgiev/fexpr v0.6.0/go.mod h1:RyGiGqmeXhEQ6+mlGdnUleLHgtzzu/VGO2WtJkF5drE=
 github.com/gliderlabs/ssh v0.3.8 h1:a4YXD1V7xMF9g5nTkdfnja3Sxy1PVDCj1Zg4Wb8vY6c=
 github.com/gliderlabs/ssh v0.3.8/go.mod h1:xYoytBv1sV0aL3CavoDuJIQNURXkkfPA/wxQ1pL1fAU=
-github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
-github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/logr v1.4.4 h1:tG4xh9yMsRCAiodLVTxyrkzSZ9+o0L1Kg/+cPVcbP/8=
+github.com/go-logr/logr v1.4.4/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
 github.com/go-ole/go-ole v1.2.6/go.mod h1:pprOEPIfldk/42T2oK7lQ4v4JSDwmV0As9GaiUsvbm0=
 github.com/go-ole/go-ole v1.3.0 h1:Dt6ye7+vXGIKZ7Xtk4s6/xVdGDQynvom7xCFEdWr6uE=
 github.com/go-ole/go-ole v1.3.0/go.mod h1:5LS6F96DhAwUc7C+1HLexzMXY1xGRSryjyPPKW6zv78=
-github.com/go-ozzo/ozzo-validation/v4 v4.3.0 h1:byhDUpfEwjsVQb1vBunvIjh2BHQ9ead57VkAEY4V+Es=
-github.com/go-ozzo/ozzo-validation/v4 v4.3.0/go.mod h1:2NKgrcHl3z6cJs+3Oo940FPRiTzuqKbvfrL2RxCj6Ew=
 github.com/go-sql-driver/mysql v1.4.1/go.mod h1:zAC/RDZ24gD3HViQzih4MyKcchzm+sOG5ZlKdlhCg5w=
 github.com/go-sql-driver/mysql v1.9.1 h1:FrjNGn/BsJQjVRuSa8CBrM5BWA9BWoXXat3KrtSb/iI=
 github.com/go-sql-driver/mysql v1.9.1/go.mod h1:qn46aNg1333BRMNU69Lq93t8du/dwxI64Gl8i5p1WMU=
@@ -58,55 +56,59 @@ github.com/golang-jwt/jwt/v5 v5.3.1/go.mod h1:fxCRLWMO43lRc8nhHWY6LGqRcf+1gQWArs
 github.com/golang/protobuf v1.3.1/go.mod h1:6lQm79b+lXiMfvg/cZm0SGofjICqVBUtrP5yJMmIC1U=
 github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
 github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
-github.com/google/pprof v0.0.0-20260115054156-294ebfa9ad83 h1:z2ogiKUYzX5Is6zr/vP9vJGqPwcdqsWjOt+V8J7+bTc=
-github.com/google/pprof v0.0.0-20260115054156-294ebfa9ad83/go.mod h1:MxpfABSjhmINe3F1It9d+8exIHFvUqtLIRCdOGNXqiI=
+github.com/google/pprof v0.0.0-20260802141513-ef3492d7dac3 h1:LMLX+LgTNWpfvCBdFebv6EsYotImrt/Ppc5cXIriCSo=
+github.com/google/pprof v0.0.0-20260802141513-ef3492d7dac3/go.mod h1:jl5iWTm0/hd5PjEYEOuwAJ57L/CibdZfrqZ5XA5GrCk=
 github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
 github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
+github.com/gorilla/websocket v1.5.3 h1:saDtZ6Pbx/0u+bgYQ3q96pZgCzfhKXGPqt7kZ72aNNg=
+github.com/gorilla/websocket v1.5.3/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
 github.com/hashicorp/golang-lru/v2 v2.0.7 h1:a+bsQ5rvGLjzHuww6tVxozPZFVghXaHOwFs4luLUK2k=
 github.com/hashicorp/golang-lru/v2 v2.0.7/go.mod h1:QeFd9opnmA6QUJc5vARoKUSoFhyfM2/ZepoAG6RGpeM=
 github.com/inconshreveable/mousetrap v1.1.0 h1:wN+x4NVGpMsO7ErUn/mUI3vEoE6Jt13X2s0bqwp9tc8=
 github.com/inconshreveable/mousetrap v1.1.0/go.mod h1:vpF70FUmC8bwa3OWnCshd2FqLfsEA9PFc4w1p2J65bw=
-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/jarcoal/httpmock v1.4.2 h1:dKwiP/9zITCPfBLsDn3kchbSOu16JrnxtVEmL0fPRcI=
+github.com/jarcoal/httpmock v1.4.2/go.mod h1:ftW1xULwo+j0R0JJkJIIi7UKigZUXCLLanykgjwBXL0=
 github.com/jessevdk/go-flags v1.4.0/go.mod h1:4FA24M0QyGHXBuZZK/XkWh8h0e1EYbRYJSGM75WSRxI=
-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/klauspost/compress v1.19.2 h1:hMRETovs/pu/dVWN7zIT1PGG8t509MwT6bO7XSi26R8=
+github.com/klauspost/compress v1.19.2/go.mod h1:cwPg85FWrGar70rWktvGQj8/hthj3wpl0PGDogxkrSQ=
 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 v0.0.0-20260216142805-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=
-github.com/mattn/go-colorable v0.1.14/go.mod h1:6LmQG8QLFO4G5z1gPvYEzlUgJ2wF+stgPZH1UqBm1s8=
-github.com/mattn/go-isatty v0.0.20 h1:xfD0iDuEKnDkl03q4limB+vH+GxLEtL/jb4xVJSWWEY=
-github.com/mattn/go-isatty v0.0.20/go.mod h1:W+V8PltTTMOvKvAeJH7IuucS94S2C6jfK/D7dTCTo3Y=
+github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5 h1:eveIIGn4BGM3qknO74omf6HYr30/exH+eVUTuAgwjZ0=
+github.com/lufia/plan9stats v0.0.0-20260802145828-341c2f0c90b5/go.mod h1:autxFIvghDt3jPTLoqZ9OZ7s9qTGNAWmYCjVFWPX/zg=
+github.com/lxzan/gws v1.10.1 h1:1xG+tDOV0lgDeVPf0wNT74u3cn0K3LpcavRrTPTrMwQ=
+github.com/lxzan/gws v1.10.1/go.mod h1:gXHSCPmTGryWJ4icuqy8Yho32E4YIMHH0fkDRYJRbdc=
+github.com/mattn/go-colorable v0.1.15 h1:+u9SLTRGnXv73cEsnsmoZBom+dMU88B2M0aDcWy0/jY=
+github.com/mattn/go-colorable v0.1.15/go.mod h1:6LmQG8QLFO4G5z1gPvYEzlUgJ2wF+stgPZH1UqBm1s8=
+github.com/mattn/go-isatty v0.0.24 h1:tGZZoVgT/KiqK1c8ocVLeDS8BSWMRd47J3Lbz7vsReI=
+github.com/mattn/go-isatty v0.0.24/go.mod h1:nMCL3Zebbrt45jsMDgnfIwz6ydEQApk5oEI3HqDio6A=
 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 v0.13.2 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/nicholas-fedor/shoutrrr v0.17.0 h1:xfp3z5QbE8jXvUhUEwWDk47SJ/b912VoB8MJJDU+q4E=
+github.com/nicholas-fedor/shoutrrr v0.17.0/go.mod h1:s4ldyLs6uwBy9lIjYrY+8lyTqJtPvZSrILw0CyMLock=
+github.com/onsi/ginkgo/v2 v2.32.0 h1:Hw7s2pVrQo/8Yz5N77qdnpHaoc+c6cC9WIV1Jce+J6E=
+github.com/onsi/ginkgo/v2 v2.32.0/go.mod h1:+aXOY+vzZ5mu2iI2HpTZUPmM//oQfsNFX6gU9kNcA44=
+github.com/onsi/gomega v1.42.1 h1:iN1rCUX+44NZ1Dc97MPoeFYbFR0vh8zxoxMFwKdyZ6I=
+github.com/onsi/gomega v1.42.1/go.mod h1:REff/hsDsodHoKlWsP2mAPhu1+5/6hVYNf9rIEBpeSg=
 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 v1.12.0 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/pocketbase/ozzo-validation/v4 v4.3.0 h1:uKBDVma7bZqgR2a6AwE+k9hkuDFfiZMpBHQdZ1z3iQs=
+github.com/pocketbase/ozzo-validation/v4 v4.3.0/go.mod h1:6XNjSTw/Jb2F8LOkKO3oyzIWExbrGiYoS4uVxVwz90g=
+github.com/pocketbase/pocketbase v0.39.10 h1:2j8TDJRuo3aAC8Y8F9WFux0SwYcxeDCgEYQxxdWkwGE=
+github.com/pocketbase/pocketbase v0.39.10/go.mod h1:tSX3anHQ7Ul6dPV9WhlEc6No1DtklGF69iwnVNW3BEE=
+github.com/power-devops/perfstat v0.0.0-20260805114148-88456608a4f6 h1:jL3a8soXdzuTCcRnKhOmtcsVOObdDTFf4O2B403HPRU=
+github.com/power-devops/perfstat v0.0.0-20260805114148-88456608a4f6/go.mod h1:OmDBASR4679mdNQnz2pUhc2G8CO2JrUAVFDRBDP/hJE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec h1:W09IVJc94icq4NjY3clb7Lk8O1qJ8BdBEF8z0ibU0rE=
 github.com/remyoudompheng/bigfft v0.0.0-20230129092748-24d4a6f8daec/go.mod h1:qqbHyh8v60DhA7CoWK5oRCqLrMHRGoxYCSS9EjAz6Eo=
 github.com/rogpeppe/go-internal v1.9.0 h1:73kH8U+JUqXU8lRuOHeVHaa/SZPifC7BkcraZVejAe8=
 github.com/rogpeppe/go-internal v1.9.0/go.mod h1:WtVeX8xhTBvf0smdhujwtBcq4Qrzq/fJaraNFVN+nFs=
 github.com/russross/blackfriday/v2 v2.1.0/go.mod h1:+Rmxgy9KzJVeS9/2gXHxylqXiyQDYRxCVz55jmeOWTM=
-github.com/shirou/gopsutil/v4 v4.26.1 h1:TOkEyriIXk2HX9d4isZJtbjXbEjf5qyKPAzbzY0JWSo=
-github.com/shirou/gopsutil/v4 v4.26.1/go.mod h1:medLI9/UNAb0dOI9Q3/7yWSqKkj00u+1tgY8nvv41pc=
+github.com/shirou/gopsutil/v4 v4.26.7 h1:IXzpHz/dkMRYAhKkOXr1HB6SuzWU3eoyyeWe7g3bNZc=
+github.com/shirou/gopsutil/v4 v4.26.7/go.mod h1:5O9FjBiXoTDFatIWjZZosqj4pV0DRtLx598xGbBehzM=
 github.com/spf13/cast v1.10.0 h1:h2x0u2shc1QuLHfxi+cTJvs30+ZAHOGRic8uyGTDWxY=
 github.com/spf13/cast v1.10.0/go.mod h1:jNfB8QC9IA6ZuY2ZjDp0KtFO2LZZlg4S/7bzP6qqeHo=
 github.com/spf13/cobra v1.10.2 h1:DMTTonx5m65Ic0GOoRY2c16WCbHxOOw6xxezuLaBpcU=
@@ -115,55 +117,57 @@ github.com/spf13/pflag v1.0.9/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An
 github.com/spf13/pflag v1.0.10 h1:4EBh2KAYBwaONj6b2Ye1GiHfwjqyROoF4RwYO+vPwFk=
 github.com/spf13/pflag v1.0.10/go.mod h1:McXfInJRrz4CZXVZOBLb0bTZqETkiAhM9Iw0y3An2Bg=
 github.com/stretchr/objx v0.1.0/go.mod h1:HFkY916IF+rwdDfMAkV7OtwuqBVzrE8GR6GFx+wExME=
+github.com/stretchr/objx v0.5.3 h1:jmXUvGomnU1o3W/V5h2VEradbpJDwGrzugQQvL0POH4=
+github.com/stretchr/objx v0.5.3/go.mod h1:rDQraq+vQZU7Fde9LOZLr8Tax6zZvy4kuNKF+QYS+U0=
 github.com/stretchr/testify v1.4.0/go.mod h1:j7eGeouHqKxXV5pUuKE4zz7dFj8WfuZ+81PSLYec5m4=
 github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
 github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
-github.com/tklauser/go-sysconf v0.3.16 h1:frioLaCQSsF5Cy1jgRBrzr6t502KIIwQ0MArYICU0nA=
-github.com/tklauser/go-sysconf v0.3.16/go.mod h1:/qNL9xxDhc7tx3HSRsLWNnuzbVfh3e7gh/BmM179nYI=
-github.com/tklauser/numcpus v0.11.0 h1:nSTwhKH5e1dMNsCdVBukSZrURJRoHbSEQjdEbY+9RXw=
-github.com/tklauser/numcpus v0.11.0/go.mod h1:z+LwcLq54uWZTX0u/bGobaV34u6V7KNlTZejzM6/3MQ=
+github.com/tklauser/go-sysconf v0.4.0 h1:7H0uAN+7RkwWRaxhYXDLqa5V3LPrJeV8wmD9dRUgPQU=
+github.com/tklauser/go-sysconf v0.4.0/go.mod h1:8mTNWyog7H+MpKijp4VmKJAd2bbYQ2zuUwkYRbUArPI=
+github.com/tklauser/numcpus v0.12.0 h1:NR85qdvHA9pFse3x3weVZ0r0ST8R6l5RHbZrlRaqob4=
+github.com/tklauser/numcpus v0.12.0/go.mod h1:ABHeXzJnr/qqwguhClkZKT1/8VABcYrsyUiUGobwWJg=
 github.com/x448/float16 v0.8.4 h1:qLwI1I70+NjRFUR3zs1JPUCgaCXSh3SW62uAKT1mSBM=
 github.com/x448/float16 v0.8.4/go.mod h1:14CWIYCyZA/cWjXOioeEpHeN/83MdbZDRQHoFcYsOfg=
 github.com/yusufpapurcu/wmi v1.2.4 h1:zFUKzehAFReQwLys1b/iSMl+JQGSCSjtVqQn9bBrPo0=
 github.com/yusufpapurcu/wmi v1.2.4/go.mod h1:SBZ9tNy3G9/m5Oi98Zks0QjeHVDvuK0qfxQmPyzfmi0=
-go.yaml.in/yaml/v3 v3.0.4 h1:tfq32ie2Jv2UxXFdLJdh3jXuOzWiL1fo0bu/FbuKpbc=
+go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
+go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
 go.yaml.in/yaml/v3 v3.0.4/go.mod h1:DhzuOOF2ATzADvBadXxruRBLzYTpT36CKvDb3+aBEFg=
+go.yaml.in/yaml/v3 v3.0.5 h1:N6y/pJk8buWs9NY5ERU2HSMfm+IuD/OtfdAnq6kESPw=
+go.yaml.in/yaml/v3 v3.0.5/go.mod h1:HVTZu1O7/Vkt2N+BFy8Zza+lnLsABggaTM2ZpNIGuKg=
 golang.org/x/crypto v0.0.0-20190308221718-c2843e01d9a2/go.mod h1:djNgcEr1/C05ACkg1iLfiJU5Ep61QUkGW8qpdssI0+w=
-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/crypto v0.55.0 h1:+KWHjbgOaAQ66dh/YlkZKHlz9ZUlq61AFirAR9ntP8M=
+golang.org/x/crypto v0.55.0/go.mod h1:uq0V9dE/fzQuJtbnL+2EhWOE63vo164FY8xqEnV9xis=
+golang.org/x/exp v0.0.0-20260811152304-ee035b5b010f h1:iXpLj9sdDH/RLYsnOMpbETK6KWtrHwvegcc4psWJHV8=
+golang.org/x/exp v0.0.0-20260811152304-ee035b5b010f/go.mod h1:EdfpwwqSu+0Li0mzskwHU6FWDV3t9Q+RZDo3QMUtL3Q=
 golang.org/x/image v0.0.0-20191009234506-e7c1f5e7dbb8/go.mod h1:FeLwcggjj3mMvU+oOTbSwawSJRM1uh48EjtB4UJZlP0=
-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/image v0.45.0 h1:FMb1nTbH5H9vF55SriQHgFw5GnNL9Jg6L25BwXKzhB0=
+golang.org/x/image v0.45.0/go.mod h1:n62x/7RqlwXDvGsSU4u6IUTUf6KghUZ9Bt7cG/T9Fx4=
+golang.org/x/mod v0.38.0 h1:MECBjubtXD7yj4HrhIUcywNaGeNVUdfVnxmPajOk4yk=
+golang.org/x/mod v0.38.0/go.mod h1:V6Xz0pq8TQ3dGqVQ1FVHuelZpAL0uNhSkk9ogYP3c40=
 golang.org/x/net v0.0.0-20190603091049-60506f45cf65/go.mod h1:HSz+uSET+XFnRR8LxR5pz3Of3rY3CfYBVs4xY44aLks=
-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/net v0.57.0 h1:K5+3DljvIuDG9/Jv9rvyMywYNFCQ9RSUY6OOTTkT+tE=
+golang.org/x/net v0.57.0/go.mod h1:KpXc8iv+r3XplLAG/f7Jsf9RPszJzdR0f58q9vGOuEU=
+golang.org/x/oauth2 v0.36.0 h1:peZ/1z27fi9hUOFCAZaHyrpWG5lwe0RJEEEeH0ThlIs=
+golang.org/x/oauth2 v0.36.0/go.mod h1:YDBUJMTkDnJS+A4BP4eZBjCqtokkg1hODuPjwiGPO7Q=
+golang.org/x/sync v0.22.0 h1:SZjpbeLmrCk4xhRSZFNZW5gFUeCeFgjekvI/+gfScek=
+golang.org/x/sync v0.22.0/go.mod h1:9xrNwdLfx4jkKbNva9FpL6vEN7evnE43NNNJQ2LF3+0=
 golang.org/x/sys v0.0.0-20190215142949-d0b11bdaac8a/go.mod h1:STP8DvDyc/dI5b8T5hshtkjS+E42TnysNCUPdjciGhY=
 golang.org/x/sys v0.0.0-20190916202348-b4ddaad3f8a3/go.mod h1:h1NjWce9XRLGQEsW7wpKNCjG9DtNlClVuFLEZdDNbEs=
 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 v0.41.0 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/sys v0.47.0 h1:o7XGOvZQCADBQQ4Y7VNq2dRWQR7JmOUW8Kxx4ZsNgWs=
+golang.org/x/sys v0.47.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/term v0.45.0 h1:NwWyBmoJCbfTHpxrWoZ9C6/VxOf7ic219I8xZZFdrf0=
+golang.org/x/term v0.45.0/go.mod h1:9aqxs0blBcrm/n0L9QW0aRVD+ktan8ssZromtqJC43w=
 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 v0.34.0 h1:oL/Qq0Kdaqxa1KbNeMKwQq0reLCCaFtqu2eNuSeNHbk=
-golang.org/x/text v0.34.0/go.mod h1:homfLqTYRFyVYemLBFl5GgL/DWEiH5wcsQ5gSh1yziA=
+golang.org/x/text v0.41.0 h1:vz/seA0lnX87Othu2f/0L24RcgrXD9/YFTSuGjj3rH8=
+golang.org/x/text v0.41.0/go.mod h1:jvf1O8ajNzZqhSrQBPbutR/EB83Cc0CFrezNQIwbb5M=
 golang.org/x/tools v0.0.0-20180917221912-90fa682c2a6e/go.mod h1:n7NCudcB/nEzxVGmLbDWY5pfWTLqBcC2KZ6jyYvM4mQ=
-golang.org/x/tools v0.42.0 h1:uNgphsn75Tdz5Ji2q36v/nsFSfR/9BRFvqhGBaJGd5k=
-golang.org/x/tools v0.42.0/go.mod h1:Ma6lCIwGZvHK6XtgbswSoWroEkhugApmsXyrUmBhfr0=
+golang.org/x/tools v0.48.0 h1:3+hClM1aLL5mjMKm5ovokw9epgRXPuu2tILgismM6RE=
+golang.org/x/tools v0.48.0/go.mod h1:08xX0orndb/F7jJxGDicx061tyd5pcMto75YMAXr6lk=
 google.golang.org/appengine v1.6.5/go.mod h1:8WjMMxjGQR8xUklV/ARdw2HLXBOI7O7uCIDZVag1xfc=
-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=
@@ -173,30 +177,30 @@ gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
 gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
 howett.net/plist v1.0.1 h1:37GdZ8tP09Q35o9ych3ehygcsL+HqKSwzctveSlarvM=
 howett.net/plist v1.0.1/go.mod h1:lqaXoTrLY4hg8tnEzNru53gicrbv7rrk+2xJA/7hw9g=
-modernc.org/cc/v4 v4.27.1 h1:9W30zRlYrefrDV2JE2O8VDtJ1yPGownxciz5rrbQZis=
-modernc.org/cc/v4 v4.27.1/go.mod h1:uVtb5OGqUKpoLWhqwNQo/8LwvoiEBLvZXIQ/SmO6mL0=
-modernc.org/ccgo/v4 v4.30.1 h1:4r4U1J6Fhj98NKfSjnPUN7Ze2c6MnAdL0hWw6+LrJpc=
-modernc.org/ccgo/v4 v4.30.1/go.mod h1:bIOeI1JL54Utlxn+LwrFyjCx2n2RDiYEaJVSrgdrRfM=
-modernc.org/fileutil v1.3.40 h1:ZGMswMNc9JOCrcrakF1HrvmergNLAmxOPjizirpfqBA=
-modernc.org/fileutil v1.3.40/go.mod h1:HxmghZSZVAz/LXcMNwZPA/DRrQZEVP9VX0V4LQGQFOc=
+modernc.org/cc/v4 v4.29.0 h1:CXgwL8cvxmyzBQZzbSl/6xFtMCryb6u8IOqDci39cgc=
+modernc.org/cc/v4 v4.29.0/go.mod h1:OnovgIhbbMXMu1aISnJ0wvVD1KnW+cAUJkIrAWh+kVI=
+modernc.org/ccgo/v4 v4.34.6 h1:sBgfIwyN0TQ9C5hwIeuqyeAKyMWnbvj2fvpF4L11uzU=
+modernc.org/ccgo/v4 v4.34.6/go.mod h1:SZ8YcN9NG7XVsQYdm6jYBvi8PQP1qi+kqB6OhjqI3Fk=
+modernc.org/fileutil v1.4.0 h1:j6ZzNTftVS054gi281TyLjHPp6CPHr2KCxEXjEbD6SM=
+modernc.org/fileutil v1.4.0/go.mod h1:EqdKFDxiByqxLk8ozOxObDSfcVOv/54xDs/DUHdvCUU=
 modernc.org/gc/v2 v2.6.5 h1:nyqdV8q46KvTpZlsw66kWqwXRHdjIlJOhG6kxiV/9xI=
 modernc.org/gc/v2 v2.6.5/go.mod h1:YgIahr1ypgfe7chRuJi2gD7DBQiKSLMPgBQe9oIiito=
-modernc.org/gc/v3 v3.1.1 h1:k8T3gkXWY9sEiytKhcgyiZ2L0DTyCQ/nvX+LoCljoRE=
-modernc.org/gc/v3 v3.1.1/go.mod h1:HFK/6AGESC7Ex+EZJhJ2Gni6cTaYpSMmU/cT9RmlfYY=
+modernc.org/gc/v3 v3.1.4 h1:2g65LGVSmFQrXeITAw97x7hCRvZFcyE1uDP+7Vng7JI=
+modernc.org/gc/v3 v3.1.4/go.mod h1:HFK/6AGESC7Ex+EZJhJ2Gni6cTaYpSMmU/cT9RmlfYY=
 modernc.org/goabi0 v0.2.0 h1:HvEowk7LxcPd0eq6mVOAEMai46V+i7Jrj13t4AzuNks=
 modernc.org/goabi0 v0.2.0/go.mod h1:CEFRnnJhKvWT1c1JTI3Avm+tgOWbkOu5oPA8eH8LnMI=
-modernc.org/libc v1.67.6 h1:eVOQvpModVLKOdT+LvBPjdQqfrZq+pC39BygcT+E7OI=
-modernc.org/libc v1.67.6/go.mod h1:JAhxUVlolfYDErnwiqaLvUqc8nfb2r6S6slAgZOnaiE=
+modernc.org/libc v1.74.1 h1:bdR4VTKFMC4966QSNZ05XLGI/VwzVa2kTUX51Dm0riQ=
+modernc.org/libc v1.74.1/go.mod h1:uH4t5bOx3G3g9Xcmj10YKlTcVISlRDwv8VoQJG9n8Os=
 modernc.org/mathutil v1.7.1 h1:GCZVGXdaN8gTqB1Mf/usp1Y/hSqgI2vAGGP4jZMCxOU=
 modernc.org/mathutil v1.7.1/go.mod h1:4p5IwJITfppl0G4sUEDtCr4DthTaT47/N3aT6MhfgJg=
-modernc.org/memory v1.11.0 h1:o4QC8aMQzmcwCK3t3Ux/ZHmwFPzE6hf2Y5LbkRs+hbI=
-modernc.org/memory v1.11.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
-modernc.org/opt v0.1.4 h1:2kNGMRiUjrp4LcaPuLY2PzUfqM/w9N23quVwhKt5Qm8=
-modernc.org/opt v0.1.4/go.mod h1:03fq9lsNfvkYSfxrfUhZCWPk1lm4cq4N+Bh//bEtgns=
+modernc.org/memory v1.12.0 h1:twkmYNkGXCvtYWzoux02jtK6eovjZbdI0uHFUYp6kuU=
+modernc.org/memory v1.12.0/go.mod h1:/JP4VbVC+K5sU2wZi9bHoq2MAkCnrt2r98UGeSK7Mjw=
+modernc.org/opt v0.2.0 h1:tGyef5ApycA7FSEOMraay9SaTk5zmbx7Tu+cJs4QKZg=
+modernc.org/opt v0.2.0/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 v1.45.0 h1:r51cSGzKpbptxnby+EIIz5fop4VuE4qFoVEjNvWoObs=
-modernc.org/sqlite v1.45.0/go.mod h1:CzbrU2lSB1DKUusvwGz7rqEKIq+NUd8GWuBBZDs9/nA=
+modernc.org/sqlite v1.55.0 h1:hIFh0MCH0rGinQ/4KYb5/UbCkRkb+UP+OkLCVWa5MTM=
+modernc.org/sqlite v1.55.0/go.mod h1:4ntCLuNmnH8+GNqjka1wNg7KJd5/Hi5FYp8K+XQ7GZw=
 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

@@ -23,6 +23,7 @@ type AlertManager struct {
 	hub           hubLike
 	stopOnce      sync.Once
 	pendingAlerts sync.Map
+	alertsCache   *AlertsCache
 }
 
 type AlertMessageData struct {
@@ -54,6 +55,7 @@ type SystemAlertStats struct {
 	Temperatures map[string]float32            `json:"t"`
 	LoadAvg      [3]float64                    `json:"la"`
 	Battery      [2]uint8                      `json:"bat"`
+	Batteries    map[string]uint8              `json:"bats"`
 	ExtraFs      map[string]SystemAlertFsStats `json:"efs"`
 }
 
@@ -63,7 +65,7 @@ type SystemAlertGPUData struct {
 
 type SystemAlertData struct {
 	systemRecord *core.Record
-	alertRecord  *core.Record
+	alertData    CachedAlertData
 	name         string
 	unit         string
 	val          float64
@@ -97,7 +99,8 @@ var supportsTitle = map[string]struct{}{
 // NewAlertManager creates a new AlertManager instance.
 func NewAlertManager(app hubLike) *AlertManager {
 	am := &AlertManager{
-		hub: app,
+		hub:         app,
+		alertsCache: NewAlertsCache(app),
 	}
 	am.bindEvents()
 	return am
@@ -110,6 +113,9 @@ func (am *AlertManager) bindEvents() {
 	am.hub.OnRecordAfterUpdateSuccess("smart_devices").BindFunc(am.handleSmartDeviceAlert)
 
 	am.hub.OnServe().BindFunc(func(e *core.ServeEvent) error {
+		// Populate all alerts into cache on startup
+		_ = am.alertsCache.PopulateFromDB(true)
+
 		if err := resolveStatusAlerts(e.App); err != nil {
 			e.App.Logger().Error("Failed to resolve stale status alerts", "err", err)
 		}
@@ -297,17 +303,12 @@ func (am *AlertManager) SendShoutrrrAlert(notificationUrl, title, message, link,
 	return nil
 }
 
-func (am *AlertManager) SendTestNotification(e *core.RequestEvent) error {
-	var data struct {
-		URL string `json:"url"`
-	}
-	err := e.BindBody(&data)
-	if err != nil || data.URL == "" {
-		return e.BadRequestError("URL is required", err)
-	}
-	err = am.SendShoutrrrAlert(data.URL, "Test Alert", "This is a notification from Beszel.", am.hub.Settings().Meta.AppURL, "View Beszel")
+// setAlertTriggered updates the "triggered" status of an alert record in the database
+func (am *AlertManager) setAlertTriggered(alert CachedAlertData, triggered bool) error {
+	alertRecord, err := am.hub.FindRecordById("alerts", alert.Id)
 	if err != nil {
-		return e.JSON(200, map[string]string{"err": err.Error()})
+		return err
 	}
-	return e.JSON(200, map[string]bool{"err": false})
+	alertRecord.Set("triggered", triggered)
+	return am.hub.Save(alertRecord)
 }

internal/alerts/alerts_api.go

@@ -3,7 +3,11 @@ package alerts
 import (
 	"database/sql"
 	"errors"
+	"net"
 	"net/http"
+	"net/url"
+	"slices"
+	"strings"
 
 	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
@@ -117,3 +121,72 @@ func DeleteUserAlerts(e *core.RequestEvent) error {
 
 	return e.JSON(http.StatusOK, map[string]any{"success": true, "count": numDeleted})
 }
+
+// SendTestNotification handles API request to send a test notification to a specified Shoutrrr URL
+func (am *AlertManager) SendTestNotification(e *core.RequestEvent) error {
+	var data struct {
+		URL string `json:"url"`
+	}
+	err := e.BindBody(&data)
+	if err != nil || data.URL == "" {
+		return e.BadRequestError("URL is required", err)
+	}
+	// Only allow admins to send test notifications to internal URLs
+	if !e.Auth.IsSuperuser() && e.Auth.GetString("role") != "admin" {
+		internalURL, err := isInternalURL(data.URL)
+		if err != nil {
+			return e.BadRequestError(err.Error(), nil)
+		}
+		if internalURL {
+			return e.ForbiddenError("Only admins can send to internal destinations", nil)
+		}
+	}
+	err = am.SendShoutrrrAlert(data.URL, "Test Alert", "This is a notification from Beszel.", am.hub.Settings().Meta.AppURL, "View Beszel")
+	if err != nil {
+		return e.JSON(200, map[string]string{"err": err.Error()})
+	}
+	return e.JSON(200, map[string]bool{"err": false})
+}
+
+// isInternalURL checks if the given shoutrrr URL points to an internal destination (localhost or private IP)
+func isInternalURL(rawURL string) (bool, error) {
+	parsedURL, err := url.Parse(rawURL)
+	if err != nil {
+		return false, err
+	}
+
+	host := parsedURL.Hostname()
+	if host == "" {
+		return false, nil
+	}
+
+	if strings.EqualFold(host, "localhost") {
+		return true, nil
+	}
+
+	if ip := net.ParseIP(host); ip != nil {
+		return isInternalIP(ip), nil
+	}
+
+	// Some Shoutrrr URLs use the host position for service identifiers rather than a
+	// network hostname (for example, discord://token@webhookid). Restrict DNS lookups
+	// to names that look like actual hostnames so valid service URLs keep working.
+	if !strings.Contains(host, ".") {
+		return false, nil
+	}
+
+	ips, err := net.LookupIP(host)
+	if err != nil {
+		return false, nil
+	}
+
+	if slices.ContainsFunc(ips, isInternalIP) {
+		return true, nil
+	}
+
+	return false, nil
+}
+
+func isInternalIP(ip net.IP) bool {
+	return ip.IsPrivate() || ip.IsLoopback() || ip.IsUnspecified()
+}

internal/alerts/alerts_api_test.go

@@ -0,0 +1,501 @@
+//go:build testing
+
+package alerts_test
+
+import (
+	"bytes"
+	"encoding/json"
+	"io"
+	"net/http"
+	"strings"
+	"testing"
+
+	"github.com/henrygd/beszel/internal/alerts"
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+	pbTests "github.com/pocketbase/pocketbase/tests"
+
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+
+	"github.com/stretchr/testify/assert"
+)
+
+// marshal to json and return an io.Reader (for use in ApiScenario.Body)
+func jsonReader(v any) io.Reader {
+	data, err := json.Marshal(v)
+	if err != nil {
+		panic(err)
+	}
+	return bytes.NewReader(data)
+}
+
+func TestIsInternalURL(t *testing.T) {
+	testCases := []struct {
+		name     string
+		url      string
+		internal bool
+	}{
+		{name: "loopback ipv4", url: "generic://127.0.0.1", internal: true},
+		{name: "localhost hostname", url: "generic://localhost", internal: true},
+		{name: "localhost hostname", url: "generic+http://localhost/api/v1/postStuff", internal: true},
+		{name: "localhost hostname", url: "generic+http://127.0.0.1:8080/api/v1/postStuff", internal: true},
+		{name: "localhost hostname", url: "generic+https://beszel.dev/api/v1/postStuff", internal: false},
+		{name: "public ipv4", url: "generic://8.8.8.8", internal: false},
+		{name: "token style service url", url: "discord://abc123@123456789", internal: false},
+		{name: "single label service url", url: "slack://token@team/channel", internal: false},
+	}
+
+	for _, testCase := range testCases {
+		t.Run(testCase.name, func(t *testing.T) {
+			internal, err := alerts.IsInternalURL(testCase.url)
+			assert.NoError(t, err)
+			assert.Equal(t, testCase.internal, internal)
+		})
+	}
+}
+
+func TestUserAlertsApi(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	hub.StartHub()
+
+	user1, _ := beszelTests.CreateUser(hub, "alertstest@example.com", "password")
+	user1Token, _ := user1.NewAuthToken()
+
+	user2, _ := beszelTests.CreateUser(hub, "alertstest2@example.com", "password")
+	user2Token, _ := user2.NewAuthToken()
+
+	system1, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system1",
+		"users": []string{user1.Id},
+		"host":  "127.0.0.1",
+	})
+
+	system2, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system2",
+		"users": []string{user1.Id, user2.Id},
+		"host":  "127.0.0.2",
+	})
+
+	userRecords, _ := hub.CountRecords("users")
+	assert.EqualValues(t, 2, userRecords, "all users should be created")
+
+	systemRecords, _ := hub.CountRecords("systems")
+	assert.EqualValues(t, 2, systemRecords, "all systems should be created")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		// {
+		// 	Name:            "GET not implemented - returns index",
+		// 	Method:          http.MethodGet,
+		// 	URL:             "/api/beszel/user-alerts",
+		// 	ExpectedStatus:  200,
+		// 	ExpectedContent: []string{"<html ", "globalThis.BESZEL"},
+		// 	TestAppFactory:  testAppFactory,
+		// },
+		{
+			Name:            "POST no auth",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST no body",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST bad data",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"invalidField": "this should cause validation error",
+				"threshold":    "not a number",
+			}),
+		},
+		{
+			Name:   "POST malformed JSON",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Bad data"},
+			TestAppFactory:  testAppFactory,
+			Body:            strings.NewReader(`{"alertType": "cpu", "threshold": 80, "enabled": true,}`),
+		},
+		{
+			Name:   "POST valid alert data multiple systems",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     69,
+				"min":       9,
+				"systems":   []string{system1.Id, system2.Id},
+				"overwrite": false,
+			}),
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				// check total alerts
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+				// check alert has correct values
+				matchingAlerts, _ := app.CountRecords("alerts", dbx.HashExp{"name": "CPU", "user": user1.Id, "system": system1.Id, "value": 69, "min": 9})
+				assert.EqualValues(t, 1, matchingAlerts, "should have 1 alert")
+			},
+		},
+		{
+			Name:   "POST valid alert data single system",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "Memory",
+				"systems": []string{system1.Id},
+				"value":   90,
+				"min":     10,
+			}),
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				user1Alerts, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 3, user1Alerts, "should have 3 alerts")
+			},
+		},
+		{
+			Name:   "Overwrite: false, should not overwrite existing alert",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     45,
+				"min":       5,
+				"systems":   []string{system1.Id},
+				"overwrite": false,
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user1.Id})
+				assert.EqualValues(t, 80, alert.Get("value"), "should have 80 as value")
+			},
+		},
+		{
+			Name:   "Overwrite: true, should overwrite existing alert",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":      "CPU",
+				"value":     45,
+				"min":       5,
+				"systems":   []string{system2.Id},
+				"overwrite": true,
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system2.Id,
+					"user":   user2.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user2.Id})
+				assert.EqualValues(t, 45, alert.Get("value"), "should have 45 as value")
+			},
+		},
+		{
+			Name:            "DELETE no auth",
+			Method:          http.MethodDelete,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system1.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 1, alerts, "should have 1 alert")
+			},
+		},
+		{
+			Name:   "DELETE alert",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system1.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system1.Id,
+					"user":   user1.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.Zero(t, alerts, "should have 0 alerts")
+			},
+		},
+		{
+			Name:   "DELETE alert multiple systems",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":2", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "Memory",
+				"systems": []string{system1.Id, system2.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				for _, systemId := range []string{system1.Id, system2.Id} {
+					_, err := beszelTests.CreateRecord(app, "alerts", map[string]any{
+						"name":   "Memory",
+						"system": systemId,
+						"user":   user1.Id,
+						"value":  90,
+						"min":    10,
+					})
+					assert.NoError(t, err, "should create alert")
+				}
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				alerts, _ := app.CountRecords("alerts")
+				assert.Zero(t, alerts, "should have 0 alerts")
+			},
+		},
+		{
+			Name:   "User 2 should not be able to delete alert of user 1",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system2.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.ClearCollection(t, app, "alerts")
+				for _, user := range []string{user1.Id, user2.Id} {
+					beszelTests.CreateRecord(app, "alerts", map[string]any{
+						"name":   "CPU",
+						"system": system2.Id,
+						"user":   user,
+						"value":  80,
+						"min":    10,
+					})
+				}
+				alerts, _ := app.CountRecords("alerts")
+				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
+				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
+				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
+				assert.EqualValues(t, 1, user2AlertCount, "should have 1 alert")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
+				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
+				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
+				assert.Zero(t, user2AlertCount, "should have 0 alerts")
+			},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}
+func TestSendTestNotification(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	userToken, err := user.NewAuthToken()
+
+	adminUser, err := beszelTests.CreateUserWithRole(hub, "admin@example.com", "password123", "admin")
+	assert.NoError(t, err, "Failed to create admin user")
+	adminUserToken, err := adminUser.NewAuthToken()
+
+	superuser, err := beszelTests.CreateSuperuser(hub, "superuser@example.com", "password123")
+	assert.NoError(t, err, "Failed to create superuser")
+	superuserToken, err := superuser.NewAuthToken()
+	assert.NoError(t, err, "Failed to create superuser auth token")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:            "POST /test-notification - no auth should fail",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/test-notification",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+		},
+		{
+			Name:           "POST /test-notification - with external auth should succeed",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic://8.8.8.8",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"err\":"},
+		},
+		{
+			Name:           "POST /test-notification - local url with user auth should fail",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic://localhost:8010",
+			}),
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Only admins"},
+		},
+		{
+			Name:           "POST /test-notification - internal url with user auth should fail",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic+http://192.168.0.5",
+			}),
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Only admins"},
+		},
+		{
+			Name:           "POST /test-notification - internal url with admin auth should succeed",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": adminUserToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"err\":"},
+		},
+		{
+			Name:           "POST /test-notification - internal url with superuser auth should succeed",
+			Method:         http.MethodPost,
+			URL:            "/api/beszel/test-notification",
+			TestAppFactory: testAppFactory,
+			Headers: map[string]string{
+				"Authorization": superuserToken,
+			},
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"err\":"},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}

internal/alerts/alerts_cache.go

@@ -0,0 +1,177 @@
+package alerts
+
+import (
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/pocketbase/pocketbase/tools/store"
+)
+
+// CachedAlertData represents the relevant fields of an alert record for status checking and updates.
+type CachedAlertData struct {
+	Id        string
+	SystemID  string
+	UserID    string
+	Name      string
+	Value     float64
+	Triggered bool
+	Min       uint8
+	// Created   types.DateTime
+}
+
+func (a *CachedAlertData) PopulateFromRecord(record *core.Record) {
+	a.Id = record.Id
+	a.SystemID = record.GetString("system")
+	a.UserID = record.GetString("user")
+	a.Name = record.GetString("name")
+	a.Value = record.GetFloat("value")
+	a.Triggered = record.GetBool("triggered")
+	a.Min = uint8(record.GetInt("min"))
+	// a.Created = record.GetDateTime("created")
+}
+
+// AlertsCache provides an in-memory cache for system alerts.
+type AlertsCache struct {
+	app       core.App
+	store     *store.Store[string, *store.Store[string, CachedAlertData]]
+	populated bool
+}
+
+// NewAlertsCache creates a new instance of SystemAlertsCache.
+func NewAlertsCache(app core.App) *AlertsCache {
+	c := AlertsCache{
+		app:   app,
+		store: store.New(map[string]*store.Store[string, CachedAlertData]{}),
+	}
+	return c.bindEvents()
+}
+
+// bindEvents sets up event listeners to keep the cache in sync with database changes.
+func (c *AlertsCache) bindEvents() *AlertsCache {
+	c.app.OnRecordAfterUpdateSuccess("alerts").BindFunc(func(e *core.RecordEvent) error {
+		// c.Delete(e.Record.Original()) // this would be needed if the system field on an existing alert was changed, however we don't currently allow that in the UI so we'll leave it commented out
+		c.Update(e.Record)
+		return e.Next()
+	})
+	c.app.OnRecordAfterDeleteSuccess("alerts").BindFunc(func(e *core.RecordEvent) error {
+		c.Delete(e.Record)
+		return e.Next()
+	})
+	c.app.OnRecordAfterCreateSuccess("alerts").BindFunc(func(e *core.RecordEvent) error {
+		c.Update(e.Record)
+		return e.Next()
+	})
+	return c
+}
+
+// PopulateFromDB clears current entries and loads all alerts from the database into the cache.
+func (c *AlertsCache) PopulateFromDB(force bool) error {
+	if !force && c.populated {
+		return nil
+	}
+	records, err := c.app.FindAllRecords("alerts")
+	if err != nil {
+		return err
+	}
+	c.store.RemoveAll()
+	for _, record := range records {
+		c.Update(record)
+	}
+	c.populated = true
+	return nil
+}
+
+// Update adds or updates an alert record in the cache.
+func (c *AlertsCache) Update(record *core.Record) {
+	systemID := record.GetString("system")
+	if systemID == "" {
+		return
+	}
+	systemStore, ok := c.store.GetOk(systemID)
+	if !ok {
+		systemStore = store.New(map[string]CachedAlertData{})
+		c.store.Set(systemID, systemStore)
+	}
+	var ca CachedAlertData
+	ca.PopulateFromRecord(record)
+	systemStore.Set(record.Id, ca)
+}
+
+// Delete removes an alert record from the cache.
+func (c *AlertsCache) Delete(record *core.Record) {
+	systemID := record.GetString("system")
+	if systemID == "" {
+		return
+	}
+	if systemStore, ok := c.store.GetOk(systemID); ok {
+		systemStore.Remove(record.Id)
+	}
+}
+
+// GetSystemAlerts returns all alerts for the specified system, lazy-loading if necessary.
+func (c *AlertsCache) GetSystemAlerts(systemID string) []CachedAlertData {
+	systemStore, ok := c.store.GetOk(systemID)
+	if !ok {
+		// Populate cache for this system
+		records, err := c.app.FindAllRecords("alerts", dbx.NewExp("system={:system}", dbx.Params{"system": systemID}))
+		if err != nil {
+			return nil
+		}
+		systemStore = store.New(map[string]CachedAlertData{})
+		for _, record := range records {
+			var ca CachedAlertData
+			ca.PopulateFromRecord(record)
+			systemStore.Set(record.Id, ca)
+		}
+		c.store.Set(systemID, systemStore)
+	}
+	all := systemStore.GetAll()
+	alerts := make([]CachedAlertData, 0, len(all))
+	for _, alert := range all {
+		alerts = append(alerts, alert)
+	}
+	return alerts
+}
+
+// GetAlert returns a specific alert by its ID from the cache.
+func (c *AlertsCache) GetAlert(systemID, alertID string) (CachedAlertData, bool) {
+	if systemStore, ok := c.store.GetOk(systemID); ok {
+		return systemStore.GetOk(alertID)
+	}
+	return CachedAlertData{}, false
+}
+
+// GetAlertsByName returns all alerts of a specific type for the specified system.
+func (c *AlertsCache) GetAlertsByName(systemID, alertName string) []CachedAlertData {
+	allAlerts := c.GetSystemAlerts(systemID)
+	var alerts []CachedAlertData
+	for _, record := range allAlerts {
+		if record.Name == alertName {
+			alerts = append(alerts, record)
+		}
+	}
+	return alerts
+}
+
+// GetAlertsExcludingNames returns all alerts for the specified system excluding the given types.
+func (c *AlertsCache) GetAlertsExcludingNames(systemID string, excludedNames ...string) []CachedAlertData {
+	excludeMap := make(map[string]struct{})
+	for _, name := range excludedNames {
+		excludeMap[name] = struct{}{}
+	}
+	allAlerts := c.GetSystemAlerts(systemID)
+	var alerts []CachedAlertData
+	for _, record := range allAlerts {
+		if _, excluded := excludeMap[record.Name]; !excluded {
+			alerts = append(alerts, record)
+		}
+	}
+	return alerts
+}
+
+// Refresh returns the latest cached copy for an alert snapshot if it still exists.
+func (c *AlertsCache) Refresh(alert CachedAlertData) (CachedAlertData, bool) {
+	if alert.Id == "" {
+		return CachedAlertData{}, false
+	}
+	return c.GetAlert(alert.SystemID, alert.Id)
+}

internal/alerts/alerts_cache_test.go

@@ -0,0 +1,215 @@
+//go:build testing
+
+package alerts_test
+
+import (
+	"testing"
+
+	"github.com/henrygd/beszel/internal/alerts"
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestSystemAlertsCachePopulateAndFilter(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	systems, err := beszelTests.CreateSystems(hub, 2, user.Id, "up")
+	require.NoError(t, err)
+	system1 := systems[0]
+	system2 := systems[1]
+
+	statusAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "Status",
+		"system": system1.Id,
+		"user":   user.Id,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	cpuAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "CPU",
+		"system": system1.Id,
+		"user":   user.Id,
+		"value":  80,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	memoryAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "Memory",
+		"system": system2.Id,
+		"user":   user.Id,
+		"value":  90,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	cache := alerts.NewAlertsCache(hub)
+	cache.PopulateFromDB(false)
+
+	statusAlerts := cache.GetAlertsByName(system1.Id, "Status")
+	require.Len(t, statusAlerts, 1)
+	assert.Equal(t, statusAlert.Id, statusAlerts[0].Id)
+
+	nonStatusAlerts := cache.GetAlertsExcludingNames(system1.Id, "Status")
+	require.Len(t, nonStatusAlerts, 1)
+	assert.Equal(t, cpuAlert.Id, nonStatusAlerts[0].Id)
+
+	system2Alerts := cache.GetSystemAlerts(system2.Id)
+	require.Len(t, system2Alerts, 1)
+	assert.Equal(t, memoryAlert.Id, system2Alerts[0].Id)
+}
+
+func TestSystemAlertsCacheLazyLoadUpdateAndDelete(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]
+
+	statusAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "Status",
+		"system": systemRecord.Id,
+		"user":   user.Id,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	cache := alerts.NewAlertsCache(hub)
+	require.Len(t, cache.GetSystemAlerts(systemRecord.Id), 1, "first lookup should lazy-load alerts for the system")
+
+	cpuAlert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "CPU",
+		"system": systemRecord.Id,
+		"user":   user.Id,
+		"value":  80,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	cache.Update(cpuAlert)
+
+	nonStatusAlerts := cache.GetAlertsExcludingNames(systemRecord.Id, "Status")
+	require.Len(t, nonStatusAlerts, 1)
+	assert.Equal(t, cpuAlert.Id, nonStatusAlerts[0].Id)
+
+	cache.Delete(statusAlert)
+	assert.Empty(t, cache.GetAlertsByName(systemRecord.Id, "Status"), "deleted alerts should be removed from the in-memory cache")
+}
+
+func TestSystemAlertsCacheRefreshReturnsLatestCopy(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+	require.NoError(t, err)
+	system := systems[0]
+
+	alert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":      "Status",
+		"system":    system.Id,
+		"user":      user.Id,
+		"min":       1,
+		"triggered": false,
+	})
+	require.NoError(t, err)
+
+	cache := alerts.NewAlertsCache(hub)
+	snapshot := cache.GetSystemAlerts(system.Id)[0]
+	assert.False(t, snapshot.Triggered)
+
+	alert.Set("triggered", true)
+	require.NoError(t, hub.Save(alert))
+
+	refreshed, ok := cache.Refresh(snapshot)
+	require.True(t, ok)
+	assert.Equal(t, snapshot.Id, refreshed.Id)
+	assert.True(t, refreshed.Triggered, "refresh should return the updated cached value rather than the stale snapshot")
+
+	require.NoError(t, hub.Delete(alert))
+	_, ok = cache.Refresh(snapshot)
+	assert.False(t, ok, "refresh should report false when the cached alert no longer exists")
+}
+
+func TestAlertManagerCacheLifecycle(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+	require.NoError(t, err)
+	system := systems[0]
+
+	// Create an alert
+	alert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   "CPU",
+		"system": system.Id,
+		"user":   user.Id,
+		"value":  80,
+		"min":    1,
+	})
+	require.NoError(t, err)
+
+	am := hub.AlertManager
+	cache := am.GetSystemAlertsCache()
+
+	// Verify it's in cache (it should be since CreateRecord triggers the event)
+	assert.Len(t, cache.GetSystemAlerts(system.Id), 1)
+	assert.Equal(t, alert.Id, cache.GetSystemAlerts(system.Id)[0].Id)
+	assert.EqualValues(t, 80, cache.GetSystemAlerts(system.Id)[0].Value)
+
+	// Update the alert through PocketBase to trigger events
+	alert.Set("value", 85)
+	require.NoError(t, hub.Save(alert))
+
+	// Check if updated value is reflected (or at least that it's still there)
+	cachedAlerts := cache.GetSystemAlerts(system.Id)
+	assert.Len(t, cachedAlerts, 1)
+	assert.EqualValues(t, 85, cachedAlerts[0].Value)
+
+	// Delete the alert through PocketBase to trigger events
+	require.NoError(t, hub.Delete(alert))
+
+	// Verify it's removed from cache
+	assert.Empty(t, cache.GetSystemAlerts(system.Id), "alert should be removed from cache after PocketBase delete")
+}
+
+// func TestAlertManagerCacheMovesAlertToNewSystemOnUpdate(t *testing.T) {
+// 	hub, user := beszelTests.GetHubWithUser(t)
+// 	defer hub.Cleanup()
+
+// 	systems, err := beszelTests.CreateSystems(hub, 2, user.Id, "up")
+// 	require.NoError(t, err)
+// 	system1 := systems[0]
+// 	system2 := systems[1]
+
+// 	alert, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+// 		"name":   "CPU",
+// 		"system": system1.Id,
+// 		"user":   user.Id,
+// 		"value":  80,
+// 		"min":    1,
+// 	})
+// 	require.NoError(t, err)
+
+// 	am := hub.AlertManager
+// 	cache := am.GetSystemAlertsCache()
+
+// 	// Initially in system1 cache
+// 	assert.Len(t, cache.Get(system1.Id), 1)
+// 	assert.Empty(t, cache.Get(system2.Id))
+
+// 	// Move alert to system2
+// 	alert.Set("system", system2.Id)
+// 	require.NoError(t, hub.Save(alert))
+
+// 	// DEBUG: print if it is found
+// 	// fmt.Printf("system1 alerts after update: %v\n", cache.Get(system1.Id))
+
+// 	// Should be removed from system1 and present in system2
+// 	assert.Empty(t, cache.GetType(system1.Id, "CPU"), "updated alerts should be evicted from the previous system cache")
+// 	require.Len(t, cache.Get(system2.Id), 1)
+// 	assert.Equal(t, alert.Id, cache.Get(system2.Id)[0].Id)
+// }

internal/alerts/alerts_quiet_hours_test.go

@@ -322,8 +322,9 @@ func TestAlertSilencedMultiUser(t *testing.T) {
 }
 
 func TestAlertSilencedWithActualAlert(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
 	synctest.Test(t, func(t *testing.T) {
-		hub, user := beszelTests.GetHubWithUser(t)
 		defer hub.Cleanup()
 
 		// Create a system

internal/alerts/alerts_status.go

@@ -5,15 +5,14 @@ import (
 	"strings"
 	"time"
 
-	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 )
 
 type alertInfo struct {
-	systemName  string
-	alertRecord *core.Record
-	expireTime  time.Time
-	timer       *time.Timer
+	systemName string
+	alertData  CachedAlertData
+	expireTime time.Time
+	timer      *time.Timer
 }
 
 // Stop cancels all pending status alert timers.
@@ -36,76 +35,61 @@ func (am *AlertManager) HandleStatusAlerts(newStatus string, systemRecord *core.
 		return nil
 	}
 
-	alertRecords, err := am.getSystemStatusAlerts(systemRecord.Id)
-	if err != nil {
-		return err
-	}
-	if len(alertRecords) == 0 {
+	alerts := am.alertsCache.GetAlertsByName(systemRecord.Id, "Status")
+	if len(alerts) == 0 {
 		return nil
 	}
 
 	systemName := systemRecord.GetString("name")
 	if newStatus == "down" {
-		am.handleSystemDown(systemName, alertRecords)
+		am.handleSystemDown(systemName, alerts)
 	} else {
-		am.handleSystemUp(systemName, alertRecords)
+		am.handleSystemUp(systemName, alerts)
 	}
 	return nil
 }
 
-// getSystemStatusAlerts retrieves all "Status" alert records for a given system ID.
-func (am *AlertManager) getSystemStatusAlerts(systemID string) ([]*core.Record, error) {
-	alertRecords, err := am.hub.FindAllRecords("alerts", dbx.HashExp{
-		"system": systemID,
-		"name":   "Status",
-	})
-	if err != nil {
-		return nil, err
-	}
-	return alertRecords, nil
-}
-
 // handleSystemDown manages the logic when a system status changes to "down". It schedules pending alerts for each alert record.
-func (am *AlertManager) handleSystemDown(systemName string, alertRecords []*core.Record) {
-	for _, alertRecord := range alertRecords {
-		min := max(1, alertRecord.GetInt("min"))
-		am.schedulePendingStatusAlert(systemName, alertRecord, time.Duration(min)*time.Minute)
+func (am *AlertManager) handleSystemDown(systemName string, alerts []CachedAlertData) {
+	for _, alertData := range alerts {
+		min := max(1, int(alertData.Min))
+		am.schedulePendingStatusAlert(systemName, alertData, time.Duration(min)*time.Minute)
 	}
 }
 
 // schedulePendingStatusAlert sets up a timer to send a "down" alert after the specified delay if the system is still down.
 // It returns true if the alert was scheduled, or false if an alert was already pending for the given alert record.
-func (am *AlertManager) schedulePendingStatusAlert(systemName string, alertRecord *core.Record, delay time.Duration) bool {
+func (am *AlertManager) schedulePendingStatusAlert(systemName string, alertData CachedAlertData, delay time.Duration) bool {
 	alert := &alertInfo{
-		systemName:  systemName,
-		alertRecord: alertRecord,
-		expireTime:  time.Now().Add(delay),
+		systemName: systemName,
+		alertData:  alertData,
+		expireTime: time.Now().Add(delay),
 	}
 
-	storedAlert, loaded := am.pendingAlerts.LoadOrStore(alertRecord.Id, alert)
+	storedAlert, loaded := am.pendingAlerts.LoadOrStore(alertData.Id, alert)
 	if loaded {
 		return false
 	}
 
 	stored := storedAlert.(*alertInfo)
 	stored.timer = time.AfterFunc(time.Until(stored.expireTime), func() {
-		am.processPendingAlert(alertRecord.Id)
+		am.processPendingAlert(alertData.Id)
 	})
 	return true
 }
 
 // handleSystemUp manages the logic when a system status changes to "up".
 // It cancels any pending alerts and sends "up" alerts.
-func (am *AlertManager) handleSystemUp(systemName string, alertRecords []*core.Record) {
-	for _, alertRecord := range alertRecords {
+func (am *AlertManager) handleSystemUp(systemName string, alerts []CachedAlertData) {
+	for _, alertData := range alerts {
 		// If alert exists for record, delete and continue (down alert not sent)
-		if am.cancelPendingAlert(alertRecord.Id) {
+		if am.cancelPendingAlert(alertData.Id) {
 			continue
 		}
-		if !alertRecord.GetBool("triggered") {
+		if !alertData.Triggered {
 			continue
 		}
-		if err := am.sendStatusAlert("up", systemName, alertRecord); err != nil {
+		if err := am.sendStatusAlert("up", systemName, alertData); err != nil {
 			am.hub.Logger().Error("Failed to send alert", "err", err)
 		}
 	}
@@ -125,6 +109,18 @@ func (am *AlertManager) cancelPendingAlert(alertID string) bool {
 	return true
 }
 
+// CancelPendingStatusAlerts cancels all pending status alert timers for a given system.
+// This is called when a system is paused to prevent delayed alerts from firing.
+func (am *AlertManager) CancelPendingStatusAlerts(systemID string) {
+	am.pendingAlerts.Range(func(key, value any) bool {
+		info := value.(*alertInfo)
+		if info.alertData.SystemID == systemID {
+			am.cancelPendingAlert(key.(string))
+		}
+		return true
+	})
+}
+
 // processPendingAlert sends a "down" alert if the pending alert has expired and the system is still down.
 func (am *AlertManager) processPendingAlert(alertID string) {
 	value, loaded := am.pendingAlerts.LoadAndDelete(alertID)
@@ -133,23 +129,22 @@ func (am *AlertManager) processPendingAlert(alertID string) {
 	}
 
 	info := value.(*alertInfo)
-	if info.alertRecord.GetBool("triggered") {
+	refreshedAlertData, ok := am.alertsCache.Refresh(info.alertData)
+	if !ok || refreshedAlertData.Triggered {
 		return
 	}
-	if err := am.sendStatusAlert("down", info.systemName, info.alertRecord); err != nil {
+	if err := am.sendStatusAlert("down", info.systemName, refreshedAlertData); err != nil {
 		am.hub.Logger().Error("Failed to send alert", "err", err)
 	}
 }
 
 // sendStatusAlert sends a status alert ("up" or "down") to the users associated with the alert records.
-func (am *AlertManager) sendStatusAlert(alertStatus string, systemName string, alertRecord *core.Record) error {
-	switch alertStatus {
-	case "up":
-		alertRecord.Set("triggered", false)
-	case "down":
-		alertRecord.Set("triggered", true)
+func (am *AlertManager) sendStatusAlert(alertStatus string, systemName string, alertData CachedAlertData) error {
+	// Update trigger state for alert record before sending alert
+	triggered := alertStatus == "down"
+	if err := am.setAlertTriggered(alertData, triggered); err != nil {
+		return err
 	}
-	am.hub.Save(alertRecord)
 
 	var emoji string
 	if alertStatus == "up" {
@@ -162,10 +157,10 @@ func (am *AlertManager) sendStatusAlert(alertStatus string, systemName string, a
 	message := strings.TrimSuffix(title, emoji)
 
 	// Get system ID for the link
-	systemID := alertRecord.GetString("system")
+	systemID := alertData.SystemID
 
 	return am.SendAlert(AlertMessageData{
-		UserID:   alertRecord.GetString("user"),
+		UserID:   alertData.UserID,
 		SystemID: systemID,
 		Title:    title,
 		Message:  message,
@@ -211,12 +206,13 @@ func resolveStatusAlerts(app core.App) error {
 func (am *AlertManager) restorePendingStatusAlerts() error {
 	type pendingStatusAlert struct {
 		AlertID    string `db:"alert_id"`
+		SystemID   string `db:"system_id"`
 		SystemName string `db:"system_name"`
 	}
 
 	var pending []pendingStatusAlert
 	err := am.hub.DB().NewQuery(`
-		SELECT a.id AS alert_id, s.name AS system_name
+		SELECT a.id AS alert_id, a.system AS system_id, s.name AS system_name
 		FROM alerts a
 		JOIN systems s ON a.system = s.id
 		WHERE a.name = 'Status'
@@ -227,13 +223,16 @@ func (am *AlertManager) restorePendingStatusAlerts() error {
 		return err
 	}
 
+	// Make sure cache is populated before trying to restore pending alerts
+	_ = am.alertsCache.PopulateFromDB(false)
+
 	for _, item := range pending {
-		alertRecord, err := am.hub.FindRecordById("alerts", item.AlertID)
-		if err != nil {
-			return err
+		alertData, ok := am.alertsCache.GetAlert(item.SystemID, item.AlertID)
+		if !ok {
+			continue
 		}
-		min := max(1, alertRecord.GetInt("min"))
-		am.schedulePendingStatusAlert(item.SystemName, alertRecord, time.Duration(min)*time.Minute)
+		min := max(1, int(alertData.Min))
+		am.schedulePendingStatusAlert(item.SystemName, alertData, time.Duration(min)*time.Minute)
 	}
 
 	return nil

internal/alerts/alerts_status_test.go

@@ -15,9 +15,23 @@ import (
 	"github.com/stretchr/testify/require"
 )
 
+func setStatusAlertEmail(t *testing.T, hub core.App, userID, email string) {
+	t.Helper()
+
+	userSettings, err := hub.FindFirstRecordByFilter("user_settings", "user={:user}", map[string]any{"user": userID})
+	require.NoError(t, err)
+
+	userSettings.Set("settings", map[string]any{
+		"emails":   []string{email},
+		"webhooks": []string{},
+	})
+	require.NoError(t, hub.Save(userSettings))
+}
+
 func TestStatusAlerts(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
 	synctest.Test(t, func(t *testing.T) {
-		hub, user := beszelTests.GetHubWithUser(t)
 		defer hub.Cleanup()
 
 		systems, err := beszelTests.CreateSystems(hub, 4, user.Id, "paused")
@@ -221,8 +235,9 @@ func TestHandleStatusAlertsDoesNotSendRecoveryWhileDownIsOnlyPending(t *testing.
 }
 
 func TestStatusAlertTimerCancellationPreventsBoundaryDelivery(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
 	synctest.Test(t, func(t *testing.T) {
-		hub, user := beszelTests.GetHubWithUser(t)
 		defer hub.Cleanup()
 
 		userSettings, err := hub.FindFirstRecordByFilter("user_settings", "user={:user}", map[string]any{"user": user.Id})
@@ -322,6 +337,183 @@ func TestStatusAlertDownFiresAfterDelayExpires(t *testing.T) {
 	assert.True(t, alertRecord.GetBool("triggered"), "alert should be marked triggered after downtime matures")
 }
 
+func TestStatusAlertMultipleUsersRespectDifferentMinutes(t *testing.T) {
+	hub, user1 := beszelTests.GetHubWithUser(t)
+
+	synctest.Test(t, func(t *testing.T) {
+		defer hub.Cleanup()
+
+		setStatusAlertEmail(t, hub, user1.Id, "user1@example.com")
+
+		user2, err := beszelTests.CreateUser(hub, "user2@example.com", "password")
+		require.NoError(t, err)
+		_, err = beszelTests.CreateRecord(hub, "user_settings", map[string]any{
+			"user": user2.Id,
+			"settings": map[string]any{
+				"emails":   []string{"user2@example.com"},
+				"webhooks": []string{},
+			},
+		})
+		require.NoError(t, err)
+
+		system, err := beszelTests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "shared-system",
+			"users": []string{user1.Id, user2.Id},
+			"host":  "127.0.0.1",
+		})
+		require.NoError(t, err)
+		system.Set("status", "up")
+		require.NoError(t, hub.SaveNoValidate(system))
+
+		alertUser1, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user1.Id,
+			"min":    1,
+		})
+		require.NoError(t, err)
+		alertUser2, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user2.Id,
+			"min":    2,
+		})
+		require.NoError(t, err)
+
+		time.Sleep(10 * time.Millisecond)
+
+		system.Set("status", "down")
+		require.NoError(t, hub.SaveNoValidate(system))
+
+		assert.Equal(t, 2, hub.GetPendingAlertsCount(), "both user alerts should be pending after the system goes down")
+
+		time.Sleep(59 * time.Second)
+		synctest.Wait()
+		assert.Zero(t, hub.TestMailer.TotalSend(), "no messages should be sent before the earliest alert minute elapses")
+
+		time.Sleep(2 * time.Second)
+		synctest.Wait()
+
+		messages := hub.TestMailer.Messages()
+		require.Len(t, messages, 1, "only the first user's alert should send after one minute")
+		require.Len(t, messages[0].To, 1)
+		assert.Equal(t, "user1@example.com", messages[0].To[0].Address)
+		assert.Contains(t, messages[0].Subject, "Connection to shared-system is down")
+		assert.Equal(t, 1, hub.GetPendingAlertsCount(), "the later user alert should still be pending")
+
+		time.Sleep(58 * time.Second)
+		synctest.Wait()
+		assert.Equal(t, 1, hub.TestMailer.TotalSend(), "the second user's alert should still be waiting before two minutes")
+
+		time.Sleep(2 * time.Second)
+		synctest.Wait()
+
+		messages = hub.TestMailer.Messages()
+		require.Len(t, messages, 2, "both users should eventually receive their own status alert")
+		require.Len(t, messages[1].To, 1)
+		assert.Equal(t, "user2@example.com", messages[1].To[0].Address)
+		assert.Contains(t, messages[1].Subject, "Connection to shared-system is down")
+		assert.Zero(t, hub.GetPendingAlertsCount(), "all pending alerts should be consumed after both timers fire")
+
+		alertUser1, err = hub.FindRecordById("alerts", alertUser1.Id)
+		require.NoError(t, err)
+		assert.True(t, alertUser1.GetBool("triggered"), "user1 alert should be marked triggered after delivery")
+
+		alertUser2, err = hub.FindRecordById("alerts", alertUser2.Id)
+		require.NoError(t, err)
+		assert.True(t, alertUser2.GetBool("triggered"), "user2 alert should be marked triggered after delivery")
+	})
+}
+
+func TestStatusAlertMultipleUsersRecoveryBetweenMinutesOnlyAlertsEarlierUser(t *testing.T) {
+	hub, user1 := beszelTests.GetHubWithUser(t)
+
+	synctest.Test(t, func(t *testing.T) {
+		defer hub.Cleanup()
+
+		setStatusAlertEmail(t, hub, user1.Id, "user1@example.com")
+
+		user2, err := beszelTests.CreateUser(hub, "user2@example.com", "password")
+		require.NoError(t, err)
+		_, err = beszelTests.CreateRecord(hub, "user_settings", map[string]any{
+			"user": user2.Id,
+			"settings": map[string]any{
+				"emails":   []string{"user2@example.com"},
+				"webhooks": []string{},
+			},
+		})
+		require.NoError(t, err)
+
+		system, err := beszelTests.CreateRecord(hub, "systems", map[string]any{
+			"name":  "shared-system",
+			"users": []string{user1.Id, user2.Id},
+			"host":  "127.0.0.1",
+		})
+		require.NoError(t, err)
+		system.Set("status", "up")
+		require.NoError(t, hub.SaveNoValidate(system))
+
+		alertUser1, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user1.Id,
+			"min":    1,
+		})
+		require.NoError(t, err)
+		alertUser2, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user2.Id,
+			"min":    2,
+		})
+		require.NoError(t, err)
+
+		time.Sleep(10 * time.Millisecond)
+
+		system.Set("status", "down")
+		require.NoError(t, hub.SaveNoValidate(system))
+
+		time.Sleep(61 * time.Second)
+		synctest.Wait()
+
+		messages := hub.TestMailer.Messages()
+		require.Len(t, messages, 1, "the first user's down alert should send before recovery")
+		require.Len(t, messages[0].To, 1)
+		assert.Equal(t, "user1@example.com", messages[0].To[0].Address)
+		assert.Contains(t, messages[0].Subject, "Connection to shared-system is down")
+		assert.Equal(t, 1, hub.GetPendingAlertsCount(), "the second user's alert should still be pending")
+
+		system.Set("status", "up")
+		require.NoError(t, hub.SaveNoValidate(system))
+
+		time.Sleep(time.Second)
+		synctest.Wait()
+
+		messages = hub.TestMailer.Messages()
+		require.Len(t, messages, 2, "recovery should notify only the user whose down alert had already triggered")
+		for _, message := range messages {
+			require.Len(t, message.To, 1)
+			assert.Equal(t, "user1@example.com", message.To[0].Address)
+		}
+		assert.Contains(t, messages[1].Subject, "Connection to shared-system is up")
+		assert.Zero(t, hub.GetPendingAlertsCount(), "recovery should cancel the later user's pending alert")
+
+		time.Sleep(61 * time.Second)
+		synctest.Wait()
+
+		messages = hub.TestMailer.Messages()
+		require.Len(t, messages, 2, "user2 should never receive a down alert once recovery cancels the pending timer")
+
+		alertUser1, err = hub.FindRecordById("alerts", alertUser1.Id)
+		require.NoError(t, err)
+		assert.False(t, alertUser1.GetBool("triggered"), "user1 alert should be cleared after recovery")
+
+		alertUser2, err = hub.FindRecordById("alerts", alertUser2.Id)
+		require.NoError(t, err)
+		assert.False(t, alertUser2.GetBool("triggered"), "user2 alert should remain untriggered because it never fired")
+	})
+}
+
 func TestStatusAlertDuplicateDownCallIsIdempotent(t *testing.T) {
 	hub, user := beszelTests.GetHubWithUser(t)
 	defer hub.Cleanup()
@@ -626,3 +818,197 @@ func TestResolveStatusAlerts(t *testing.T) {
 	assert.EqualValues(t, 1, alertHistoryCount, "Should have exactly one unresolved alert history record")
 
 }
+
+func TestAlertsHistoryStatus(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
+	synctest.Test(t, func(t *testing.T) {
+		defer hub.Cleanup()
+
+		// Create a system
+		systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+		assert.NoError(t, err)
+		system := systems[0]
+
+		// Create a status alertRecord for the system
+		alertRecord, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user.Id,
+			"min":    1,
+		})
+		assert.NoError(t, err)
+
+		// Verify alert is not triggered initially
+		assert.False(t, alertRecord.GetBool("triggered"), "Alert should not be triggered initially")
+
+		// Set the system to 'down' (this should trigger the alert)
+		system.Set("status", "down")
+		err = hub.Save(system)
+		assert.NoError(t, err)
+
+		time.Sleep(time.Second * 30)
+		synctest.Wait()
+
+		alertFresh, _ := hub.FindRecordById("alerts", alertRecord.Id)
+		assert.False(t, alertFresh.GetBool("triggered"), "Alert should not be triggered after 30 seconds")
+
+		time.Sleep(time.Minute)
+		synctest.Wait()
+
+		// Verify alert is triggered after setting system to down
+		alertFresh, err = hub.FindRecordById("alerts", alertRecord.Id)
+		assert.NoError(t, err)
+		assert.True(t, alertFresh.GetBool("triggered"), "Alert should be triggered after one minute")
+
+		// Verify we have one unresolved alert history record
+		alertHistoryCount, err := hub.CountRecords("alerts_history", dbx.HashExp{"resolved": ""})
+		assert.NoError(t, err)
+		assert.EqualValues(t, 1, alertHistoryCount, "Should have exactly one unresolved alert history record")
+
+		// Set the system back to 'up' (this should resolve the alert)
+		system.Set("status", "up")
+		err = hub.Save(system)
+		assert.NoError(t, err)
+
+		time.Sleep(time.Second)
+		synctest.Wait()
+
+		// Verify alert is not triggered after setting system back to up
+		alertFresh, err = hub.FindRecordById("alerts", alertRecord.Id)
+		assert.NoError(t, err)
+		assert.False(t, alertFresh.GetBool("triggered"), "Alert should not be triggered after system recovers")
+
+		// Verify the alert history record is resolved
+		alertHistoryCount, err = hub.CountRecords("alerts_history", dbx.HashExp{"resolved": ""})
+		assert.NoError(t, err)
+		assert.EqualValues(t, 0, alertHistoryCount, "Should have no unresolved alert history records")
+	})
+}
+
+func TestStatusAlertClearedBeforeSend(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
+	synctest.Test(t, func(t *testing.T) {
+		defer hub.Cleanup()
+
+		// Create a system
+		systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+		assert.NoError(t, err)
+		system := systems[0]
+
+		// Ensure user settings have an email
+		userSettings, _ := hub.FindFirstRecordByFilter("user_settings", "user={:user}", map[string]any{"user": user.Id})
+		userSettings.Set("settings", `{"emails":["test@example.com"],"webhooks":[]}`)
+		hub.Save(userSettings)
+
+		// Initial email count
+		initialEmailCount := hub.TestMailer.TotalSend()
+
+		// Create a status alertRecord for the system
+		alertRecord, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+			"name":   "Status",
+			"system": system.Id,
+			"user":   user.Id,
+			"min":    1,
+		})
+		assert.NoError(t, err)
+
+		// Verify alert is not triggered initially
+		assert.False(t, alertRecord.GetBool("triggered"), "Alert should not be triggered initially")
+
+		// Set the system to 'down' (this should trigger the alert)
+		system.Set("status", "down")
+		err = hub.Save(system)
+		assert.NoError(t, err)
+
+		time.Sleep(time.Second * 30)
+		synctest.Wait()
+
+		// Set system back up to clear the pending alert before it triggers
+		system.Set("status", "up")
+		err = hub.Save(system)
+		assert.NoError(t, err)
+
+		time.Sleep(time.Minute)
+		synctest.Wait()
+
+		// Verify that we have not sent any emails since the system recovered before the alert triggered
+		assert.Equal(t, initialEmailCount, hub.TestMailer.TotalSend(), "No email should be sent if system recovers before alert triggers")
+
+		// Verify alert is not triggered after setting system back to up
+		alertFresh, err := hub.FindRecordById("alerts", alertRecord.Id)
+		assert.NoError(t, err)
+		assert.False(t, alertFresh.GetBool("triggered"), "Alert should not be triggered after system recovers")
+
+		// Verify that no alert history record was created since the alert never triggered
+		alertHistoryCount, err := hub.CountRecords("alerts_history")
+		assert.NoError(t, err)
+		assert.EqualValues(t, 0, alertHistoryCount, "Should have no unresolved alert history records since alert never triggered")
+	})
+}
+
+func TestCancelPendingStatusAlertsClearsAllAlertsForSystem(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	userSettings, err := hub.FindFirstRecordByFilter("user_settings", "user={:user}", map[string]any{"user": user.Id})
+	require.NoError(t, err)
+	userSettings.Set("settings", `{"emails":["test@example.com"],"webhooks":[]}`)
+	require.NoError(t, hub.Save(userSettings))
+
+	systemCollection, err := hub.FindCollectionByNameOrId("systems")
+	require.NoError(t, err)
+
+	system1 := core.NewRecord(systemCollection)
+	system1.Set("name", "system-1")
+	system1.Set("status", "up")
+	system1.Set("host", "127.0.0.1")
+	system1.Set("users", []string{user.Id})
+	require.NoError(t, hub.Save(system1))
+
+	system2 := core.NewRecord(systemCollection)
+	system2.Set("name", "system-2")
+	system2.Set("status", "up")
+	system2.Set("host", "127.0.0.2")
+	system2.Set("users", []string{user.Id})
+	require.NoError(t, hub.Save(system2))
+
+	alertCollection, err := hub.FindCollectionByNameOrId("alerts")
+	require.NoError(t, err)
+
+	alert1 := core.NewRecord(alertCollection)
+	alert1.Set("user", user.Id)
+	alert1.Set("system", system1.Id)
+	alert1.Set("name", "Status")
+	alert1.Set("triggered", false)
+	alert1.Set("min", 5)
+	require.NoError(t, hub.Save(alert1))
+
+	alert2 := core.NewRecord(alertCollection)
+	alert2.Set("user", user.Id)
+	alert2.Set("system", system2.Id)
+	alert2.Set("name", "Status")
+	alert2.Set("triggered", false)
+	alert2.Set("min", 5)
+	require.NoError(t, hub.Save(alert2))
+
+	am := alerts.NewTestAlertManagerWithoutWorker(hub)
+	initialEmailCount := hub.TestMailer.TotalSend()
+
+	// Both systems go down
+	require.NoError(t, am.HandleStatusAlerts("down", system1))
+	require.NoError(t, am.HandleStatusAlerts("down", system2))
+	assert.Equal(t, 2, am.GetPendingAlertsCount(), "both systems should have pending alerts")
+
+	// System 1 is paused — cancel its pending alerts
+	am.CancelPendingStatusAlerts(system1.Id)
+	assert.Equal(t, 1, am.GetPendingAlertsCount(), "only system2 alert should remain pending after pausing system1")
+
+	// Expire and process remaining alerts — only system2 should fire
+	am.ForceExpirePendingAlerts()
+	processed, err := am.ProcessPendingAlerts()
+	require.NoError(t, err)
+	assert.Len(t, processed, 1, "only the non-paused system's alert should be processed")
+	assert.Equal(t, initialEmailCount+1, hub.TestMailer.TotalSend(), "only system2 should send a down notification")
+}

internal/alerts/alerts_system.go

@@ -14,11 +14,8 @@ import (
 )
 
 func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error {
-	alertRecords, err := am.hub.FindAllRecords("alerts",
-		dbx.NewExp("system={:system} AND name!='Status'", dbx.Params{"system": systemRecord.Id}),
-	)
-	if err != nil || len(alertRecords) == 0 {
-		// log.Println("no alerts found for system")
+	alerts := am.alertsCache.GetAlertsExcludingNames(systemRecord.Id, "Status")
+	if len(alerts) == 0 {
 		return nil
 	}
 
@@ -26,8 +23,8 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 	now := systemRecord.GetDateTime("updated").Time().UTC()
 	oldestTime := now
 
-	for _, alertRecord := range alertRecords {
-		name := alertRecord.GetString("name")
+	for _, alertData := range alerts {
+		name := alertData.Name
 		var val float64
 		unit := "%"
 
@@ -66,14 +63,14 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 		case "GPU":
 			val = data.Info.GpuPct
 		case "Battery":
-			if data.Stats.Battery[0] == 0 {
+			if !hasRepresentativeBattery(data.Stats.Battery, data.Stats.Batteries) {
 				continue
 			}
 			val = float64(data.Stats.Battery[0])
 		}
 
-		triggered := alertRecord.GetBool("triggered")
-		threshold := alertRecord.GetFloat("value")
+		triggered := alertData.Triggered
+		threshold := alertData.Value
 
 		// Battery alert has inverted logic: trigger when value is BELOW threshold
 		lowAlert := isLowAlert(name)
@@ -91,11 +88,11 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 			}
 		}
 
-		min := max(1, uint8(alertRecord.GetInt("min")))
+		min := max(1, alertData.Min)
 
 		alert := SystemAlertData{
 			systemRecord: systemRecord,
-			alertRecord:  alertRecord,
+			alertData:    alertData,
 			name:         name,
 			unit:         unit,
 			val:          val,
@@ -128,7 +125,7 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 		Created types.DateTime `db:"created"`
 	}{}
 
-	err = am.hub.DB().
+	err := am.hub.DB().
 		Select("stats", "created").
 		From("system_stats").
 		Where(dbx.NewExp(
@@ -170,6 +167,7 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 		stat := systemStats[i]
 		// subtract 10 seconds to give a small time buffer
 		systemStatsCreation := stat.Created.Time().Add(-time.Second * 10)
+		stats = SystemAlertStats{}
 		if err := json.Unmarshal(stat.Stats, &stats); err != nil {
 			return err
 		}
@@ -238,6 +236,9 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 				}
 				alert.val += maxUsage
 			case "Battery":
+				if !hasRepresentativeBattery(stats.Battery, stats.Batteries) {
+					continue
+				}
 				alert.val += float64(stats.Battery[0])
 			default:
 				continue
@@ -300,6 +301,10 @@ func (am *AlertManager) HandleSystemAlerts(systemRecord *core.Record, data *syst
 	return nil
 }
 
+func hasRepresentativeBattery(legacy [2]uint8, batteries map[string]uint8) bool {
+	return legacy != [2]uint8{} || len(batteries) > 0
+}
+
 func (am *AlertManager) sendSystemAlert(alert SystemAlertData) {
 	// log.Printf("Sending alert %s: val %f | count %d | threshold %f\n", alert.name, alert.val, alert.count, alert.threshold)
 	systemName := alert.systemRecord.GetString("name")
@@ -343,13 +348,12 @@ func (am *AlertManager) sendSystemAlert(alert SystemAlertData) {
 	}
 	body := fmt.Sprintf("%s averaged %.2f%s for the previous %v %s.", alert.descriptor, alert.val, alert.unit, alert.min, minutesLabel)
 
-	alert.alertRecord.Set("triggered", alert.triggered)
-	if err := am.hub.Save(alert.alertRecord); err != nil {
+	if err := am.setAlertTriggered(alert.alertData, alert.triggered); err != nil {
 		// app.Logger().Error("failed to save alert record", "err", err)
 		return
 	}
 	am.SendAlert(AlertMessageData{
-		UserID:   alert.alertRecord.GetString("user"),
+		UserID:   alert.alertData.UserID,
 		SystemID: alert.systemRecord.Id,
 		Title:    subject,
 		Message:  body,

internal/alerts/alerts_system_test.go

@@ -0,0 +1,216 @@
+//go:build testing
+
+package alerts_test
+
+import (
+	"testing"
+	"testing/synctest"
+	"time"
+
+	"github.com/henrygd/beszel/internal/entities/system"
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+type systemAlertValueSetter[T any] func(info *system.Info, stats *system.Stats, value T)
+
+type systemAlertTestFixture struct {
+	hub     *beszelTests.TestHub
+	alertID string
+	submit  func(*system.CombinedData) error
+}
+
+func createCombinedData[T any](value T, setValue systemAlertValueSetter[T]) *system.CombinedData {
+	var data system.CombinedData
+	setValue(&data.Info, &data.Stats, value)
+	return &data
+}
+
+func newSystemAlertTestFixture(t *testing.T, alertName string, min int, threshold float64) *systemAlertTestFixture {
+	t.Helper()
+
+	hub, user := beszelTests.GetHubWithUser(t)
+
+	systems, err := beszelTests.CreateSystems(hub, 1, user.Id, "up")
+	require.NoError(t, err)
+	systemRecord := systems[0]
+
+	sysManagerSystem, err := hub.GetSystemManager().GetSystemFromStore(systemRecord.Id)
+	require.NoError(t, err)
+	require.NotNil(t, sysManagerSystem)
+	sysManagerSystem.StopUpdater()
+
+	userSettings, err := hub.FindFirstRecordByFilter("user_settings", "user={:user}", map[string]any{"user": user.Id})
+	require.NoError(t, err)
+	userSettings.Set("settings", `{"emails":["test@example.com"],"webhooks":[]}`)
+	require.NoError(t, hub.Save(userSettings))
+
+	alertRecord, err := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":   alertName,
+		"system": systemRecord.Id,
+		"user":   user.Id,
+		"min":    min,
+		"value":  threshold,
+	})
+	require.NoError(t, err)
+
+	assert.False(t, alertRecord.GetBool("triggered"), "Alert should not be triggered initially")
+
+	alertsCache := hub.GetAlertManager().GetSystemAlertsCache()
+	cachedAlerts := alertsCache.GetAlertsExcludingNames(systemRecord.Id, "Status")
+	assert.Len(t, cachedAlerts, 1, "Alert should be in cache")
+
+	return &systemAlertTestFixture{
+		hub:     hub,
+		alertID: alertRecord.Id,
+		submit: func(data *system.CombinedData) error {
+			_, err := sysManagerSystem.CreateRecords(data)
+			return err
+		},
+	}
+}
+
+func (fixture *systemAlertTestFixture) cleanup() {
+	fixture.hub.Cleanup()
+}
+
+func submitValue[T any](fixture *systemAlertTestFixture, t *testing.T, value T, setValue systemAlertValueSetter[T]) {
+	t.Helper()
+	require.NoError(t, fixture.submit(createCombinedData(value, setValue)))
+}
+
+func (fixture *systemAlertTestFixture) assertTriggered(t *testing.T, triggered bool, message string) {
+	t.Helper()
+
+	alertRecord, err := fixture.hub.FindRecordById("alerts", fixture.alertID)
+	require.NoError(t, err)
+	assert.Equal(t, triggered, alertRecord.GetBool("triggered"), message)
+}
+
+func waitForSystemAlert(d time.Duration) {
+	time.Sleep(d)
+	synctest.Wait()
+}
+
+func testOneMinuteSystemAlert[T any](t *testing.T, alertName string, threshold float64, setValue systemAlertValueSetter[T], triggerValue, resolveValue T) {
+	t.Helper()
+	fixture := newSystemAlertTestFixture(t, alertName, 1, threshold)
+	defer fixture.cleanup()
+
+	synctest.Test(t, func(t *testing.T) {
+		submitValue(fixture, t, triggerValue, setValue)
+		waitForSystemAlert(time.Second)
+
+		fixture.assertTriggered(t, true, "Alert should be triggered")
+		assert.Equal(t, 1, fixture.hub.TestMailer.TotalSend(), "An email should have been sent")
+
+		submitValue(fixture, t, resolveValue, setValue)
+		waitForSystemAlert(time.Second)
+
+		fixture.assertTriggered(t, false, "Alert should be untriggered")
+		assert.Equal(t, 2, fixture.hub.TestMailer.TotalSend(), "A second email should have been sent for untriggering the alert")
+
+		waitForSystemAlert(time.Minute)
+	})
+}
+
+func testMultiMinuteSystemAlert[T any](t *testing.T, alertName string, threshold float64, min int, setValue systemAlertValueSetter[T], baselineValue, triggerValue, resolveValue T) {
+	t.Helper()
+	fixture := newSystemAlertTestFixture(t, alertName, min, threshold)
+	defer fixture.cleanup()
+
+	synctest.Test(t, func(t *testing.T) {
+		submitValue(fixture, t, baselineValue, setValue)
+		waitForSystemAlert(time.Minute + time.Second)
+		fixture.assertTriggered(t, false, "Alert should not be triggered yet")
+
+		submitValue(fixture, t, triggerValue, setValue)
+		waitForSystemAlert(time.Minute)
+		fixture.assertTriggered(t, false, "Alert should not be triggered until the history window is full")
+
+		submitValue(fixture, t, triggerValue, setValue)
+		waitForSystemAlert(time.Second)
+		fixture.assertTriggered(t, true, "Alert should be triggered")
+		assert.Equal(t, 1, fixture.hub.TestMailer.TotalSend(), "An email should have been sent")
+
+		submitValue(fixture, t, resolveValue, setValue)
+		waitForSystemAlert(time.Second)
+		fixture.assertTriggered(t, false, "Alert should be untriggered")
+		assert.Equal(t, 2, fixture.hub.TestMailer.TotalSend(), "A second email should have been sent for untriggering the alert")
+	})
+}
+
+func setCPUAlertValue(info *system.Info, stats *system.Stats, value float64) {
+	info.Cpu = value
+	stats.Cpu = value
+}
+
+func setMemoryAlertValue(info *system.Info, stats *system.Stats, value float64) {
+	info.MemPct = value
+	stats.MemPct = value
+}
+
+func setDiskAlertValue(info *system.Info, stats *system.Stats, value float64) {
+	info.DiskPct = value
+	stats.DiskPct = value
+}
+
+func setBandwidthAlertValue(info *system.Info, stats *system.Stats, value [2]uint64) {
+	info.BandwidthBytes = value[0] + value[1]
+	stats.Bandwidth = value
+}
+
+func megabytesToBytes(mb uint64) uint64 {
+	return mb * 1024 * 1024
+}
+
+func setGPUAlertValue(info *system.Info, stats *system.Stats, value float64) {
+	info.GpuPct = value
+	stats.GPUData = map[string]system.GPUData{
+		"GPU0": {Usage: value},
+	}
+}
+
+func setTemperatureAlertValue(info *system.Info, stats *system.Stats, value float64) {
+	info.DashboardTemp = value
+	stats.Temperatures = map[string]float64{
+		"Temp0": value,
+	}
+}
+
+func setLoadAvgAlertValue(info *system.Info, stats *system.Stats, value [3]float64) {
+	info.LoadAvg = value
+	stats.LoadAvg = value
+}
+
+func setBatteryAlertValue(info *system.Info, stats *system.Stats, value [2]uint8) {
+	info.Battery = value
+	stats.Battery = value
+}
+
+func TestSystemAlertsOneMin(t *testing.T) {
+	testOneMinuteSystemAlert(t, "CPU", 50, setCPUAlertValue, 51, 49)
+	testOneMinuteSystemAlert(t, "Memory", 50, setMemoryAlertValue, 51, 49)
+	testOneMinuteSystemAlert(t, "Disk", 50, setDiskAlertValue, 51, 49)
+	testOneMinuteSystemAlert(t, "Bandwidth", 50, setBandwidthAlertValue, [2]uint64{megabytesToBytes(26), megabytesToBytes(25)}, [2]uint64{megabytesToBytes(25), megabytesToBytes(24)})
+	testOneMinuteSystemAlert(t, "GPU", 50, setGPUAlertValue, 51, 49)
+	testOneMinuteSystemAlert(t, "Temperature", 70, setTemperatureAlertValue, 71, 69)
+	testOneMinuteSystemAlert(t, "LoadAvg1", 4, setLoadAvgAlertValue, [3]float64{4.1, 0, 0}, [3]float64{3.9, 0, 0})
+	testOneMinuteSystemAlert(t, "LoadAvg5", 4, setLoadAvgAlertValue, [3]float64{0, 4.1, 0}, [3]float64{0, 3.9, 0})
+	testOneMinuteSystemAlert(t, "LoadAvg15", 4, setLoadAvgAlertValue, [3]float64{0, 0, 4.1}, [3]float64{0, 0, 3.9})
+	testOneMinuteSystemAlert(t, "Battery", 20, setBatteryAlertValue, [2]uint8{0, 1}, [2]uint8{21, 0})
+}
+
+func TestSystemAlertsTwoMin(t *testing.T) {
+	testMultiMinuteSystemAlert(t, "CPU", 50, 2, setCPUAlertValue, 10, 51, 48)
+	testMultiMinuteSystemAlert(t, "Memory", 50, 2, setMemoryAlertValue, 10, 51, 48)
+	testMultiMinuteSystemAlert(t, "Disk", 50, 2, setDiskAlertValue, 10, 51, 48)
+	testMultiMinuteSystemAlert(t, "Bandwidth", 50, 2, setBandwidthAlertValue, [2]uint64{megabytesToBytes(10), megabytesToBytes(10)}, [2]uint64{megabytesToBytes(26), megabytesToBytes(25)}, [2]uint64{megabytesToBytes(10), megabytesToBytes(10)})
+	testMultiMinuteSystemAlert(t, "GPU", 50, 2, setGPUAlertValue, 10, 51, 48)
+	testMultiMinuteSystemAlert(t, "Temperature", 70, 2, setTemperatureAlertValue, 10, 71, 67)
+	testMultiMinuteSystemAlert(t, "LoadAvg1", 4, 2, setLoadAvgAlertValue, [3]float64{0, 0, 0}, [3]float64{4.1, 0, 0}, [3]float64{3.5, 0, 0})
+	testMultiMinuteSystemAlert(t, "LoadAvg5", 4, 2, setLoadAvgAlertValue, [3]float64{0, 2, 0}, [3]float64{0, 4.1, 0}, [3]float64{0, 3.5, 0})
+	testMultiMinuteSystemAlert(t, "LoadAvg15", 4, 2, setLoadAvgAlertValue, [3]float64{0, 0, 2}, [3]float64{0, 0, 4.1}, [3]float64{0, 0, 3.5})
+	testMultiMinuteSystemAlert(t, "Battery", 20, 2, setBatteryAlertValue, [2]uint8{21, 0}, [2]uint8{19, 0}, [2]uint8{25, 1})
+}

internal/alerts/alerts_test.go

@@ -3,374 +3,21 @@
 package alerts_test
 
 import (
-	"bytes"
-	"encoding/json"
-	"io"
-	"net/http"
-	"strings"
 	"testing"
 	"testing/synctest"
 	"time"
 
 	beszelTests "github.com/henrygd/beszel/internal/tests"
 
+	"github.com/henrygd/beszel/internal/alerts"
 	"github.com/pocketbase/dbx"
-	"github.com/pocketbase/pocketbase/core"
-	pbTests "github.com/pocketbase/pocketbase/tests"
 	"github.com/stretchr/testify/assert"
 )
 
-// marshal to json and return an io.Reader (for use in ApiScenario.Body)
-func jsonReader(v any) io.Reader {
-	data, err := json.Marshal(v)
-	if err != nil {
-		panic(err)
-	}
-	return bytes.NewReader(data)
-}
-
-func TestUserAlertsApi(t *testing.T) {
-	hub, _ := beszelTests.NewTestHub(t.TempDir())
-	defer hub.Cleanup()
-
-	hub.StartHub()
-
-	user1, _ := beszelTests.CreateUser(hub, "alertstest@example.com", "password")
-	user1Token, _ := user1.NewAuthToken()
-
-	user2, _ := beszelTests.CreateUser(hub, "alertstest2@example.com", "password")
-	user2Token, _ := user2.NewAuthToken()
-
-	system1, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
-		"name":  "system1",
-		"users": []string{user1.Id},
-		"host":  "127.0.0.1",
-	})
-
-	system2, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
-		"name":  "system2",
-		"users": []string{user1.Id, user2.Id},
-		"host":  "127.0.0.2",
-	})
-
-	userRecords, _ := hub.CountRecords("users")
-	assert.EqualValues(t, 2, userRecords, "all users should be created")
-
-	systemRecords, _ := hub.CountRecords("systems")
-	assert.EqualValues(t, 2, systemRecords, "all systems should be created")
-
-	testAppFactory := func(t testing.TB) *pbTests.TestApp {
-		return hub.TestApp
-	}
-
-	scenarios := []beszelTests.ApiScenario{
-		// {
-		// 	Name:            "GET not implemented - returns index",
-		// 	Method:          http.MethodGet,
-		// 	URL:             "/api/beszel/user-alerts",
-		// 	ExpectedStatus:  200,
-		// 	ExpectedContent: []string{"<html ", "globalThis.BESZEL"},
-		// 	TestAppFactory:  testAppFactory,
-		// },
-		{
-			Name:            "POST no auth",
-			Method:          http.MethodPost,
-			URL:             "/api/beszel/user-alerts",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "POST no body",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  400,
-			ExpectedContent: []string{"Bad data"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "POST bad data",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  400,
-			ExpectedContent: []string{"Bad data"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"invalidField": "this should cause validation error",
-				"threshold":    "not a number",
-			}),
-		},
-		{
-			Name:   "POST malformed JSON",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  400,
-			ExpectedContent: []string{"Bad data"},
-			TestAppFactory:  testAppFactory,
-			Body:            strings.NewReader(`{"alertType": "cpu", "threshold": 80, "enabled": true,}`),
-		},
-		{
-			Name:   "POST valid alert data multiple systems",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":      "CPU",
-				"value":     69,
-				"min":       9,
-				"systems":   []string{system1.Id, system2.Id},
-				"overwrite": false,
-			}),
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				// check total alerts
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
-				// check alert has correct values
-				matchingAlerts, _ := app.CountRecords("alerts", dbx.HashExp{"name": "CPU", "user": user1.Id, "system": system1.Id, "value": 69, "min": 9})
-				assert.EqualValues(t, 1, matchingAlerts, "should have 1 alert")
-			},
-		},
-		{
-			Name:   "POST valid alert data single system",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "Memory",
-				"systems": []string{system1.Id},
-				"value":   90,
-				"min":     10,
-			}),
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				user1Alerts, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
-				assert.EqualValues(t, 3, user1Alerts, "should have 3 alerts")
-			},
-		},
-		{
-			Name:   "Overwrite: false, should not overwrite existing alert",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":      "CPU",
-				"value":     45,
-				"min":       5,
-				"systems":   []string{system1.Id},
-				"overwrite": false,
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				beszelTests.CreateRecord(app, "alerts", map[string]any{
-					"name":   "CPU",
-					"system": system1.Id,
-					"user":   user1.Id,
-					"value":  80,
-					"min":    10,
-				})
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 1, alerts, "should have 1 alert")
-				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user1.Id})
-				assert.EqualValues(t, 80, alert.Get("value"), "should have 80 as value")
-			},
-		},
-		{
-			Name:   "Overwrite: true, should overwrite existing alert",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user2Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":      "CPU",
-				"value":     45,
-				"min":       5,
-				"systems":   []string{system2.Id},
-				"overwrite": true,
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				beszelTests.CreateRecord(app, "alerts", map[string]any{
-					"name":   "CPU",
-					"system": system2.Id,
-					"user":   user2.Id,
-					"value":  80,
-					"min":    10,
-				})
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 1, alerts, "should have 1 alert")
-				alert, _ := app.FindFirstRecordByFilter("alerts", "name = 'CPU' && user = {:user}", dbx.Params{"user": user2.Id})
-				assert.EqualValues(t, 45, alert.Get("value"), "should have 45 as value")
-			},
-		},
-		{
-			Name:            "DELETE no auth",
-			Method:          http.MethodDelete,
-			URL:             "/api/beszel/user-alerts",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"systems": []string{system1.Id},
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				beszelTests.CreateRecord(app, "alerts", map[string]any{
-					"name":   "CPU",
-					"system": system1.Id,
-					"user":   user1.Id,
-					"value":  80,
-					"min":    10,
-				})
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 1, alerts, "should have 1 alert")
-			},
-		},
-		{
-			Name:   "DELETE alert",
-			Method: http.MethodDelete,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"systems": []string{system1.Id},
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				beszelTests.CreateRecord(app, "alerts", map[string]any{
-					"name":   "CPU",
-					"system": system1.Id,
-					"user":   user1.Id,
-					"value":  80,
-					"min":    10,
-				})
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				alerts, _ := app.CountRecords("alerts")
-				assert.Zero(t, alerts, "should have 0 alerts")
-			},
-		},
-		{
-			Name:   "DELETE alert multiple systems",
-			Method: http.MethodDelete,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user1Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"count\":2", "\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "Memory",
-				"systems": []string{system1.Id, system2.Id},
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				for _, systemId := range []string{system1.Id, system2.Id} {
-					_, err := beszelTests.CreateRecord(app, "alerts", map[string]any{
-						"name":   "Memory",
-						"system": systemId,
-						"user":   user1.Id,
-						"value":  90,
-						"min":    10,
-					})
-					assert.NoError(t, err, "should create alert")
-				}
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				alerts, _ := app.CountRecords("alerts")
-				assert.Zero(t, alerts, "should have 0 alerts")
-			},
-		},
-		{
-			Name:   "User 2 should not be able to delete alert of user 1",
-			Method: http.MethodDelete,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": user2Token,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"count\":1", "\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"systems": []string{system2.Id},
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.ClearCollection(t, app, "alerts")
-				for _, user := range []string{user1.Id, user2.Id} {
-					beszelTests.CreateRecord(app, "alerts", map[string]any{
-						"name":   "CPU",
-						"system": system2.Id,
-						"user":   user,
-						"value":  80,
-						"min":    10,
-					})
-				}
-				alerts, _ := app.CountRecords("alerts")
-				assert.EqualValues(t, 2, alerts, "should have 2 alerts")
-				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
-				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
-				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
-				assert.EqualValues(t, 1, user2AlertCount, "should have 1 alert")
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				user1AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user1.Id})
-				assert.EqualValues(t, 1, user1AlertCount, "should have 1 alert")
-				user2AlertCount, _ := app.CountRecords("alerts", dbx.HashExp{"user": user2.Id})
-				assert.Zero(t, user2AlertCount, "should have 0 alerts")
-			},
-		},
-	}
-
-	for _, scenario := range scenarios {
-		scenario.Test(t)
-	}
-}
-
 func TestAlertsHistory(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+
 	synctest.Test(t, func(t *testing.T) {
-		hub, user := beszelTests.GetHubWithUser(t)
 		defer hub.Cleanup()
 
 		// Create systems and alerts
@@ -496,3 +143,46 @@ func TestAlertsHistory(t *testing.T) {
 		assert.EqualValues(t, 2, totalHistoryCount, "Should have 2 total alert history records")
 	})
 }
+
+func TestSetAlertTriggered(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	hub.StartHub()
+
+	user, _ := beszelTests.CreateUser(hub, "test@example.com", "password")
+	system, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "test-system",
+		"users": []string{user.Id},
+		"host":  "127.0.0.1",
+	})
+
+	alertRecord, _ := beszelTests.CreateRecord(hub, "alerts", map[string]any{
+		"name":      "CPU",
+		"system":    system.Id,
+		"user":      user.Id,
+		"value":     80,
+		"triggered": false,
+	})
+
+	am := alerts.NewAlertManager(hub)
+
+	var alert alerts.CachedAlertData
+	alert.PopulateFromRecord(alertRecord)
+
+	// Test triggering the alert
+	err := am.SetAlertTriggered(alert, true)
+	assert.NoError(t, err)
+
+	updatedRecord, err := hub.FindRecordById("alerts", alert.Id)
+	assert.NoError(t, err)
+	assert.True(t, updatedRecord.GetBool("triggered"))
+
+	// Test un-triggering the alert
+	err = am.SetAlertTriggered(alert, false)
+	assert.NoError(t, err)
+
+	updatedRecord, err = hub.FindRecordById("alerts", alert.Id)
+	assert.NoError(t, err)
+	assert.False(t, updatedRecord.GetBool("triggered"))
+}

internal/alerts/alerts_test_helpers.go

@@ -11,10 +11,16 @@ import (
 
 func NewTestAlertManagerWithoutWorker(app hubLike) *AlertManager {
 	return &AlertManager{
-		hub: app,
+		hub:         app,
+		alertsCache: NewAlertsCache(app),
 	}
 }
 
+// GetSystemAlertsCache returns the internal system alerts cache.
+func (am *AlertManager) GetSystemAlertsCache() *AlertsCache {
+	return am.alertsCache
+}
+
 func (am *AlertManager) GetAlertManager() *AlertManager {
 	return am
 }
@@ -33,10 +39,10 @@ func (am *AlertManager) GetPendingAlertsCount() int {
 }
 
 // ProcessPendingAlerts manually processes all expired alerts (for testing)
-func (am *AlertManager) ProcessPendingAlerts() ([]*core.Record, error) {
+func (am *AlertManager) ProcessPendingAlerts() ([]CachedAlertData, error) {
 	now := time.Now()
 	var lastErr error
-	var processedAlerts []*core.Record
+	var processedAlerts []CachedAlertData
 	am.pendingAlerts.Range(func(key, value any) bool {
 		info := value.(*alertInfo)
 		if now.After(info.expireTime) {
@@ -44,7 +50,7 @@ func (am *AlertManager) ProcessPendingAlerts() ([]*core.Record, error) {
 				info.timer.Stop()
 			}
 			am.processPendingAlert(key.(string))
-			processedAlerts = append(processedAlerts, info.alertRecord)
+			processedAlerts = append(processedAlerts, info.alertData)
 		}
 		return true
 	})
@@ -85,3 +91,11 @@ func ResolveStatusAlerts(app core.App) error {
 func (am *AlertManager) RestorePendingStatusAlerts() error {
 	return am.restorePendingStatusAlerts()
 }
+
+func (am *AlertManager) SetAlertTriggered(alert CachedAlertData, triggered bool) error {
+	return am.setAlertTriggered(alert, triggered)
+}
+
+func IsInternalURL(rawURL string) (bool, error) {
+	return isInternalURL(rawURL)
+}

internal/cmd/agent/agent.go

@@ -195,6 +195,6 @@ func main() {
 	}
 
 	if err := a.Start(serverConfig); err != nil {
-		log.Fatal("Failed to start server: ", err)
+		log.Fatal("Failed to start: ", err)
 	}
 }

internal/cmd/hub/hub.go

@@ -28,8 +28,8 @@ func main() {
 	}
 
 	baseApp := getBaseApp()
-	h := hub.NewHub(baseApp)
-	if err := h.StartHub(); err != nil {
+	hub := hub.NewHub(baseApp)
+	if err := hub.StartHub(); err != nil {
 		log.Fatal(err)
 	}
 }

internal/common/common-ws.go

@@ -42,7 +42,8 @@ type AgentResponse struct {
 	SmartData   map[string]smart.SmartData `cbor:"5,keyasint,omitempty,omitzero"` // Legacy (<= 0.17)
 	ServiceInfo systemd.ServiceDetails     `cbor:"6,keyasint,omitempty,omitzero"` // Legacy (<= 0.17)
 	// Data is the generic response payload for new endpoints (0.18+)
-	Data cbor.RawMessage `cbor:"7,keyasint,omitempty,omitzero"`
+	Data          cbor.RawMessage `cbor:"7,keyasint,omitempty,omitzero"`
+	SmartComplete bool            `cbor:"8,keyasint,omitempty,omitzero"`
 }
 
 type FingerprintRequest struct {

internal/entities/container/container.go

@@ -10,10 +10,19 @@ type ApiInfo struct {
 	Status  string
 	State   string
 	Image   string
+	Health  struct {
+		Status string
+		// FailingStreak int
+	}
+	Ports []struct {
+		// PrivatePort uint16
+		PublicPort uint16
+		IP         string
+		// Type        string
+	}
 	// ImageID string
 	// Command string
 	// Created int64
-	// Ports      []Port
 	// SizeRw     int64 `json:",omitempty"`
 	// SizeRootFs int64 `json:",omitempty"`
 	// Labels     map[string]string
@@ -43,6 +52,17 @@ type HostInfo struct {
 }
 
 func (s *ApiStats) CalculateCpuPercentLinux(prevCpuContainer uint64, prevCpuSystem uint64) float64 {
+	// A counter can read lower than the stored previous value when a stats
+	// response is processed after a newer one for the same container, or when an
+	// accounting counter resets. Unsigned subtraction wraps to ~2^64 instead of
+	// going negative: on the container counter that surfaces as an absurd
+	// percentage the caller rejects, discarding the whole sample; on the system
+	// counter it inflates the divisor and silently reports near-zero CPU.
+	// Treat either direction as a new baseline.
+	if s.CPUStats.CPUUsage.TotalUsage < prevCpuContainer || s.CPUStats.SystemUsage < prevCpuSystem {
+		return 0.0
+	}
+
 	cpuDelta := s.CPUStats.CPUUsage.TotalUsage - prevCpuContainer
 	systemDelta := s.CPUStats.SystemUsage - prevCpuSystem
 
@@ -61,7 +81,11 @@ func (s *ApiStats) CalculateCpuPercentWindows(prevCpuUsage uint64, prevRead time
 	possIntervals /= 100                // Convert to number of 100ns intervals
 	possIntervals *= uint64(s.NumProcs) // Multiple by the number of processors
 
-	// Intervals used
+	// Intervals used. Same rollback guard as the Linux path: an out-of-order or
+	// reset counter would wrap the subtraction to ~2^64.
+	if s.CPUStats.CPUUsage.TotalUsage < prevCpuUsage {
+		return 0.0
+	}
 	intervalsUsed := s.CPUStats.CPUUsage.TotalUsage - prevCpuUsage
 
 	// Percentage avoiding divide-by-zero
@@ -140,6 +164,7 @@ type Stats struct {
 	Status string       `json:"-" cbor:"6,keyasint"`
 	Id     string       `json:"-" cbor:"7,keyasint"`
 	Image  string       `json:"-" cbor:"8,keyasint"`
+	Ports  string       `json:"-" cbor:"10,keyasint"`
 	// PrevCpu     [2]uint64    `json:"-"`
 	CpuSystem    uint64       `json:"-"`
 	CpuContainer uint64       `json:"-"`

internal/entities/smart/smart.go

@@ -494,7 +494,7 @@ type SmartInfoForNvme struct {
 	FirmwareVersion string           `json:"firmware_version"`
 	// NVMePCIVendor                 NVMePCIVendor                 `json:"nvme_pci_vendor"`
 	// NVMeIEEEOUIIdentifier         uint32                        `json:"nvme_ieee_oui_identifier"`
-	// NVMeTotalCapacity             uint64                        `json:"nvme_total_capacity"`
+	NVMeTotalCapacity uint64 `json:"nvme_total_capacity"`
 	// NVMeUnallocatedCapacity       uint64                        `json:"nvme_unallocated_capacity"`
 	// NVMeControllerID              uint16                        `json:"nvme_controller_id"`
 	// NVMeVersion                   VersionStringInfo             `json:"nvme_version"`
@@ -531,6 +531,13 @@ type SmartData struct {
 	Attributes      []*SmartAttribute `json:"a,omitempty" cbor:"9,keyasint,omitempty"`
 }
 
+// SmartDataResponse contains the collected data and whether every discovered
+// device was collected. Older agents omit Complete, so hubs must not prune from it.
+type SmartDataResponse struct {
+	Data     map[string]SmartData `json:"data" cbor:"0,keyasint"`
+	Complete bool                 `json:"complete" cbor:"1,keyasint,omitempty"` // Whether every discovered device was collected
+}
+
 type SmartAttribute struct {
 	ID         uint16 `json:"id,omitempty" cbor:"0,keyasint,omitempty"`
 	Name       string `json:"n" cbor:"1,keyasint"`

internal/entities/system/system.go

@@ -33,6 +33,8 @@ type Stats struct {
 	MaxNetworkSent float64             `json:"nsm,omitempty" cbor:"-"`
 	MaxNetworkRecv float64             `json:"nrm,omitempty" cbor:"-"`
 	Temperatures   map[string]float64  `json:"t,omitempty" cbor:"20,keyasint,omitempty"`
+	Fans           map[string]uint16   `json:"f,omitempty" cbor:"36,keyasint,omitempty"`
+	Batteries      map[string]uint8    `json:"bats,omitempty" cbor:"37,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"`
@@ -42,12 +44,14 @@ type Stats struct {
 	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]
+	Battery           [2]uint8             `json:"bat,omitzero" cbor:"29,keyasint,omitzero"`    // [percent, charge state]
 	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]
 	CpuBreakdown      []float64            `json:"cpub,omitempty" cbor:"33,keyasint,omitempty"` // [user, system, iowait, steal, idle]
 	CpuCoresUsage     Uint8Slice           `json:"cpus,omitempty" cbor:"34,keyasint,omitempty"` // per-core busy usage [CPU0..]
+	DiskIoStats       [6]float64           `json:"dios,omitzero" cbor:"35,keyasint,omitzero"`   // [read time %, write time %, io utilization %, r_await ms, w_await ms, weighted io %]
+	MaxDiskIoStats    [6]float64           `json:"diosm,omitzero" cbor:"-"`                     // max values for DiskIoStats
 }
 
 // Uint8Slice wraps []uint8 to customize JSON encoding while keeping CBOR efficient.
@@ -93,10 +97,12 @@ type FsStats struct {
 	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"`
-	MaxDiskReadBytes  uint64 `json:"rbm,omitempty" cbor:"-"`
-	MaxDiskWriteBytes uint64 `json:"wbm,omitempty" cbor:"-"`
+	DiskReadBytes     uint64     `json:"rb" cbor:"6,keyasint,omitempty"`
+	DiskWriteBytes    uint64     `json:"wb" cbor:"7,keyasint,omitempty"`
+	MaxDiskReadBytes  uint64     `json:"rbm,omitempty" cbor:"-"`
+	MaxDiskWriteBytes uint64     `json:"wbm,omitempty" cbor:"-"`
+	DiskIoStats       [6]float64 `json:"dios,omitzero" cbor:"8,keyasint,omitzero"` // [read time %, write time %, io utilization %, r_await ms, w_await ms, weighted io %]
+	MaxDiskIoStats    [6]float64 `json:"diosm,omitzero" cbor:"-"`                  // max values for DiskIoStats
 }
 
 type NetIoStats struct {

internal/entities/system/system_test.go

@@ -0,0 +1,37 @@
+package system
+
+import (
+	"encoding/json"
+	"testing"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestStatsBatteryTransport(t *testing.T) {
+	stats := Stats{Battery: [2]uint8{0, 1}, Batteries: map[string]uint8{"Primary": 0, "Mouse": 75}}
+
+	jsonData, err := json.Marshal(stats)
+	require.NoError(t, err)
+	var jsonPayload map[string]any
+	require.NoError(t, json.Unmarshal(jsonData, &jsonPayload))
+	assert.Equal(t, []any{float64(0), float64(1)}, jsonPayload["bat"])
+	assert.Equal(t, map[string]any{"Primary": float64(0), "Mouse": float64(75)}, jsonPayload["bats"])
+
+	cborData, err := cbor.Marshal(stats)
+	require.NoError(t, err)
+	var decoded Stats
+	require.NoError(t, cbor.Unmarshal(cborData, &decoded))
+	assert.Equal(t, stats.Battery, decoded.Battery)
+	assert.Equal(t, stats.Batteries, decoded.Batteries)
+}
+
+func TestStatsLegacyBatteryPayload(t *testing.T) {
+	data, err := json.Marshal(Stats{Battery: [2]uint8{50, 4}})
+	require.NoError(t, err)
+	var payload map[string]any
+	require.NoError(t, json.Unmarshal(data, &payload))
+	assert.Contains(t, payload, "bat")
+	assert.NotContains(t, payload, "bats")
+}

internal/ghupdate/checksum.go

@@ -0,0 +1,43 @@
+package ghupdate
+
+import (
+	"bytes"
+	"crypto/sha256"
+	"encoding/hex"
+	"fmt"
+	"io"
+	"os"
+	"strings"
+)
+
+func verifyAssetChecksum(path, digest string) error {
+	algorithm, expectedHex, ok := strings.Cut(digest, ":")
+	if !ok || algorithm == "" || expectedHex == "" {
+		return fmt.Errorf("invalid release digest %q", digest)
+	}
+	if !strings.EqualFold(algorithm, "sha256") {
+		return fmt.Errorf("unsupported release digest algorithm %q", algorithm)
+	}
+
+	expected, err := hex.DecodeString(expectedHex)
+	if err != nil || len(expected) != sha256.Size {
+		return fmt.Errorf("invalid SHA-256 release digest %q", digest)
+	}
+
+	file, err := os.Open(path)
+	if err != nil {
+		return fmt.Errorf("failed to open release for checksum verification: %w", err)
+	}
+	defer file.Close()
+
+	hash := sha256.New()
+	if _, err := io.Copy(hash, file); err != nil {
+		return fmt.Errorf("failed to calculate release checksum: %w", err)
+	}
+	actual := hash.Sum(nil)
+	if !bytes.Equal(actual, expected) {
+		return fmt.Errorf("release checksum mismatch: expected %s, got %s", expectedHex, hex.EncodeToString(actual))
+	}
+
+	return nil
+}

internal/ghupdate/checksum_test.go

@@ -0,0 +1,57 @@
+package ghupdate
+
+import (
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+)
+
+func TestVerifyAssetChecksum(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "asset")
+	if err := os.WriteFile(path, []byte("beszel release asset"), 0600); err != nil {
+		t.Fatal(err)
+	}
+
+	tests := []struct {
+		name    string
+		digest  string
+		wantErr string
+	}{
+		{
+			name:   "valid",
+			digest: "sha256:2316f86af2c3af2f0ef595ad5359cdd19b329d4829a5425460ca9ffbf92671ab",
+		},
+		{
+			name:    "mismatch",
+			digest:  "sha256:0316f86af2c3af2f0ef595ad5359cdd19b329d4829a5425460ca9ffbf92671ab",
+			wantErr: "checksum mismatch",
+		},
+		{
+			name:    "malformed",
+			digest:  "sha256:not-a-checksum",
+			wantErr: "invalid SHA-256",
+		},
+		{
+			name:    "missing",
+			wantErr: "invalid release digest",
+		},
+		{
+			name:    "unsupported algorithm",
+			digest:  "sha512:2316f86af2c3af2f0ef595ad5359cdd19b329d4829a5425460ca9ffbf92671ab",
+			wantErr: "unsupported",
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			err := verifyAssetChecksum(path, tt.digest)
+			if tt.wantErr == "" && err != nil {
+				t.Fatalf("expected checksum to verify, got %v", err)
+			}
+			if tt.wantErr != "" && (err == nil || !strings.Contains(err.Error(), tt.wantErr)) {
+				t.Fatalf("expected error containing %q, got %v", tt.wantErr, err)
+			}
+		})
+	}
+}

internal/ghupdate/extract.go

@@ -46,18 +46,23 @@ func extractTarGz(srcPath, destDir string) error {
 			return err
 		}
 
+		path, err := archivePath(destDir, header.Name)
+		if err != nil {
+			return err
+		}
+
 		if header.Typeflag == tar.TypeDir {
-			if err := os.MkdirAll(filepath.Join(destDir, header.Name), 0755); err != nil {
+			if err := os.MkdirAll(path, 0755); err != nil {
 				return err
 			}
 			continue
 		}
 
-		if err := os.MkdirAll(filepath.Dir(filepath.Join(destDir, header.Name)), 0755); err != nil {
+		if err := os.MkdirAll(filepath.Dir(path), 0755); err != nil {
 			return err
 		}
 
-		outFile, err := os.Create(filepath.Join(destDir, header.Name))
+		outFile, err := os.Create(path)
 		if err != nil {
 			return err
 		}
@@ -72,6 +77,14 @@ func extractTarGz(srcPath, destDir string) error {
 	return nil
 }
 
+// archivePath returns a path within destDir, rejecting path traversal entries.
+func archivePath(destDir, name string) (string, error) {
+	if !filepath.IsLocal(name) {
+		return "", fmt.Errorf("invalid file path: %q", name)
+	}
+	return filepath.Join(destDir, name), nil
+}
+
 // extractZip extracts the zip archive at "src" to "dest".
 //
 // Note that only dirs and regular files will be extracted.
@@ -84,9 +97,6 @@ func extractZip(src, dest string) error {
 	}
 	defer zr.Close()
 
-	// normalize dest path to check later for Zip Slip
-	dest = filepath.Clean(dest) + string(os.PathSeparator)
-
 	for _, f := range zr.File {
 		err := extractFile(f, dest)
 		if err != nil {
@@ -100,11 +110,9 @@ func extractZip(src, dest string) error {
 // extractFile extracts the provided zipFile into "basePath/zipFileName" path,
 // creating all the necessary path directories.
 func extractFile(zipFile *zip.File, basePath string) error {
-	path := filepath.Join(basePath, zipFile.Name)
-
-	// check for Zip Slip
-	if !strings.HasPrefix(path, basePath) {
-		return fmt.Errorf("invalid file path: %s", path)
+	path, err := archivePath(basePath, zipFile.Name)
+	if err != nil {
+		return err
 	}
 
 	r, err := zipFile.Open()

internal/ghupdate/ghupdate.go

@@ -110,21 +110,13 @@ func (p *updater) update() (updated bool, err error) {
 	}
 
 	var latest *release
-	var useMirror bool
 
-	// Determine the API endpoint based on UseMirror flag
-	apiURL := fmt.Sprintf("https://api.github.com/repos/%s/%s/releases/latest", p.config.Owner, p.config.Repo)
+	apiURL := getApiURL(p.config.UseMirror, p.config.Owner, p.config.Repo)
 	if p.config.UseMirror {
-		useMirror = true
-		apiURL = fmt.Sprintf("https://gh.beszel.dev/repos/%s/%s/releases/latest?api=true", p.config.Owner, p.config.Repo)
 		ColorPrint(ColorYellow, "Using mirror for update.")
 	}
 
-	latest, err = fetchLatestRelease(
-		p.config.Context,
-		p.config.HttpClient,
-		apiURL,
-	)
+	latest, err = FetchLatestRelease(p.config.Context, p.config.HttpClient, apiURL)
 	if err != nil {
 		return false, err
 	}
@@ -143,21 +135,33 @@ func (p *updater) update() (updated bool, err error) {
 		return false, err
 	}
 
-	releaseDir := filepath.Join(p.config.DataDir, ".beszel_update")
+	if err := os.MkdirAll(p.config.DataDir, 0755); err != nil {
+		return false, fmt.Errorf("failed to create update data directory: %w", err)
+	}
+	releaseDir, err := os.MkdirTemp(p.config.DataDir, ".beszel_update-")
+	if err != nil {
+		return false, fmt.Errorf("failed to create update directory: %w", err)
+	}
 	defer os.RemoveAll(releaseDir)
 
 	ColorPrintf(ColorYellow, "Downloading %s...", asset.Name)
 
 	// download the release asset
-	assetPath := filepath.Join(releaseDir, asset.Name)
-	if err := downloadFile(p.config.Context, p.config.HttpClient, asset.DownloadUrl, assetPath, useMirror); err != nil {
+	assetPath, err := archivePath(releaseDir, asset.Name)
+	if err != nil {
+		return false, err
+	}
+	if err := downloadFile(p.config.Context, p.config.HttpClient, asset.DownloadUrl, assetPath, p.config.UseMirror); err != nil {
+		return false, err
+	}
+	ColorPrint(ColorYellow, "Verifying checksum...")
+	if err := verifyAssetChecksum(assetPath, asset.Digest); err != nil {
 		return false, err
 	}
 
 	ColorPrintf(ColorYellow, "Extracting %s...", asset.Name)
 
-	extractDir := filepath.Join(releaseDir, "extracted_"+asset.Name)
-	defer os.RemoveAll(extractDir)
+	extractDir := filepath.Join(releaseDir, "extracted")
 
 	// Extract the archive (automatically detects format)
 	if err := extract(assetPath, extractDir); err != nil {
@@ -226,11 +230,11 @@ func (p *updater) update() (updated bool, err error) {
 	return true, nil
 }
 
-func fetchLatestRelease(
-	ctx context.Context,
-	client HttpClient,
-	url string,
-) (*release, error) {
+func FetchLatestRelease(ctx context.Context, client HttpClient, url string) (*release, error) {
+	if url == "" {
+		url = getApiURL(false, "henrygd", "beszel")
+	}
+
 	req, err := http.NewRequestWithContext(ctx, "GET", url, nil)
 	if err != nil {
 		return nil, err
@@ -375,3 +379,10 @@ func isGlibc() bool {
 	}
 	return false
 }
+
+func getApiURL(useMirror bool, owner, repo string) string {
+	if useMirror {
+		return fmt.Sprintf("https://gh.beszel.dev/repos/%s/%s/releases/latest?api=true", owner, repo)
+	}
+	return fmt.Sprintf("https://api.github.com/repos/%s/%s/releases/latest", owner, repo)
+}

internal/ghupdate/ghupdate_test.go

@@ -1,6 +1,9 @@
 package ghupdate
 
 import (
+	"archive/tar"
+	"compress/gzip"
+	"os"
 	"path/filepath"
 	"testing"
 )
@@ -43,3 +46,59 @@ func TestExtractFailure(t *testing.T) {
 		t.Fatal("Expected Extract to fail due to missing tar.gz file")
 	}
 }
+
+func TestArchivePath(t *testing.T) {
+	destDir := t.TempDir()
+	for _, name := range []string{
+		"",
+		"..",
+		filepath.Join("..", "file"),
+		filepath.Join("dir", "..", "..", "file"),
+		string(os.PathSeparator) + filepath.Join("tmp", "file"),
+	} {
+		if _, err := archivePath(destDir, name); err == nil {
+			t.Errorf("expected %q to be rejected", name)
+		}
+	}
+
+	name := filepath.Join("dir", "file")
+	if path, err := archivePath(destDir, name); err != nil || path != filepath.Join(destDir, name) {
+		t.Errorf("archivePath(%q) = %q, %v", name, path, err)
+	}
+}
+
+func TestExtractTarGzRejectsPathTraversal(t *testing.T) {
+	testDir := t.TempDir()
+	archivePath := filepath.Join(testDir, "malicious.tar.gz")
+	destDir := filepath.Join(testDir, "extract")
+	escapedPath := filepath.Join(testDir, "escaped")
+
+	archive, err := os.Create(archivePath)
+	if err != nil {
+		t.Fatal(err)
+	}
+	gz := gzip.NewWriter(archive)
+	tw := tar.NewWriter(gz)
+	if err := tw.WriteHeader(&tar.Header{Name: "../escaped", Mode: 0600, Size: 1}); err != nil {
+		t.Fatal(err)
+	}
+	if _, err := tw.Write([]byte("x")); err != nil {
+		t.Fatal(err)
+	}
+	if err := tw.Close(); err != nil {
+		t.Fatal(err)
+	}
+	if err := gz.Close(); err != nil {
+		t.Fatal(err)
+	}
+	if err := archive.Close(); err != nil {
+		t.Fatal(err)
+	}
+
+	if err := extract(archivePath, destDir); err == nil {
+		t.Fatal("expected path traversal archive to be rejected")
+	}
+	if _, err := os.Stat(escapedPath); !os.IsNotExist(err) {
+		t.Fatalf("path traversal wrote %s", escapedPath)
+	}
+}

internal/ghupdate/release.go

@@ -8,6 +8,7 @@ import (
 type releaseAsset struct {
 	Name        string `json:"name"`
 	DownloadUrl string `json:"browser_download_url"`
+	Digest      string `json:"digest"`
 	Id          int    `json:"id"`
 	Size        int    `json:"size"`
 }

internal/hub/agent_connect.go

@@ -317,6 +317,9 @@ func getRealIP(r *http.Request) string {
 	if ip := r.Header.Get("CF-Connecting-IP"); ip != "" {
 		return ip
 	}
+	if ip := r.Header.Get("X-Real-IP"); ip != "" {
+		return ip
+	}
 	if ip := r.Header.Get("X-Forwarded-For"); ip != "" {
 		// X-Forwarded-For can contain a comma-separated list: "client_ip, proxy1, proxy2"
 		// Take the first one

internal/hub/agent_connect_test.go

@@ -32,7 +32,7 @@ func createTestHub(t testing.TB) (*Hub, *pbtests.TestApp, error) {
 	if err != nil {
 		return nil, nil, err
 	}
-	return NewHub(testApp), testApp, nil
+	return NewHub(testApp), testApp, err
 }
 
 // cleanupTestHub stops background system goroutines before tearing down the app.
@@ -897,12 +897,8 @@ func TestAgentWebSocketIntegration(t *testing.T) {
 			require.NoError(t, err)
 
 			// Set up environment variables for the agent
-			os.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
-			os.Setenv("BESZEL_AGENT_TOKEN", tc.agentToken)
-			defer func() {
-				os.Unsetenv("BESZEL_AGENT_HUB_URL")
-				os.Unsetenv("BESZEL_AGENT_TOKEN")
-			}()
+			t.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
+			t.Setenv("BESZEL_AGENT_TOKEN", tc.agentToken)
 
 			// Start agent in background
 			done := make(chan error, 1)
@@ -1080,12 +1076,8 @@ func TestMultipleSystemsWithSameUniversalToken(t *testing.T) {
 			require.NoError(t, err)
 
 			// Set up environment variables for the agent
-			os.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
-			os.Setenv("BESZEL_AGENT_TOKEN", universalToken)
-			defer func() {
-				os.Unsetenv("BESZEL_AGENT_HUB_URL")
-				os.Unsetenv("BESZEL_AGENT_TOKEN")
-			}()
+			t.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
+			t.Setenv("BESZEL_AGENT_TOKEN", universalToken)
 
 			// Count systems before connection
 			systemsBefore, err := testApp.FindRecordsByFilter("systems", "users ~ {:userId}", "", -1, 0, map[string]any{"userId": userRecord.Id})
@@ -1243,12 +1235,8 @@ func TestPermanentUniversalTokenFromDB(t *testing.T) {
 	require.NoError(t, err)
 
 	// Set up environment variables for the agent
-	os.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
-	os.Setenv("BESZEL_AGENT_TOKEN", universalToken)
-	defer func() {
-		os.Unsetenv("BESZEL_AGENT_HUB_URL")
-		os.Unsetenv("BESZEL_AGENT_TOKEN")
-	}()
+	t.Setenv("BESZEL_AGENT_HUB_URL", ts.URL)
+	t.Setenv("BESZEL_AGENT_TOKEN", universalToken)
 
 	// Start agent in background
 	done := make(chan error, 1)
@@ -1808,6 +1796,24 @@ func TestGetRealIP(t *testing.T) {
 			remoteAddr: "127.0.0.1:12345",
 			expectedIP: "192.168.1.8",
 		},
+		{
+			name:       "X-Real-IP header",
+			headers:    map[string]string{"X-Real-IP": "10.8.0.4"},
+			remoteAddr: "172.21.0.1:12345",
+			expectedIP: "10.8.0.4",
+		},
+		{
+			name:       "X-Real-IP takes precedence over X-Forwarded-For",
+			headers:    map[string]string{"X-Real-IP": "10.8.0.4", "X-Forwarded-For": "10.8.0.5"},
+			remoteAddr: "172.21.0.1:12345",
+			expectedIP: "10.8.0.4",
+		},
+		{
+			name:       "CF-Connecting-IP takes precedence over X-Real-IP",
+			headers:    map[string]string{"CF-Connecting-IP": "1.2.3.4", "X-Real-IP": "10.8.0.4"},
+			remoteAddr: "172.21.0.1:12345",
+			expectedIP: "1.2.3.4",
+		},
 	}
 
 	for _, tc := range testCases {

internal/hub/api.go

@@ -0,0 +1,391 @@
+package hub
+
+import (
+	"context"
+	"net/http"
+	"regexp"
+	"strings"
+	"time"
+
+	"github.com/blang/semver"
+	"github.com/google/uuid"
+	"github.com/henrygd/beszel"
+	"github.com/henrygd/beszel/internal/alerts"
+	"github.com/henrygd/beszel/internal/ghupdate"
+	"github.com/henrygd/beszel/internal/hub/config"
+	"github.com/henrygd/beszel/internal/hub/systems"
+	"github.com/henrygd/beszel/internal/hub/utils"
+	"github.com/pocketbase/dbx"
+	"github.com/pocketbase/pocketbase/apis"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+// UpdateInfo holds information about the latest update check
+type UpdateInfo struct {
+	lastCheck time.Time
+	Version   string `json:"v"`
+	Url       string `json:"url"`
+}
+
+var containerIDPattern = regexp.MustCompile(`^[a-fA-F0-9]{12,64}$`)
+
+// Middleware to allow only admin role users
+var requireAdminRole = customAuthMiddleware(func(e *core.RequestEvent) bool {
+	return e.Auth.GetString("role") == "admin"
+})
+
+// Middleware to exclude readonly users
+var excludeReadOnlyRole = customAuthMiddleware(func(e *core.RequestEvent) bool {
+	return e.Auth.GetString("role") != "readonly"
+})
+
+// customAuthMiddleware handles boilerplate for custom authentication middlewares. fn should
+// return true if the request is allowed, false otherwise. e.Auth is guaranteed to be non-nil.
+func customAuthMiddleware(fn func(*core.RequestEvent) bool) func(*core.RequestEvent) error {
+	return func(e *core.RequestEvent) error {
+		if e.Auth == nil {
+			return e.UnauthorizedError("The request requires valid record authorization token.", nil)
+		}
+		if !fn(e) {
+			return e.ForbiddenError("The authorized record is not allowed to perform this action.", nil)
+		}
+		return e.Next()
+	}
+}
+
+// registerMiddlewares registers custom middlewares
+func (h *Hub) registerMiddlewares(se *core.ServeEvent) {
+	// authorizes request with user matching the provided email
+	authorizeRequestWithEmail := func(e *core.RequestEvent, email string) (err error) {
+		if e.Auth != nil || email == "" {
+			return e.Next()
+		}
+		isAuthRefresh := e.Request.URL.Path == "/api/collections/users/auth-refresh" && e.Request.Method == http.MethodPost
+		e.Auth, err = e.App.FindAuthRecordByEmail("users", email)
+		if err != nil || !isAuthRefresh {
+			return e.Next()
+		}
+		// auth refresh endpoint, make sure token is set in header
+		token, _ := e.Auth.NewAuthToken()
+		e.Request.Header.Set("Authorization", token)
+		return e.Next()
+	}
+	// authenticate with trusted header
+	if autoLogin, _ := utils.GetEnv("AUTO_LOGIN"); autoLogin != "" {
+		se.Router.BindFunc(func(e *core.RequestEvent) error {
+			return authorizeRequestWithEmail(e, autoLogin)
+		})
+	}
+	// authenticate with trusted header
+	if trustedHeader, _ := utils.GetEnv("TRUSTED_AUTH_HEADER"); trustedHeader != "" {
+		se.Router.BindFunc(func(e *core.RequestEvent) error {
+			return authorizeRequestWithEmail(e, e.Request.Header.Get(trustedHeader))
+		})
+	}
+}
+
+// registerApiRoutes registers custom API routes
+func (h *Hub) registerApiRoutes(se *core.ServeEvent) error {
+	// auth protected routes
+	apiAuth := se.Router.Group("/api/beszel")
+	apiAuth.Bind(apis.RequireAuth())
+	// auth optional routes
+	apiNoAuth := se.Router.Group("/api/beszel")
+
+	// create first user endpoint only needed if no users exist
+	if totalUsers, _ := se.App.CountRecords("users"); totalUsers == 0 {
+		apiNoAuth.POST("/create-user", h.um.CreateFirstUser)
+	}
+	// check if first time setup on login page
+	apiNoAuth.GET("/first-run", func(e *core.RequestEvent) error {
+		total, err := e.App.CountRecords("users")
+		return e.JSON(http.StatusOK, map[string]bool{"firstRun": err == nil && total == 0})
+	})
+	// get public key and version
+	apiAuth.GET("/info", h.getInfo)
+	apiAuth.GET("/getkey", h.getInfo) // deprecated - keep for compatibility w/ integrations
+	// check for updates
+	if optIn, _ := utils.GetEnv("CHECK_UPDATES"); optIn == "true" {
+		var updateInfo UpdateInfo
+		apiAuth.GET("/update", updateInfo.getUpdate)
+	}
+	// send test notification
+	apiAuth.POST("/test-notification", h.SendTestNotification)
+	// heartbeat status and test
+	apiAuth.GET("/heartbeat-status", h.getHeartbeatStatus).BindFunc(requireAdminRole)
+	apiAuth.POST("/test-heartbeat", h.testHeartbeat).BindFunc(requireAdminRole)
+	// get config.yml content
+	apiAuth.GET("/config-yaml", config.GetYamlConfig).BindFunc(requireAdminRole)
+	// handle agent websocket connection
+	apiNoAuth.GET("/agent-connect", h.handleAgentConnect)
+	// get or create universal tokens
+	apiAuth.GET("/universal-token", h.getUniversalToken).BindFunc(excludeReadOnlyRole)
+	// update / delete user alerts
+	apiAuth.POST("/user-alerts", alerts.UpsertUserAlerts)
+	apiAuth.DELETE("/user-alerts", alerts.DeleteUserAlerts)
+	// refresh SMART devices for a system
+	apiAuth.POST("/smart/refresh", h.refreshSmartData).BindFunc(excludeReadOnlyRole)
+	// get systemd service details
+	apiAuth.GET("/systemd/info", h.getSystemdInfo)
+	// /containers routes
+	if enabled, _ := utils.GetEnv("CONTAINER_DETAILS"); enabled != "false" {
+		// get container logs
+		apiAuth.GET("/containers/logs", h.getContainerLogs)
+		// get container info
+		apiAuth.GET("/containers/info", h.getContainerInfo)
+	}
+	return nil
+}
+
+// getInfo returns data needed by authenticated users, such as the public key and current version
+func (h *Hub) getInfo(e *core.RequestEvent) error {
+	type infoResponse struct {
+		Key         string `json:"key"`
+		Version     string `json:"v"`
+		CheckUpdate bool   `json:"cu"`
+	}
+	info := infoResponse{
+		Key:     h.pubKey,
+		Version: beszel.Version,
+	}
+	if optIn, _ := utils.GetEnv("CHECK_UPDATES"); optIn == "true" {
+		info.CheckUpdate = true
+	}
+	return e.JSON(http.StatusOK, info)
+}
+
+// getUpdate checks for the latest release on GitHub and returns update info if a newer version is available
+func (info *UpdateInfo) getUpdate(e *core.RequestEvent) error {
+	if time.Since(info.lastCheck) < 6*time.Hour {
+		return e.JSON(http.StatusOK, info)
+	}
+	info.lastCheck = time.Now()
+	latestRelease, err := ghupdate.FetchLatestRelease(context.Background(), http.DefaultClient, "")
+	if err != nil {
+		return err
+	}
+	currentVersion, err := semver.Parse(strings.TrimPrefix(beszel.Version, "v"))
+	if err != nil {
+		return err
+	}
+	latestVersion, err := semver.Parse(strings.TrimPrefix(latestRelease.Tag, "v"))
+	if err != nil {
+		return err
+	}
+	if latestVersion.GT(currentVersion) {
+		info.Version = strings.TrimPrefix(latestRelease.Tag, "v")
+		info.Url = latestRelease.Url
+	}
+	return e.JSON(http.StatusOK, info)
+}
+
+// GetUniversalToken handles the universal token API endpoint (create, read, delete)
+func (h *Hub) getUniversalToken(e *core.RequestEvent) error {
+	if e.Auth.IsSuperuser() {
+		return e.ForbiddenError("Superusers cannot use universal tokens", nil)
+	}
+
+	tokenMap := universalTokenMap.GetMap()
+	userID := e.Auth.Id
+	query := e.Request.URL.Query()
+	token := query.Get("token")
+	enable := query.Get("enable")
+	permanent := query.Get("permanent")
+
+	// helper for deleting any existing permanent token record for this user
+	deletePermanent := func() error {
+		rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID})
+		if err != nil {
+			return nil // no record
+		}
+		return h.Delete(rec)
+	}
+
+	// helper for upserting a permanent token record for this user
+	upsertPermanent := func(token string) error {
+		rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID})
+		if err == nil {
+			rec.Set("token", token)
+			return h.Save(rec)
+		}
+
+		col, err := h.FindCachedCollectionByNameOrId("universal_tokens")
+		if err != nil {
+			return err
+		}
+		newRec := core.NewRecord(col)
+		newRec.Set("user", userID)
+		newRec.Set("token", token)
+		return h.Save(newRec)
+	}
+
+	// Disable universal tokens (both ephemeral and permanent)
+	if enable == "0" {
+		tokenMap.RemovebyValue(userID)
+		_ = deletePermanent()
+		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": false, "permanent": false})
+	}
+
+	// Enable universal token (ephemeral or permanent)
+	if enable == "1" {
+		if token == "" {
+			token = uuid.New().String()
+		}
+
+		if permanent == "1" {
+			// make token permanent (persist across restarts)
+			tokenMap.RemovebyValue(userID)
+			if err := upsertPermanent(token); err != nil {
+				return err
+			}
+			return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": true})
+		}
+
+		// default: ephemeral mode (1 hour)
+		_ = deletePermanent()
+		tokenMap.Set(token, userID, time.Hour)
+		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": false})
+	}
+
+	// Read current state
+	// Prefer permanent token if it exists.
+	if rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID}); err == nil {
+		dbToken := rec.GetString("token")
+		// If no token was provided, or the caller is asking about their permanent token, return it.
+		if token == "" || token == dbToken {
+			return e.JSON(http.StatusOK, map[string]any{"token": dbToken, "active": true, "permanent": true})
+		}
+		// Token doesn't match their permanent token (avoid leaking other info)
+		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": false, "permanent": false})
+	}
+
+	// No permanent token; fall back to ephemeral token map.
+	if token == "" {
+		// return existing token if it exists
+		if token, _, ok := tokenMap.GetByValue(userID); ok {
+			return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": false})
+		}
+		// if no token is provided, generate a new one
+		token = uuid.New().String()
+	}
+
+	// Token is considered active only if it belongs to the current user.
+	activeUser, ok := tokenMap.GetOk(token)
+	active := ok && activeUser == userID
+	response := map[string]any{"token": token, "active": active, "permanent": false}
+	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 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 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")
+	containerID := e.Request.URL.Query().Get("container")
+
+	if systemID == "" || containerID == "" || !containerIDPattern.MatchString(containerID) {
+		return e.BadRequestError("Invalid system or container parameter", nil)
+	}
+
+	system, err := h.sm.GetSystem(systemID)
+	if err != nil || !system.HasUser(e.App, e.Auth) {
+		return e.NotFoundError("", nil)
+	}
+
+	data, err := fetchFunc(system, containerID)
+	if err != nil {
+		return e.InternalServerError("", err)
+	}
+
+	return e.JSON(http.StatusOK, map[string]string{responseKey: data})
+}
+
+// getContainerLogs handles GET /api/beszel/containers/logs requests
+func (h *Hub) getContainerLogs(e *core.RequestEvent) error {
+	return h.containerRequestHandler(e, func(system *systems.System, containerID string) (string, error) {
+		return system.FetchContainerLogsFromAgent(containerID)
+	}, "logs")
+}
+
+func (h *Hub) getContainerInfo(e *core.RequestEvent) error {
+	return h.containerRequestHandler(e, func(system *systems.System, containerID string) (string, error) {
+		return system.FetchContainerInfoFromAgent(containerID)
+	}, "info")
+}
+
+// getSystemdInfo handles GET /api/beszel/systemd/info requests
+func (h *Hub) getSystemdInfo(e *core.RequestEvent) error {
+	query := e.Request.URL.Query()
+	systemID := query.Get("system")
+	serviceName := query.Get("service")
+
+	if systemID == "" || serviceName == "" {
+		return e.BadRequestError("Invalid system or service parameter", nil)
+	}
+	system, err := h.sm.GetSystem(systemID)
+	if err != nil || !system.HasUser(e.App, e.Auth) {
+		return e.NotFoundError("", nil)
+	}
+	// verify service exists before fetching details
+	_, err = e.App.FindFirstRecordByFilter("systemd_services", "system = {:system} && name = {:name}", dbx.Params{
+		"system": systemID,
+		"name":   serviceName,
+	})
+	if err != nil {
+		return e.NotFoundError("", err)
+	}
+	details, err := system.FetchSystemdInfoFromAgent(serviceName)
+	if err != nil {
+		return e.InternalServerError("", err)
+	}
+	e.Response.Header().Set("Cache-Control", "public, max-age=60")
+	return e.JSON(http.StatusOK, map[string]any{"details": details})
+}
+
+// refreshSmartData handles POST /api/beszel/smart/refresh requests
+// Fetches fresh SMART data from the agent and updates the collection
+func (h *Hub) refreshSmartData(e *core.RequestEvent) error {
+	systemID := e.Request.URL.Query().Get("system")
+	if systemID == "" {
+		return e.BadRequestError("Invalid system parameter", nil)
+	}
+
+	system, err := h.sm.GetSystem(systemID)
+	if err != nil || !system.HasUser(e.App, e.Auth) {
+		return e.NotFoundError("", nil)
+	}
+
+	if err := system.FetchAndSaveSmartDevices(); err != nil {
+		return e.InternalServerError("", err)
+	}
+
+	return e.JSON(http.StatusOK, map[string]string{"status": "ok"})
+}

internal/hub/api_test.go

@@ -0,0 +1,972 @@
+package hub_test
+
+import (
+	"bytes"
+	"encoding/json"
+	"fmt"
+	"io"
+	"net/http"
+	"testing"
+
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+
+	"github.com/henrygd/beszel/internal/migrations"
+	"github.com/pocketbase/pocketbase/core"
+	pbTests "github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/require"
+)
+
+// marshal to json and return an io.Reader (for use in ApiScenario.Body)
+func jsonReader(v any) io.Reader {
+	data, err := json.Marshal(v)
+	if err != nil {
+		panic(err)
+	}
+	return bytes.NewReader(data)
+}
+
+func TestApiRoutesAuthentication(t *testing.T) {
+	hub, user := beszelTests.GetHubWithUser(t)
+	defer hub.Cleanup()
+
+	userToken, err := user.NewAuthToken()
+	require.NoError(t, err, "Failed to create auth token")
+
+	// Create test user and get auth token
+	user2, err := beszelTests.CreateUser(hub, "testuser@example.com", "password123")
+	require.NoError(t, err, "Failed to create test user")
+	user2Token, err := user2.NewAuthToken()
+	require.NoError(t, err, "Failed to create user2 auth token")
+
+	adminUser, err := beszelTests.CreateUserWithRole(hub, "admin@example.com", "password123", "admin")
+	require.NoError(t, err, "Failed to create admin user")
+	adminUserToken, err := adminUser.NewAuthToken()
+
+	readOnlyUser, err := beszelTests.CreateUserWithRole(hub, "readonly@example.com", "password123", "readonly")
+	require.NoError(t, err, "Failed to create readonly user")
+	readOnlyUserToken, err := readOnlyUser.NewAuthToken()
+	require.NoError(t, err, "Failed to create readonly user auth token")
+
+	superuser, err := beszelTests.CreateSuperuser(hub, "superuser@example.com", "password123")
+	require.NoError(t, err, "Failed to create superuser")
+	superuserToken, err := superuser.NewAuthToken()
+	require.NoError(t, err, "Failed to create superuser auth token")
+
+	// Create test system
+	system, err := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "test-system",
+		"users": []string{user.Id},
+		"host":  "127.0.0.1",
+	})
+	require.NoError(t, err, "Failed to create test system")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		// Auth Protected Routes - Should require authentication
+		{
+			Name:            "GET /config-yaml - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/config-yaml",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /config-yaml - with user auth should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/config-yaml",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"The authorized record is not allowed to perform this action."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /config-yaml - with admin auth should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/config-yaml",
+			Headers: map[string]string{
+				"Authorization": adminUserToken,
+			},
+			ExpectedStatus:  200,
+			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{"The authorized record is not allowed to perform this action."},
+			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{"The authorized record is not allowed to perform this action."},
+			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,
+			URL:             "/api/beszel/universal-token",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /universal-token - with auth should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/universal-token",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"active", "token", "permanent"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /universal-token - enable permanent should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/universal-token?enable=1&permanent=1&token=permanent-token-123",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"permanent\":true", "permanent-token-123"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /universal-token - superuser should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/universal-token",
+			Headers: map[string]string{
+				"Authorization": superuserToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"Superusers cannot use universal tokens"},
+			TestAppFactory: func(t testing.TB) *pbTests.TestApp {
+				return hub.TestApp
+			},
+		},
+		{
+			Name:   "GET /universal-token - with readonly auth should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/universal-token",
+			Headers: map[string]string{
+				"Authorization": readOnlyUserToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"The authorized record is not allowed to perform this action."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /smart/refresh - missing system should fail 400 with user auth",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/smart/refresh",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Invalid", "system", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /smart/refresh - with readonly auth should fail",
+			Method: http.MethodPost,
+			URL:    fmt.Sprintf("/api/beszel/smart/refresh?system=%s", system.Id),
+			Headers: map[string]string{
+				"Authorization": readOnlyUserToken,
+			},
+			ExpectedStatus:  403,
+			ExpectedContent: []string{"The authorized record is not allowed to perform this action."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /smart/refresh - non-user system should fail",
+			Method: http.MethodPost,
+			URL:    fmt.Sprintf("/api/beszel/smart/refresh?system=%s", system.Id),
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /smart/refresh - good user should pass validation",
+			Method: http.MethodPost,
+			URL:    fmt.Sprintf("/api/beszel/smart/refresh?system=%s", system.Id),
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  500,
+			ExpectedContent: []string{"Something went wrong while processing your request."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "POST /user-alerts - no auth should fail",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:   "POST /user-alerts - with auth should succeed",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:            "DELETE /user-alerts - no auth should fail",
+			Method:          http.MethodDelete,
+			URL:             "/api/beszel/user-alerts",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system.Id},
+			}),
+		},
+		{
+			Name:   "DELETE /user-alerts - with auth should succeed",
+			Method: http.MethodDelete,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"success\":true"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"systems": []string{system.Id},
+			}),
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				// Create an alert to delete
+				beszelTests.CreateRecord(app, "alerts", map[string]any{
+					"name":   "CPU",
+					"system": system.Id,
+					"user":   user.Id,
+					"value":  80,
+					"min":    10,
+				})
+			},
+		},
+		{
+			Name:            "GET /containers/logs - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/containers/logs?system=test-system&container=abababababab",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /containers/logs - request for valid non-user system should fail",
+			Method:          http.MethodGet,
+			URL:             fmt.Sprintf("/api/beszel/containers/logs?system=%s&container=abababababab", system.Id),
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+		},
+		{
+			Name:            "GET /containers/info - request for valid non-user system should fail",
+			Method:          http.MethodGet,
+			URL:             fmt.Sprintf("/api/beszel/containers/info?system=%s&container=abababababab", system.Id),
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+		},
+		{
+			Name:            "GET /containers/info - SHARE_ALL_SYSTEMS allows non-member user",
+			Method:          http.MethodGet,
+			URL:             fmt.Sprintf("/api/beszel/containers/info?system=%s&container=abababababab", system.Id),
+			ExpectedStatus:  500,
+			ExpectedContent: []string{"Something went wrong while processing your request."},
+			TestAppFactory:  testAppFactory,
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "true")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "")
+			},
+		},
+		{
+			Name:   "GET /containers/logs - with auth but missing system param should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/logs?container=abababababab",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Invalid", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/logs - with auth but missing container param should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/logs?system=test-system",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Invalid", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/logs - with auth but invalid system should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/logs?system=invalid-system&container=0123456789ab",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't 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", "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", "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", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/logs - good user should pass validation",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/logs?system=" + system.Id + "&container=0123456789ab",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  500,
+			ExpectedContent: []string{"Something went wrong while processing your request."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /containers/info - good user should pass validation",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/containers/info?system=" + system.Id + "&container=0123456789ab",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  500,
+			ExpectedContent: []string{"Something went wrong while processing your request."},
+			TestAppFactory:  testAppFactory,
+		},
+		// /systemd routes
+		{
+			Name:            "GET /systemd/info - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             fmt.Sprintf("/api/beszel/systemd/info?system=%s&service=nginx.service", system.Id),
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /systemd/info - request for valid non-user system should fail",
+			Method:          http.MethodGet,
+			URL:             fmt.Sprintf("/api/beszel/systemd/info?system=%s&service=nginx.service", system.Id),
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+			Headers: map[string]string{
+				"Authorization": user2Token,
+			},
+		},
+		{
+			Name:   "GET /systemd/info - with auth but missing system param should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/systemd/info?service=nginx.service",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Invalid", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /systemd/info - with auth but missing service param should fail",
+			Method: http.MethodGet,
+			URL:    fmt.Sprintf("/api/beszel/systemd/info?system=%s", system.Id),
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  400,
+			ExpectedContent: []string{"Invalid", "parameter"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /systemd/info - with auth but invalid system should fail",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/systemd/info?system=invalid-system&service=nginx.service",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /systemd/info - service not in systemd_services collection should fail",
+			Method: http.MethodGet,
+			URL:    fmt.Sprintf("/api/beszel/systemd/info?system=%s&service=notregistered.service", system.Id),
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"The requested resource wasn't found."},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /systemd/info - with auth and existing service record should pass validation",
+			Method: http.MethodGet,
+			URL:    fmt.Sprintf("/api/beszel/systemd/info?system=%s&service=nginx.service", system.Id),
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  500,
+			ExpectedContent: []string{"Something went wrong while processing your request."},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.CreateRecord(app, "systemd_services", map[string]any{
+					"system": system.Id,
+					"name":   "nginx.service",
+					"state":  0,
+					"sub":    1,
+				})
+			},
+		},
+
+		// Auth Optional Routes - Should work without authentication
+		{
+			Name:            "GET /getkey - no auth should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /getkey - with auth should also succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/getkey",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"key\":", "\"v\":"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /info - should return the same as /getkey",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/info",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"key\":", "\"v\":"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /first-run - no auth should succeed",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/first-run",
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"firstRun\":false"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /first-run - with auth should also succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/first-run",
+			Headers: map[string]string{
+				"Authorization": userToken,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"firstRun\":false"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /agent-connect - no auth should succeed (websocket upgrade fails but route is accessible)",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/agent-connect",
+			ExpectedStatus:  400,
+			ExpectedContent: []string{},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /test-notification - invalid auth token should fail",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/test-notification",
+			Body: jsonReader(map[string]any{
+				"url": "generic://127.0.0.1",
+			}),
+			Headers: map[string]string{
+				"Authorization": "invalid-token",
+			},
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "POST /user-alerts - invalid auth token should fail",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/user-alerts",
+			Headers: map[string]string{
+				"Authorization": "invalid-token",
+			},
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+			Body: jsonReader(map[string]any{
+				"name":    "CPU",
+				"value":   80,
+				"min":     10,
+				"systems": []string{system.Id},
+			}),
+		},
+		// this works but diff behavior on prod vs dev.
+		// dev returns 502; prod returns 200 with static html page 404
+		// TODO: align dev and prod behavior and re-enable this test
+		// {
+		// 	Name:               "GET /update - shouldn't exist without CHECK_UPDATES env var",
+		// 	Method:             http.MethodGet,
+		// 	URL:                "/api/beszel/update",
+		// 	NotExpectedContent: []string{"v:", "\"v\":"},
+		// 	ExpectedStatus: 502,
+		// 	TestAppFactory: testAppFactory,
+		// },
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}
+
+func TestFirstUserCreation(t *testing.T) {
+	t.Run("CreateUserEndpoint available when no users exist", func(t *testing.T) {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		hub.StartHub()
+
+		testAppFactoryExisting := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenarios := []beszelTests.ApiScenario{
+			{
+				Name:   "POST /create-user - should be available when no users exist",
+				Method: http.MethodPost,
+				URL:    "/api/beszel/create-user",
+				Body: jsonReader(map[string]any{
+					"email":    "firstuser@example.com",
+					"password": "password123",
+				}),
+				ExpectedStatus:  200,
+				ExpectedContent: []string{"User created"},
+				TestAppFactory:  testAppFactoryExisting,
+				BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+					userCount, err := hub.CountRecords("users")
+					require.NoError(t, err)
+					require.Zero(t, userCount, "Should start with no users")
+					superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
+					require.NoError(t, err)
+					require.EqualValues(t, 1, len(superusers), "Should start with one temporary superuser")
+					require.EqualValues(t, migrations.TempAdminEmail, superusers[0].GetString("email"), "Should have created one temporary superuser")
+				},
+				AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+					userCount, err := hub.CountRecords("users")
+					require.NoError(t, err)
+					require.EqualValues(t, 1, userCount, "Should have created one user")
+					superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
+					require.NoError(t, err)
+					require.EqualValues(t, 1, len(superusers), "Should have created one superuser")
+					require.EqualValues(t, "firstuser@example.com", superusers[0].GetString("email"), "Should have created one superuser")
+				},
+			},
+			{
+				Name:   "POST /create-user - should not be available when users exist",
+				Method: http.MethodPost,
+				URL:    "/api/beszel/create-user",
+				Body: jsonReader(map[string]any{
+					"email":    "firstuser@example.com",
+					"password": "password123",
+				}),
+				ExpectedStatus:  404,
+				ExpectedContent: []string{"wasn't found"},
+				TestAppFactory:  testAppFactoryExisting,
+			},
+		}
+
+		for _, scenario := range scenarios {
+			scenario.Test(t)
+		}
+	})
+
+	t.Run("CreateUserEndpoint not available when USER_EMAIL, USER_PASSWORD are set", func(t *testing.T) {
+		t.Setenv("BESZEL_HUB_USER_EMAIL", "me@example.com")
+		t.Setenv("BESZEL_HUB_USER_PASSWORD", "password123")
+
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		hub.StartHub()
+
+		testAppFactory := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenario := beszelTests.ApiScenario{
+			Name:            "POST /create-user - should not be available when USER_EMAIL, USER_PASSWORD are set",
+			Method:          http.MethodPost,
+			URL:             "/api/beszel/create-user",
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"wasn't found"},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				users, err := hub.FindAllRecords("users")
+				require.NoError(t, err)
+				require.EqualValues(t, 1, len(users), "Should start with one user")
+				require.EqualValues(t, "me@example.com", users[0].GetString("email"), "Should have created one user")
+				superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
+				require.NoError(t, err)
+				require.EqualValues(t, 1, len(superusers), "Should start with one superuser")
+				require.EqualValues(t, "me@example.com", superusers[0].GetString("email"), "Should have created one superuser")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				users, err := hub.FindAllRecords("users")
+				require.NoError(t, err)
+				require.EqualValues(t, 1, len(users), "Should still have one user")
+				require.EqualValues(t, "me@example.com", users[0].GetString("email"), "Should have created one user")
+				superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
+				require.NoError(t, err)
+				require.EqualValues(t, 1, len(superusers), "Should still have one superuser")
+				require.EqualValues(t, "me@example.com", superusers[0].GetString("email"), "Should have created one superuser")
+			},
+		}
+
+		scenario.Test(t)
+	})
+}
+
+func TestCreateUserEndpointAvailability(t *testing.T) {
+	t.Run("CreateUserEndpoint available when no users exist", func(t *testing.T) {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		// Ensure no users exist
+		userCount, err := hub.CountRecords("users")
+		require.NoError(t, err)
+		require.Zero(t, userCount, "Should start with no users")
+
+		hub.StartHub()
+
+		testAppFactory := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenario := beszelTests.ApiScenario{
+			Name:   "POST /create-user - should be available when no users exist",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/create-user",
+			Body: jsonReader(map[string]any{
+				"email":    "firstuser@example.com",
+				"password": "password123",
+			}),
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"User created"},
+			TestAppFactory:  testAppFactory,
+		}
+
+		scenario.Test(t)
+
+		// Verify user was created
+		userCount, err = hub.CountRecords("users")
+		require.NoError(t, err)
+		require.EqualValues(t, 1, userCount, "Should have created one user")
+	})
+
+	t.Run("CreateUserEndpoint not available when users exist", func(t *testing.T) {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		defer hub.Cleanup()
+
+		// Create a user first
+		_, err := beszelTests.CreateUser(hub, "existing@example.com", "password")
+		require.NoError(t, err)
+
+		hub.StartHub()
+
+		testAppFactory := func(t testing.TB) *pbTests.TestApp {
+			return hub.TestApp
+		}
+
+		scenario := beszelTests.ApiScenario{
+			Name:   "POST /create-user - should not be available when users exist",
+			Method: http.MethodPost,
+			URL:    "/api/beszel/create-user",
+			Body: jsonReader(map[string]any{
+				"email":    "another@example.com",
+				"password": "password123",
+			}),
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"wasn't found"},
+			TestAppFactory:  testAppFactory,
+		}
+
+		scenario.Test(t)
+	})
+}
+
+func TestAutoLoginMiddleware(t *testing.T) {
+	var hubs []*beszelTests.TestHub
+
+	defer func() {
+		for _, hub := range hubs {
+			hub.Cleanup()
+		}
+	}()
+
+	t.Setenv("AUTO_LOGIN", "user@test.com")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		hubs = append(hubs, hub)
+		hub.StartHub()
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:            "GET /getkey - without auto login should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /getkey - with auto login should fail if no matching user",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:            "GET /getkey - with auto login should succeed",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"key\":", "\"v\":"},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.CreateUser(app, "user@test.com", "password123")
+			},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}
+
+func TestTrustedHeaderMiddleware(t *testing.T) {
+	var hubs []*beszelTests.TestHub
+
+	defer func() {
+		for _, hub := range hubs {
+			hub.Cleanup()
+		}
+	}()
+
+	t.Setenv("TRUSTED_AUTH_HEADER", "X-Beszel-Trusted")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		hub, _ := beszelTests.NewTestHub(t.TempDir())
+		hubs = append(hubs, hub)
+		hub.StartHub()
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:            "GET /getkey - without trusted header should fail",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/getkey",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /getkey - with trusted header should fail if no matching user",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/getkey",
+			Headers: map[string]string{
+				"X-Beszel-Trusted": "user@test.com",
+			},
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		{
+			Name:   "GET /getkey - with trusted header should succeed",
+			Method: http.MethodGet,
+			URL:    "/api/beszel/getkey",
+			Headers: map[string]string{
+				"X-Beszel-Trusted": "user@test.com",
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{"\"key\":", "\"v\":"},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				beszelTests.CreateUser(app, "user@test.com", "password123")
+			},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}
+
+func TestUpdateEndpoint(t *testing.T) {
+	t.Setenv("CHECK_UPDATES", "true")
+
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+	hub.StartHub()
+
+	// Create test user and get auth token
+	// user, err := beszelTests.CreateUser(hub, "testuser@example.com", "password123")
+	// require.NoError(t, err, "Failed to create test user")
+	// userToken, err := user.NewAuthToken()
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:            "update endpoint shouldn't work without auth",
+			Method:          http.MethodGet,
+			URL:             "/api/beszel/update",
+			ExpectedStatus:  401,
+			ExpectedContent: []string{"requires valid"},
+			TestAppFactory:  testAppFactory,
+		},
+		// leave this out for now since it actually makes a request to github
+		// {
+		// 	Name:   "GET /update - with valid auth should succeed",
+		// 	Method: http.MethodGet,
+		// 	URL:    "/api/beszel/update",
+		// 	Headers: map[string]string{
+		// 		"Authorization": userToken,
+		// 	},
+		// 	ExpectedStatus:  200,
+		// 	ExpectedContent: []string{`"v":`},
+		// 	TestAppFactory:  testAppFactory,
+		// },
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}

internal/hub/collections.go

@@ -0,0 +1,131 @@
+package hub
+
+import (
+	"github.com/henrygd/beszel/internal/hub/utils"
+	"github.com/pocketbase/pocketbase/core"
+)
+
+type collectionRules struct {
+	list   *string
+	view   *string
+	create *string
+	update *string
+	delete *string
+}
+
+// setCollectionAuthSettings applies Beszel's collection auth settings.
+func setCollectionAuthSettings(app core.App) error {
+	usersCollection, err := app.FindCollectionByNameOrId("users")
+	if err != nil {
+		return err
+	}
+	superusersCollection, err := app.FindCollectionByNameOrId(core.CollectionNameSuperusers)
+	if err != nil {
+		return err
+	}
+
+	// disable email auth if DISABLE_PASSWORD_AUTH env var is set
+	disablePasswordAuth, _ := utils.GetEnv("DISABLE_PASSWORD_AUTH")
+	usersCollection.PasswordAuth.Enabled = disablePasswordAuth != "true"
+	usersCollection.PasswordAuth.IdentityFields = []string{"email"}
+	// allow oauth user creation if USER_CREATION is set
+	if userCreation, _ := utils.GetEnv("USER_CREATION"); userCreation == "true" {
+		cr := "@request.context = 'oauth2'"
+		usersCollection.CreateRule = &cr
+	} else {
+		usersCollection.CreateRule = nil
+	}
+
+	// enable mfaOtp mfa if MFA_OTP env var is set
+	mfaOtp, _ := utils.GetEnv("MFA_OTP")
+	usersCollection.OTP.Length = 6
+	superusersCollection.OTP.Length = 6
+	usersCollection.OTP.Enabled = mfaOtp == "true"
+	usersCollection.MFA.Enabled = mfaOtp == "true"
+	superusersCollection.OTP.Enabled = mfaOtp == "true" || mfaOtp == "superusers"
+	superusersCollection.MFA.Enabled = mfaOtp == "true" || mfaOtp == "superusers"
+	if err := app.Save(superusersCollection); err != nil {
+		return err
+	}
+	if err := app.Save(usersCollection); err != nil {
+		return err
+	}
+
+	// When SHARE_ALL_SYSTEMS is enabled, any authenticated user can read
+	// system-scoped data. Write rules continue to block readonly users.
+	shareAllSystems, _ := utils.GetEnv("SHARE_ALL_SYSTEMS")
+
+	authenticatedRule := "@request.auth.id != \"\""
+	systemsMemberRule := authenticatedRule + " && users.id ?= @request.auth.id"
+	systemMemberRule := authenticatedRule + " && system.users.id ?= @request.auth.id"
+
+	systemsReadRule := systemsMemberRule
+	systemScopedReadRule := systemMemberRule
+	if shareAllSystems == "true" {
+		systemsReadRule = authenticatedRule
+		systemScopedReadRule = authenticatedRule
+	}
+	systemsWriteRule := systemsReadRule + " && @request.auth.role != \"readonly\""
+	systemScopedWriteRule := systemScopedReadRule + " && @request.auth.role != \"readonly\""
+
+	if err := applyCollectionRules(app, []string{"systems"}, collectionRules{
+		list:   &systemsReadRule,
+		view:   &systemsReadRule,
+		create: &systemsWriteRule,
+		update: &systemsWriteRule,
+		delete: &systemsWriteRule,
+	}); err != nil {
+		return err
+	}
+
+	if err := applyCollectionRules(app, []string{"containers", "container_stats", "system_stats", "systemd_services"}, collectionRules{
+		list: &systemScopedReadRule,
+	}); err != nil {
+		return err
+	}
+
+	if err := applyCollectionRules(app, []string{"smart_devices"}, collectionRules{
+		list:   &systemScopedReadRule,
+		view:   &systemScopedReadRule,
+		delete: &systemScopedWriteRule,
+	}); err != nil {
+		return err
+	}
+
+	if err := applyCollectionRules(app, []string{"fingerprints"}, collectionRules{
+		list:   &systemScopedReadRule,
+		view:   &systemScopedReadRule,
+		create: &systemScopedWriteRule,
+		update: &systemScopedWriteRule,
+		delete: &systemScopedWriteRule,
+	}); err != nil {
+		return err
+	}
+
+	if err := applyCollectionRules(app, []string{"system_details"}, collectionRules{
+		list: &systemScopedReadRule,
+		view: &systemScopedReadRule,
+	}); err != nil {
+		return err
+	}
+
+	return nil
+}
+
+func applyCollectionRules(app core.App, collectionNames []string, rules collectionRules) error {
+	for _, collectionName := range collectionNames {
+		collection, err := app.FindCollectionByNameOrId(collectionName)
+		if err != nil {
+			return err
+		}
+		collection.ListRule = rules.list
+		collection.ViewRule = rules.view
+		collection.CreateRule = rules.create
+		collection.UpdateRule = rules.update
+		collection.DeleteRule = rules.delete
+		if err := app.Save(collection); err != nil {
+			return err
+		}
+	}
+	return nil
+}

internal/hub/collections_test.go

@@ -0,0 +1,527 @@
+//go:build testing
+
+package hub_test
+
+import (
+	"fmt"
+	"net/http"
+	"testing"
+
+	beszelTests "github.com/henrygd/beszel/internal/tests"
+	"github.com/pocketbase/pocketbase/core"
+	pbTests "github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+)
+
+func TestCollectionRulesDefault(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	const isUserMatchesUser = `@request.auth.id != "" && user = @request.auth.id`
+
+	const isUserInUsers = `@request.auth.id != "" && users.id ?= @request.auth.id`
+	const isUserInUsersNotReadonly = `@request.auth.id != "" && users.id ?= @request.auth.id && @request.auth.role != "readonly"`
+
+	const isUserInSystemUsers = `@request.auth.id != "" && system.users.id ?= @request.auth.id`
+	const isUserInSystemUsersNotReadonly = `@request.auth.id != "" && system.users.id ?= @request.auth.id && @request.auth.role != "readonly"`
+
+	// users collection
+	usersCollection, err := hub.FindCollectionByNameOrId("users")
+	assert.NoError(t, err, "Failed to find users collection")
+	assert.True(t, usersCollection.PasswordAuth.Enabled)
+	assert.Equal(t, usersCollection.PasswordAuth.IdentityFields, []string{"email"})
+	assert.Nil(t, usersCollection.CreateRule)
+	assert.False(t, usersCollection.MFA.Enabled)
+
+	// superusers collection
+	superusersCollection, err := hub.FindCollectionByNameOrId(core.CollectionNameSuperusers)
+	assert.NoError(t, err, "Failed to find superusers collection")
+	assert.True(t, superusersCollection.PasswordAuth.Enabled)
+	assert.Equal(t, superusersCollection.PasswordAuth.IdentityFields, []string{"email"})
+	assert.Nil(t, superusersCollection.CreateRule)
+	assert.False(t, superusersCollection.MFA.Enabled)
+
+	// alerts collection
+	alertsCollection, err := hub.FindCollectionByNameOrId("alerts")
+	require.NoError(t, err, "Failed to find alerts collection")
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.ListRule)
+	assert.Nil(t, alertsCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.DeleteRule)
+
+	// alerts_history collection
+	alertsHistoryCollection, err := hub.FindCollectionByNameOrId("alerts_history")
+	require.NoError(t, err, "Failed to find alerts_history collection")
+	assert.Equal(t, isUserMatchesUser, *alertsHistoryCollection.ListRule)
+	assert.Nil(t, alertsHistoryCollection.ViewRule)
+	assert.Nil(t, alertsHistoryCollection.CreateRule)
+	assert.Nil(t, alertsHistoryCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsHistoryCollection.DeleteRule)
+
+	// containers collection
+	containersCollection, err := hub.FindCollectionByNameOrId("containers")
+	require.NoError(t, err, "Failed to find containers collection")
+	assert.Equal(t, isUserInSystemUsers, *containersCollection.ListRule)
+	assert.Nil(t, containersCollection.ViewRule)
+	assert.Nil(t, containersCollection.CreateRule)
+	assert.Nil(t, containersCollection.UpdateRule)
+	assert.Nil(t, containersCollection.DeleteRule)
+
+	// container_stats collection
+	containerStatsCollection, err := hub.FindCollectionByNameOrId("container_stats")
+	require.NoError(t, err, "Failed to find container_stats collection")
+	assert.Equal(t, isUserInSystemUsers, *containerStatsCollection.ListRule)
+	assert.Nil(t, containerStatsCollection.ViewRule)
+	assert.Nil(t, containerStatsCollection.CreateRule)
+	assert.Nil(t, containerStatsCollection.UpdateRule)
+	assert.Nil(t, containerStatsCollection.DeleteRule)
+
+	// fingerprints collection
+	fingerprintsCollection, err := hub.FindCollectionByNameOrId("fingerprints")
+	require.NoError(t, err, "Failed to find fingerprints collection")
+	assert.Equal(t, isUserInSystemUsers, *fingerprintsCollection.ListRule)
+	assert.Equal(t, isUserInSystemUsers, *fingerprintsCollection.ViewRule)
+	assert.Equal(t, isUserInSystemUsersNotReadonly, *fingerprintsCollection.CreateRule)
+	assert.Equal(t, isUserInSystemUsersNotReadonly, *fingerprintsCollection.UpdateRule)
+	assert.Equal(t, isUserInSystemUsersNotReadonly, *fingerprintsCollection.DeleteRule)
+
+	// quiet_hours collection
+	quietHoursCollection, err := hub.FindCollectionByNameOrId("quiet_hours")
+	require.NoError(t, err, "Failed to find quiet_hours collection")
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.ListRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.DeleteRule)
+
+	// smart_devices collection
+	smartDevicesCollection, err := hub.FindCollectionByNameOrId("smart_devices")
+	require.NoError(t, err, "Failed to find smart_devices collection")
+	assert.Equal(t, isUserInSystemUsers, *smartDevicesCollection.ListRule)
+	assert.Equal(t, isUserInSystemUsers, *smartDevicesCollection.ViewRule)
+	assert.Nil(t, smartDevicesCollection.CreateRule)
+	assert.Nil(t, smartDevicesCollection.UpdateRule)
+	assert.Equal(t, isUserInSystemUsersNotReadonly, *smartDevicesCollection.DeleteRule)
+
+	// system_details collection
+	systemDetailsCollection, err := hub.FindCollectionByNameOrId("system_details")
+	require.NoError(t, err, "Failed to find system_details collection")
+	assert.Equal(t, isUserInSystemUsers, *systemDetailsCollection.ListRule)
+	assert.Equal(t, isUserInSystemUsers, *systemDetailsCollection.ViewRule)
+	assert.Nil(t, systemDetailsCollection.CreateRule)
+	assert.Nil(t, systemDetailsCollection.UpdateRule)
+	assert.Nil(t, systemDetailsCollection.DeleteRule)
+
+	// system_stats collection
+	systemStatsCollection, err := hub.FindCollectionByNameOrId("system_stats")
+	require.NoError(t, err, "Failed to find system_stats collection")
+	assert.Equal(t, isUserInSystemUsers, *systemStatsCollection.ListRule)
+	assert.Nil(t, systemStatsCollection.ViewRule)
+	assert.Nil(t, systemStatsCollection.CreateRule)
+	assert.Nil(t, systemStatsCollection.UpdateRule)
+	assert.Nil(t, systemStatsCollection.DeleteRule)
+
+	// systemd_services collection
+	systemdServicesCollection, err := hub.FindCollectionByNameOrId("systemd_services")
+	require.NoError(t, err, "Failed to find systemd_services collection")
+	assert.Equal(t, isUserInSystemUsers, *systemdServicesCollection.ListRule)
+	assert.Nil(t, systemdServicesCollection.ViewRule)
+	assert.Nil(t, systemdServicesCollection.CreateRule)
+	assert.Nil(t, systemdServicesCollection.UpdateRule)
+	assert.Nil(t, systemdServicesCollection.DeleteRule)
+
+	// systems collection
+	systemsCollection, err := hub.FindCollectionByNameOrId("systems")
+	require.NoError(t, err, "Failed to find systems collection")
+	assert.Equal(t, isUserInUsers, *systemsCollection.ListRule)
+	assert.Equal(t, isUserInUsers, *systemsCollection.ViewRule)
+	assert.Equal(t, isUserInUsersNotReadonly, *systemsCollection.CreateRule)
+	assert.Equal(t, isUserInUsersNotReadonly, *systemsCollection.UpdateRule)
+	assert.Equal(t, isUserInUsersNotReadonly, *systemsCollection.DeleteRule)
+
+	// universal_tokens collection
+	universalTokensCollection, err := hub.FindCollectionByNameOrId("universal_tokens")
+	require.NoError(t, err, "Failed to find universal_tokens collection")
+	assert.Nil(t, universalTokensCollection.ListRule)
+	assert.Nil(t, universalTokensCollection.ViewRule)
+	assert.Nil(t, universalTokensCollection.CreateRule)
+	assert.Nil(t, universalTokensCollection.UpdateRule)
+	assert.Nil(t, universalTokensCollection.DeleteRule)
+
+	// user_settings collection
+	userSettingsCollection, err := hub.FindCollectionByNameOrId("user_settings")
+	require.NoError(t, err, "Failed to find user_settings collection")
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.ListRule)
+	assert.Nil(t, userSettingsCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.UpdateRule)
+	assert.Nil(t, userSettingsCollection.DeleteRule)
+}
+
+func TestCollectionRulesShareAllSystems(t *testing.T) {
+	t.Setenv("SHARE_ALL_SYSTEMS", "true")
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	const isUser = `@request.auth.id != ""`
+	const isUserNotReadonly = `@request.auth.id != "" && @request.auth.role != "readonly"`
+
+	const isUserMatchesUser = `@request.auth.id != "" && user = @request.auth.id`
+
+	// alerts collection
+	alertsCollection, err := hub.FindCollectionByNameOrId("alerts")
+	require.NoError(t, err, "Failed to find alerts collection")
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.ListRule)
+	assert.Nil(t, alertsCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsCollection.DeleteRule)
+
+	// alerts_history collection
+	alertsHistoryCollection, err := hub.FindCollectionByNameOrId("alerts_history")
+	require.NoError(t, err, "Failed to find alerts_history collection")
+	assert.Equal(t, isUserMatchesUser, *alertsHistoryCollection.ListRule)
+	assert.Nil(t, alertsHistoryCollection.ViewRule)
+	assert.Nil(t, alertsHistoryCollection.CreateRule)
+	assert.Nil(t, alertsHistoryCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *alertsHistoryCollection.DeleteRule)
+
+	// containers collection
+	containersCollection, err := hub.FindCollectionByNameOrId("containers")
+	require.NoError(t, err, "Failed to find containers collection")
+	assert.Equal(t, isUser, *containersCollection.ListRule)
+	assert.Nil(t, containersCollection.ViewRule)
+	assert.Nil(t, containersCollection.CreateRule)
+	assert.Nil(t, containersCollection.UpdateRule)
+	assert.Nil(t, containersCollection.DeleteRule)
+
+	// container_stats collection
+	containerStatsCollection, err := hub.FindCollectionByNameOrId("container_stats")
+	require.NoError(t, err, "Failed to find container_stats collection")
+	assert.Equal(t, isUser, *containerStatsCollection.ListRule)
+	assert.Nil(t, containerStatsCollection.ViewRule)
+	assert.Nil(t, containerStatsCollection.CreateRule)
+	assert.Nil(t, containerStatsCollection.UpdateRule)
+	assert.Nil(t, containerStatsCollection.DeleteRule)
+
+	// fingerprints collection
+	fingerprintsCollection, err := hub.FindCollectionByNameOrId("fingerprints")
+	require.NoError(t, err, "Failed to find fingerprints collection")
+	assert.Equal(t, isUser, *fingerprintsCollection.ListRule)
+	assert.Equal(t, isUser, *fingerprintsCollection.ViewRule)
+	assert.Equal(t, isUserNotReadonly, *fingerprintsCollection.CreateRule)
+	assert.Equal(t, isUserNotReadonly, *fingerprintsCollection.UpdateRule)
+	assert.Equal(t, isUserNotReadonly, *fingerprintsCollection.DeleteRule)
+
+	// quiet_hours collection
+	quietHoursCollection, err := hub.FindCollectionByNameOrId("quiet_hours")
+	require.NoError(t, err, "Failed to find quiet_hours collection")
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.ListRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.UpdateRule)
+	assert.Equal(t, isUserMatchesUser, *quietHoursCollection.DeleteRule)
+
+	// smart_devices collection
+	smartDevicesCollection, err := hub.FindCollectionByNameOrId("smart_devices")
+	require.NoError(t, err, "Failed to find smart_devices collection")
+	assert.Equal(t, isUser, *smartDevicesCollection.ListRule)
+	assert.Equal(t, isUser, *smartDevicesCollection.ViewRule)
+	assert.Nil(t, smartDevicesCollection.CreateRule)
+	assert.Nil(t, smartDevicesCollection.UpdateRule)
+	assert.Equal(t, isUserNotReadonly, *smartDevicesCollection.DeleteRule)
+
+	// system_details collection
+	systemDetailsCollection, err := hub.FindCollectionByNameOrId("system_details")
+	require.NoError(t, err, "Failed to find system_details collection")
+	assert.Equal(t, isUser, *systemDetailsCollection.ListRule)
+	assert.Equal(t, isUser, *systemDetailsCollection.ViewRule)
+	assert.Nil(t, systemDetailsCollection.CreateRule)
+	assert.Nil(t, systemDetailsCollection.UpdateRule)
+	assert.Nil(t, systemDetailsCollection.DeleteRule)
+
+	// system_stats collection
+	systemStatsCollection, err := hub.FindCollectionByNameOrId("system_stats")
+	require.NoError(t, err, "Failed to find system_stats collection")
+	assert.Equal(t, isUser, *systemStatsCollection.ListRule)
+	assert.Nil(t, systemStatsCollection.ViewRule)
+	assert.Nil(t, systemStatsCollection.CreateRule)
+	assert.Nil(t, systemStatsCollection.UpdateRule)
+	assert.Nil(t, systemStatsCollection.DeleteRule)
+
+	// systemd_services collection
+	systemdServicesCollection, err := hub.FindCollectionByNameOrId("systemd_services")
+	require.NoError(t, err, "Failed to find systemd_services collection")
+	assert.Equal(t, isUser, *systemdServicesCollection.ListRule)
+	assert.Nil(t, systemdServicesCollection.ViewRule)
+	assert.Nil(t, systemdServicesCollection.CreateRule)
+	assert.Nil(t, systemdServicesCollection.UpdateRule)
+	assert.Nil(t, systemdServicesCollection.DeleteRule)
+
+	// systems collection
+	systemsCollection, err := hub.FindCollectionByNameOrId("systems")
+	require.NoError(t, err, "Failed to find systems collection")
+	assert.Equal(t, isUser, *systemsCollection.ListRule)
+	assert.Equal(t, isUser, *systemsCollection.ViewRule)
+	assert.Equal(t, isUserNotReadonly, *systemsCollection.CreateRule)
+	assert.Equal(t, isUserNotReadonly, *systemsCollection.UpdateRule)
+	assert.Equal(t, isUserNotReadonly, *systemsCollection.DeleteRule)
+
+	// universal_tokens collection
+	universalTokensCollection, err := hub.FindCollectionByNameOrId("universal_tokens")
+	require.NoError(t, err, "Failed to find universal_tokens collection")
+	assert.Nil(t, universalTokensCollection.ListRule)
+	assert.Nil(t, universalTokensCollection.ViewRule)
+	assert.Nil(t, universalTokensCollection.CreateRule)
+	assert.Nil(t, universalTokensCollection.UpdateRule)
+	assert.Nil(t, universalTokensCollection.DeleteRule)
+
+	// user_settings collection
+	userSettingsCollection, err := hub.FindCollectionByNameOrId("user_settings")
+	require.NoError(t, err, "Failed to find user_settings collection")
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.ListRule)
+	assert.Nil(t, userSettingsCollection.ViewRule)
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.CreateRule)
+	assert.Equal(t, isUserMatchesUser, *userSettingsCollection.UpdateRule)
+	assert.Nil(t, userSettingsCollection.DeleteRule)
+}
+
+func TestDisablePasswordAuth(t *testing.T) {
+	t.Setenv("DISABLE_PASSWORD_AUTH", "true")
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	usersCollection, err := hub.FindCollectionByNameOrId("users")
+	assert.NoError(t, err)
+	assert.False(t, usersCollection.PasswordAuth.Enabled)
+}
+
+func TestUserCreation(t *testing.T) {
+	t.Setenv("USER_CREATION", "true")
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	usersCollection, err := hub.FindCollectionByNameOrId("users")
+	assert.NoError(t, err)
+	assert.Equal(t, "@request.context = 'oauth2'", *usersCollection.CreateRule)
+}
+
+func TestMFAOtp(t *testing.T) {
+	t.Setenv("MFA_OTP", "true")
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	usersCollection, err := hub.FindCollectionByNameOrId("users")
+	assert.NoError(t, err)
+	assert.True(t, usersCollection.OTP.Enabled)
+	assert.True(t, usersCollection.MFA.Enabled)
+
+	superusersCollection, err := hub.FindCollectionByNameOrId(core.CollectionNameSuperusers)
+	assert.NoError(t, err)
+	assert.True(t, superusersCollection.OTP.Enabled)
+	assert.True(t, superusersCollection.MFA.Enabled)
+}
+
+func TestApiCollectionsAuthRules(t *testing.T) {
+	hub, _ := beszelTests.NewTestHub(t.TempDir())
+	defer hub.Cleanup()
+
+	hub.StartHub()
+
+	user1, _ := beszelTests.CreateUser(hub, "user1@example.com", "password")
+	user1Token, _ := user1.NewAuthToken()
+
+	user2, _ := beszelTests.CreateUser(hub, "user2@example.com", "password")
+	// user2Token, _ := user2.NewAuthToken()
+
+	userReadonly, _ := beszelTests.CreateUserWithRole(hub, "userreadonly@example.com", "password", "readonly")
+	userReadonlyToken, _ := userReadonly.NewAuthToken()
+
+	userOneSystem, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system1",
+		"users": []string{user1.Id},
+		"host":  "127.0.0.1",
+	})
+
+	sharedSystem, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system2",
+		"users": []string{user1.Id, user2.Id},
+		"host":  "127.0.0.2",
+	})
+
+	userTwoSystem, _ := beszelTests.CreateRecord(hub, "systems", map[string]any{
+		"name":  "system3",
+		"users": []string{user2.Id},
+		"host":  "127.0.0.2",
+	})
+
+	userRecords, _ := hub.CountRecords("users")
+	assert.EqualValues(t, 3, userRecords, "all users should be created")
+
+	systemRecords, _ := hub.CountRecords("systems")
+	assert.EqualValues(t, 3, systemRecords, "all systems should be created")
+
+	testAppFactory := func(t testing.TB) *pbTests.TestApp {
+		return hub.TestApp
+	}
+
+	scenarios := []beszelTests.ApiScenario{
+		{
+			Name:               "Unauthorized user cannot list systems",
+			Method:             http.MethodGet,
+			URL:                "/api/collections/systems/records",
+			ExpectedStatus:     200, // https://github.com/pocketbase/pocketbase/discussions/1570
+			TestAppFactory:     testAppFactory,
+			ExpectedContent:    []string{`"items":[]`, `"totalItems":0`},
+			NotExpectedContent: []string{userOneSystem.Id, sharedSystem.Id, userTwoSystem.Id},
+		},
+		{
+			Name:               "Unauthorized user cannot delete a system",
+			Method:             http.MethodDelete,
+			URL:                fmt.Sprintf("/api/collections/systems/records/%s", userOneSystem.Id),
+			ExpectedStatus:     404,
+			TestAppFactory:     testAppFactory,
+			ExpectedContent:    []string{"resource wasn't found"},
+			NotExpectedContent: []string{userOneSystem.Id},
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 3, systemsCount, "should have 3 systems before deletion")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 3, systemsCount, "should still have 3 systems after failed deletion")
+			},
+		},
+		{
+			Name:   "User 1 can list their own systems",
+			Method: http.MethodGet,
+			URL:    "/api/collections/systems/records",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:     200,
+			ExpectedContent:    []string{userOneSystem.Id, sharedSystem.Id},
+			NotExpectedContent: []string{userTwoSystem.Id},
+			TestAppFactory:     testAppFactory,
+		},
+		{
+			Name:   "User 1 cannot list user 2's system",
+			Method: http.MethodGet,
+			URL:    "/api/collections/systems/records",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:     200,
+			ExpectedContent:    []string{userOneSystem.Id, sharedSystem.Id},
+			NotExpectedContent: []string{userTwoSystem.Id},
+			TestAppFactory:     testAppFactory,
+		},
+		{
+			Name:   "User 1 can see user 2's system if SHARE_ALL_SYSTEMS is enabled",
+			Method: http.MethodGet,
+			URL:    "/api/collections/systems/records",
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  200,
+			ExpectedContent: []string{userOneSystem.Id, sharedSystem.Id, userTwoSystem.Id},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "true")
+				hub.SetCollectionAuthSettings()
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "")
+				hub.SetCollectionAuthSettings()
+			},
+		},
+		{
+			Name:   "User 1 can delete their own system",
+			Method: http.MethodDelete,
+			URL:    fmt.Sprintf("/api/collections/systems/records/%s", userOneSystem.Id),
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus: 204,
+			TestAppFactory: testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 3, systemsCount, "should have 3 systems before deletion")
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount, "should have 2 systems after deletion")
+			},
+		},
+		{
+			Name:   "User 1 cannot delete user 2's system",
+			Method: http.MethodDelete,
+			URL:    fmt.Sprintf("/api/collections/systems/records/%s", userTwoSystem.Id),
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus:  404,
+			TestAppFactory:  testAppFactory,
+			ExpectedContent: []string{"resource wasn't found"},
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount)
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount)
+			},
+		},
+		{
+			Name:   "Readonly cannot delete a system even if SHARE_ALL_SYSTEMS is enabled",
+			Method: http.MethodDelete,
+			URL:    fmt.Sprintf("/api/collections/systems/records/%s", sharedSystem.Id),
+			Headers: map[string]string{
+				"Authorization": userReadonlyToken,
+			},
+			ExpectedStatus:  404,
+			ExpectedContent: []string{"resource wasn't found"},
+			TestAppFactory:  testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "true")
+				hub.SetCollectionAuthSettings()
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount)
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "")
+				hub.SetCollectionAuthSettings()
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount)
+			},
+		},
+		{
+			Name:   "User 1 can delete user 2's system if SHARE_ALL_SYSTEMS is enabled",
+			Method: http.MethodDelete,
+			URL:    fmt.Sprintf("/api/collections/systems/records/%s", userTwoSystem.Id),
+			Headers: map[string]string{
+				"Authorization": user1Token,
+			},
+			ExpectedStatus: 204,
+			TestAppFactory: testAppFactory,
+			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "true")
+				hub.SetCollectionAuthSettings()
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 2, systemsCount)
+			},
+			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
+				t.Setenv("SHARE_ALL_SYSTEMS", "")
+				hub.SetCollectionAuthSettings()
+				systemsCount, _ := app.CountRecords("systems")
+				assert.EqualValues(t, 1, systemsCount)
+			},
+		},
+	}
+
+	for _, scenario := range scenarios {
+		scenario.Test(t)
+	}
+}

internal/hub/config/config.go

@@ -279,9 +279,6 @@ func createFingerprintRecord(app core.App, systemID, token string) error {
 
 // Returns the current config.yml file as a JSON object
 func GetYamlConfig(e *core.RequestEvent) error {
-	if e.Auth.GetString("role") != "admin" {
-		return e.ForbiddenError("Requires admin role", nil)
-	}
 	configContent, err := generateYAML(e.App)
 	if err != nil {
 		return err

internal/hub/config/config_test.go

@@ -64,7 +64,7 @@ func createConfigTestFingerprint(app core.App, systemID, token, fingerprint stri
 
 // TestConfigSyncWithTokens tests the config.SyncSystems function with various token scenarios
 func TestConfigSyncWithTokens(t *testing.T) {
-	testHub, err := tests.NewTestHub()
+	testHub, err := tests.NewTestHub(t.TempDir())
 	require.NoError(t, err)
 	defer testHub.Cleanup()
 

internal/hub/expirymap/expirymap.go

@@ -16,7 +16,7 @@ type val[T comparable] struct {
 }
 
 type ExpiryMap[T comparable] struct {
-	store    store.Store[string, val[T]]
+	store    *store.Store[string, val[T]]
 	stopChan chan struct{}
 	stopOnce sync.Once
 }
@@ -24,7 +24,7 @@ type ExpiryMap[T comparable] struct {
 // New creates a new expiry map with custom cleanup interval
 func New[T comparable](cleanupInterval time.Duration) *ExpiryMap[T] {
 	m := &ExpiryMap[T]{
-		store:    *store.New(map[string]val[T]{}),
+		store:    store.New(map[string]val[T]{}),
 		stopChan: make(chan struct{}),
 	}
 	go m.startCleaner(cleanupInterval)

internal/hub/hub.go

@@ -4,31 +4,27 @@ package hub
 import (
 	"crypto/ed25519"
 	"encoding/pem"
+	"errors"
 	"fmt"
-	"net/http"
 	"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/hub/utils"
 	"github.com/henrygd/beszel/internal/records"
 	"github.com/henrygd/beszel/internal/users"
 
-	"github.com/google/uuid"
-	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase"
-	"github.com/pocketbase/pocketbase/apis"
 	"github.com/pocketbase/pocketbase/core"
 	"golang.org/x/crypto/ssh"
 )
 
+// Hub is the application. It embeds the PocketBase app and keeps references to subcomponents.
 type Hub struct {
 	core.App
 	*alerts.AlertManager
@@ -42,40 +38,43 @@ type Hub struct {
 	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{}
-	hub.App = app
-
+	hub := &Hub{App: app}
 	hub.AlertManager = alerts.NewAlertManager(hub)
 	hub.um = users.NewUserManager(hub)
 	hub.rm = records.NewRecordManager(hub)
 	hub.sm = systems.NewSystemManager(hub)
-	hub.appURL, _ = GetEnv("APP_URL")
-	hub.hb = heartbeat.New(app, GetEnv)
+	hub.hb = heartbeat.New(app, utils.GetEnv)
 	if hub.hb != nil {
 		hub.hbStop = make(chan struct{})
 	}
+	_ = onAfterBootstrapAndMigrations(app, hub.initialize)
 	return hub
 }
 
-// GetEnv retrieves an environment variable with a "BESZEL_HUB_" prefix, or falls back to the unprefixed key.
-func GetEnv(key string) (value string, exists bool) {
-	if value, exists = os.LookupEnv("BESZEL_HUB_" + key); exists {
-		return value, exists
+// onAfterBootstrapAndMigrations ensures the provided function runs after the database is set up and migrations are applied.
+// This is a workaround for behavior in PocketBase where onBootstrap runs before migrations, forcing use of onServe for this purpose.
+// However, PB's tests.TestApp is already bootstrapped, generally doesn't serve, but does handle migrations.
+// So this ensures that the provided function runs at the right time either way, after DB is ready and migrations are done.
+func onAfterBootstrapAndMigrations(app core.App, fn func(app core.App) error) error {
+	// pb tests.TestApp is already bootstrapped and doesn't serve
+	if app.IsBootstrapped() {
+		return fn(app)
 	}
-	// Fallback to the old unprefixed key
-	return os.LookupEnv(key)
-}
-
-func (h *Hub) StartHub() error {
-	h.App.OnServe().BindFunc(func(e *core.ServeEvent) error {
-		// initialize settings / collections
-		if err := h.initialize(e); err != nil {
+	// Must use OnServe because OnBootstrap appears to run before migrations, even if calling e.Next() before anything else
+	app.OnServe().BindFunc(func(e *core.ServeEvent) error {
+		if err := fn(e.App); err != nil {
 			return err
 		}
+		return e.Next()
+	})
+	return nil
+}
+
+// StartHub sets up event handlers and starts the PocketBase server
+func (h *Hub) StartHub() error {
+	h.App.OnServe().BindFunc(func(e *core.ServeEvent) error {
 		// sync systems with config
 		if err := config.SyncSystems(e); err != nil {
 			return err
@@ -110,132 +109,29 @@ func (h *Hub) StartHub() error {
 	h.App.OnRecordCreate("users").BindFunc(h.um.InitializeUserRole)
 	h.App.OnRecordCreate("user_settings").BindFunc(h.um.InitializeUserSettings)
 
-	if pb, ok := h.App.(*pocketbase.PocketBase); ok {
-		// log.Println("Starting pocketbase")
-		err := pb.Start()
-		if err != nil {
-			return err
-		}
+	pb, ok := h.App.(*pocketbase.PocketBase)
+	if !ok {
+		return errors.New("not a pocketbase app")
 	}
-
-	return nil
+	return pb.Start()
 }
 
 // initialize sets up initial configuration (collections, settings, etc.)
-func (h *Hub) initialize(e *core.ServeEvent) error {
+func (h *Hub) initialize(app core.App) error {
 	// set general settings
-	settings := e.App.Settings()
-	// batch requests (for global alerts)
+	settings := app.Settings()
+	// batch requests (for alerts)
 	settings.Batch.Enabled = true
-	// set URL if BASE_URL env is set
-	if h.appURL != "" {
-		settings.Meta.AppURL = h.appURL
+	// set URL if APP_URL env is set
+	if appURL, isSet := utils.GetEnv("APP_URL"); isSet {
+		h.appURL = appURL
+		settings.Meta.AppURL = appURL
 	}
-	if err := e.App.Save(settings); err != nil {
+	if err := app.Save(settings); err != nil {
 		return err
 	}
 	// set auth settings
-	if err := setCollectionAuthSettings(e.App); err != nil {
-		return err
-	}
-	return nil
-}
-
-// setCollectionAuthSettings sets up default authentication settings for the app
-func setCollectionAuthSettings(app core.App) error {
-	usersCollection, err := app.FindCollectionByNameOrId("users")
-	if err != nil {
-		return err
-	}
-	superusersCollection, err := app.FindCollectionByNameOrId(core.CollectionNameSuperusers)
-	if err != nil {
-		return err
-	}
-
-	// disable email auth if DISABLE_PASSWORD_AUTH env var is set
-	disablePasswordAuth, _ := GetEnv("DISABLE_PASSWORD_AUTH")
-	usersCollection.PasswordAuth.Enabled = disablePasswordAuth != "true"
-	usersCollection.PasswordAuth.IdentityFields = []string{"email"}
-	// allow oauth user creation if USER_CREATION is set
-	if userCreation, _ := GetEnv("USER_CREATION"); userCreation == "true" {
-		cr := "@request.context = 'oauth2'"
-		usersCollection.CreateRule = &cr
-	} else {
-		usersCollection.CreateRule = nil
-	}
-
-	// enable mfaOtp mfa if MFA_OTP env var is set
-	mfaOtp, _ := GetEnv("MFA_OTP")
-	usersCollection.OTP.Length = 6
-	superusersCollection.OTP.Length = 6
-	usersCollection.OTP.Enabled = mfaOtp == "true"
-	usersCollection.MFA.Enabled = mfaOtp == "true"
-	superusersCollection.OTP.Enabled = mfaOtp == "true" || mfaOtp == "superusers"
-	superusersCollection.MFA.Enabled = mfaOtp == "true" || mfaOtp == "superusers"
-	if err := app.Save(superusersCollection); err != nil {
-		return err
-	}
-	if err := app.Save(usersCollection); err != nil {
-		return err
-	}
-
-	shareAllSystems, _ := GetEnv("SHARE_ALL_SYSTEMS")
-
-	// allow all users to access systems if SHARE_ALL_SYSTEMS is set
-	systemsCollection, err := app.FindCollectionByNameOrId("systems")
-	if err != nil {
-		return err
-	}
-	var systemsReadRule string
-	if shareAllSystems == "true" {
-		systemsReadRule = "@request.auth.id != \"\""
-	} else {
-		systemsReadRule = "@request.auth.id != \"\" && users.id ?= @request.auth.id"
-	}
-	updateDeleteRule := systemsReadRule + " && @request.auth.role != \"readonly\""
-	systemsCollection.ListRule = &systemsReadRule
-	systemsCollection.ViewRule = &systemsReadRule
-	systemsCollection.UpdateRule = &updateDeleteRule
-	systemsCollection.DeleteRule = &updateDeleteRule
-	if err := app.Save(systemsCollection); err != nil {
-		return err
-	}
-
-	// allow all users to access all containers if SHARE_ALL_SYSTEMS is set
-	containersCollection, err := app.FindCollectionByNameOrId("containers")
-	if err != nil {
-		return err
-	}
-	containersListRule := strings.Replace(systemsReadRule, "users.id", "system.users.id", 1)
-	containersCollection.ListRule = &containersListRule
-	if err := app.Save(containersCollection); err != nil {
-		return err
-	}
-
-	// allow all users to access system-related collections if SHARE_ALL_SYSTEMS is set
-	// these collections all have a "system" relation field
-	systemRelatedCollections := []string{"system_details", "smart_devices", "systemd_services"}
-	for _, collectionName := range systemRelatedCollections {
-		collection, err := app.FindCollectionByNameOrId(collectionName)
-		if err != nil {
-			return err
-		}
-		collection.ListRule = &containersListRule
-		// set viewRule for collections that need it (system_details, smart_devices)
-		if collection.ViewRule != nil {
-			collection.ViewRule = &containersListRule
-		}
-		// set deleteRule for smart_devices (allows user to dismiss disk warnings)
-		if collectionName == "smart_devices" {
-			deleteRule := containersListRule + " && @request.auth.role != \"readonly\""
-			collection.DeleteRule = &deleteRule
-		}
-		if err := app.Save(collection); err != nil {
-			return err
-		}
-	}
-
-	return nil
+	return setCollectionAuthSettings(app)
 }
 
 // registerCronJobs sets up scheduled tasks
@@ -247,296 +143,7 @@ func (h *Hub) registerCronJobs(_ *core.ServeEvent) error {
 	return nil
 }
 
-// custom middlewares
-func (h *Hub) registerMiddlewares(se *core.ServeEvent) {
-	// authorizes request with user matching the provided email
-	authorizeRequestWithEmail := func(e *core.RequestEvent, email string) (err error) {
-		if e.Auth != nil || email == "" {
-			return e.Next()
-		}
-		isAuthRefresh := e.Request.URL.Path == "/api/collections/users/auth-refresh" && e.Request.Method == http.MethodPost
-		e.Auth, err = e.App.FindFirstRecordByData("users", "email", email)
-		if err != nil || !isAuthRefresh {
-			return e.Next()
-		}
-		// auth refresh endpoint, make sure token is set in header
-		token, _ := e.Auth.NewAuthToken()
-		e.Request.Header.Set("Authorization", token)
-		return e.Next()
-	}
-	// authenticate with trusted header
-	if autoLogin, _ := GetEnv("AUTO_LOGIN"); autoLogin != "" {
-		se.Router.BindFunc(func(e *core.RequestEvent) error {
-			return authorizeRequestWithEmail(e, autoLogin)
-		})
-	}
-	// authenticate with trusted header
-	if trustedHeader, _ := GetEnv("TRUSTED_AUTH_HEADER"); trustedHeader != "" {
-		se.Router.BindFunc(func(e *core.RequestEvent) error {
-			return authorizeRequestWithEmail(e, e.Request.Header.Get(trustedHeader))
-		})
-	}
-}
-
-// custom api routes
-func (h *Hub) registerApiRoutes(se *core.ServeEvent) error {
-	// auth protected routes
-	apiAuth := se.Router.Group("/api/beszel")
-	apiAuth.Bind(apis.RequireAuth())
-	// auth optional routes
-	apiNoAuth := se.Router.Group("/api/beszel")
-
-	// create first user endpoint only needed if no users exist
-	if totalUsers, _ := se.App.CountRecords("users"); totalUsers == 0 {
-		apiNoAuth.POST("/create-user", h.um.CreateFirstUser)
-	}
-	// check if first time setup on login page
-	apiNoAuth.GET("/first-run", func(e *core.RequestEvent) error {
-		total, err := e.App.CountRecords("users")
-		return e.JSON(http.StatusOK, map[string]bool{"firstRun": err == nil && total == 0})
-	})
-	// get public key and version
-	apiAuth.GET("/getkey", func(e *core.RequestEvent) error {
-		return e.JSON(http.StatusOK, map[string]string{"key": h.pubKey, "v": beszel.Version})
-	})
-	// 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
-	apiNoAuth.GET("/agent-connect", h.handleAgentConnect)
-	// get or create universal tokens
-	apiAuth.GET("/universal-token", h.getUniversalToken)
-	// update / delete user alerts
-	apiAuth.POST("/user-alerts", alerts.UpsertUserAlerts)
-	apiAuth.DELETE("/user-alerts", alerts.DeleteUserAlerts)
-	// refresh SMART devices for a system
-	apiAuth.POST("/smart/refresh", h.refreshSmartData)
-	// get systemd service details
-	apiAuth.GET("/systemd/info", h.getSystemdInfo)
-	// /containers routes
-	if enabled, _ := GetEnv("CONTAINER_DETAILS"); enabled != "false" {
-		// get container logs
-		apiAuth.GET("/containers/logs", h.getContainerLogs)
-		// get container info
-		apiAuth.GET("/containers/info", h.getContainerInfo)
-	}
-	return nil
-}
-
-// Handler for universal token API endpoint (create, read, delete)
-func (h *Hub) getUniversalToken(e *core.RequestEvent) error {
-	tokenMap := universalTokenMap.GetMap()
-	userID := e.Auth.Id
-	query := e.Request.URL.Query()
-	token := query.Get("token")
-	enable := query.Get("enable")
-	permanent := query.Get("permanent")
-
-	// helper for deleting any existing permanent token record for this user
-	deletePermanent := func() error {
-		rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID})
-		if err != nil {
-			return nil // no record
-		}
-		return h.Delete(rec)
-	}
-
-	// helper for upserting a permanent token record for this user
-	upsertPermanent := func(token string) error {
-		rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID})
-		if err == nil {
-			rec.Set("token", token)
-			return h.Save(rec)
-		}
-
-		col, err := h.FindCachedCollectionByNameOrId("universal_tokens")
-		if err != nil {
-			return err
-		}
-		newRec := core.NewRecord(col)
-		newRec.Set("user", userID)
-		newRec.Set("token", token)
-		return h.Save(newRec)
-	}
-
-	// Disable universal tokens (both ephemeral and permanent)
-	if enable == "0" {
-		tokenMap.RemovebyValue(userID)
-		_ = deletePermanent()
-		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": false, "permanent": false})
-	}
-
-	// Enable universal token (ephemeral or permanent)
-	if enable == "1" {
-		if token == "" {
-			token = uuid.New().String()
-		}
-
-		if permanent == "1" {
-			// make token permanent (persist across restarts)
-			tokenMap.RemovebyValue(userID)
-			if err := upsertPermanent(token); err != nil {
-				return err
-			}
-			return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": true})
-		}
-
-		// default: ephemeral mode (1 hour)
-		_ = deletePermanent()
-		tokenMap.Set(token, userID, time.Hour)
-		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": false})
-	}
-
-	// Read current state
-	// Prefer permanent token if it exists.
-	if rec, err := h.FindFirstRecordByFilter("universal_tokens", "user = {:user}", dbx.Params{"user": userID}); err == nil {
-		dbToken := rec.GetString("token")
-		// If no token was provided, or the caller is asking about their permanent token, return it.
-		if token == "" || token == dbToken {
-			return e.JSON(http.StatusOK, map[string]any{"token": dbToken, "active": true, "permanent": true})
-		}
-		// Token doesn't match their permanent token (avoid leaking other info)
-		return e.JSON(http.StatusOK, map[string]any{"token": token, "active": false, "permanent": false})
-	}
-
-	// No permanent token; fall back to ephemeral token map.
-	if token == "" {
-		// return existing token if it exists
-		if token, _, ok := tokenMap.GetByValue(userID); ok {
-			return e.JSON(http.StatusOK, map[string]any{"token": token, "active": true, "permanent": false})
-		}
-		// if no token is provided, generate a new one
-		token = uuid.New().String()
-	}
-
-	// Token is considered active only if it belongs to the current user.
-	activeUser, ok := tokenMap.GetOk(token)
-	active := ok && activeUser == userID
-	response := map[string]any{"token": token, "active": active, "permanent": false}
-	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")
-	containerID := e.Request.URL.Query().Get("container")
-
-	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 {
-		return e.JSON(http.StatusNotFound, map[string]string{"error": "system not found"})
-	}
-
-	data, err := fetchFunc(system, containerID)
-	if err != nil {
-		return e.JSON(http.StatusNotFound, map[string]string{"error": err.Error()})
-	}
-
-	return e.JSON(http.StatusOK, map[string]string{responseKey: data})
-}
-
-// getContainerLogs handles GET /api/beszel/containers/logs requests
-func (h *Hub) getContainerLogs(e *core.RequestEvent) error {
-	return h.containerRequestHandler(e, func(system *systems.System, containerID string) (string, error) {
-		return system.FetchContainerLogsFromAgent(containerID)
-	}, "logs")
-}
-
-func (h *Hub) getContainerInfo(e *core.RequestEvent) error {
-	return h.containerRequestHandler(e, func(system *systems.System, containerID string) (string, error) {
-		return system.FetchContainerInfoFromAgent(containerID)
-	}, "info")
-}
-
-// getSystemdInfo handles GET /api/beszel/systemd/info requests
-func (h *Hub) getSystemdInfo(e *core.RequestEvent) error {
-	query := e.Request.URL.Query()
-	systemID := query.Get("system")
-	serviceName := query.Get("service")
-
-	if systemID == "" || serviceName == "" {
-		return e.JSON(http.StatusBadRequest, map[string]string{"error": "system and service parameters are required"})
-	}
-	system, err := h.sm.GetSystem(systemID)
-	if err != nil {
-		return e.JSON(http.StatusNotFound, map[string]string{"error": "system not found"})
-	}
-	details, err := system.FetchSystemdInfoFromAgent(serviceName)
-	if err != nil {
-		return e.JSON(http.StatusNotFound, map[string]string{"error": err.Error()})
-	}
-	e.Response.Header().Set("Cache-Control", "public, max-age=60")
-	return e.JSON(http.StatusOK, map[string]any{"details": details})
-}
-
-// refreshSmartData handles POST /api/beszel/smart/refresh requests
-// Fetches fresh SMART data from the agent and updates the collection
-func (h *Hub) refreshSmartData(e *core.RequestEvent) error {
-	systemID := e.Request.URL.Query().Get("system")
-	if systemID == "" {
-		return e.JSON(http.StatusBadRequest, map[string]string{"error": "system parameter is required"})
-	}
-
-	system, err := h.sm.GetSystem(systemID)
-	if err != nil {
-		return e.JSON(http.StatusNotFound, map[string]string{"error": "system not found"})
-	}
-
-	// Fetch and save SMART devices
-	if err := system.FetchAndSaveSmartDevices(); err != nil {
-		return e.JSON(http.StatusInternalServerError, map[string]string{"error": err.Error()})
-	}
-
-	return e.JSON(http.StatusOK, map[string]string{"status": "ok"})
-}
-
-// generates key pair if it doesn't exist and returns signer
+// GetSSHKey generates key pair if it doesn't exist and returns signer
 func (h *Hub) GetSSHKey(dataDir string) (ssh.Signer, error) {
 	if h.signer != nil {
 		return h.signer, nil

internal/hub/hub_test.go

@@ -3,36 +3,20 @@
 package hub_test
 
 import (
-	"bytes"
 	"crypto/ed25519"
-	"encoding/json"
 	"encoding/pem"
-	"io"
-	"net/http"
 	"os"
 	"path/filepath"
 	"strings"
 	"testing"
 
-	"github.com/henrygd/beszel/internal/migrations"
 	beszelTests "github.com/henrygd/beszel/internal/tests"
 
-	"github.com/pocketbase/pocketbase/core"
-	pbTests "github.com/pocketbase/pocketbase/tests"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
 	"golang.org/x/crypto/ssh"
 )
 
-// marshal to json and return an io.Reader (for use in ApiScenario.Body)
-func jsonReader(v any) io.Reader {
-	data, err := json.Marshal(v)
-	if err != nil {
-		panic(err)
-	}
-	return bytes.NewReader(data)
-}
-
 func TestMakeLink(t *testing.T) {
 	hub, _ := beszelTests.NewTestHub(t.TempDir())
 
@@ -265,699 +249,20 @@ func TestGetSSHKey(t *testing.T) {
 	})
 }
 
-func TestApiRoutesAuthentication(t *testing.T) {
-	hub, _ := beszelTests.NewTestHub(t.TempDir())
-	defer hub.Cleanup()
-
-	hub.StartHub()
-
-	// Create test user and get auth token
-	user, err := beszelTests.CreateUser(hub, "testuser@example.com", "password123")
-	require.NoError(t, err, "Failed to create test user")
-
-	adminUser, err := beszelTests.CreateRecord(hub, "users", map[string]any{
-		"email":    "admin@example.com",
-		"password": "password123",
-		"role":     "admin",
-	})
-	require.NoError(t, err, "Failed to create admin user")
-	adminUserToken, err := adminUser.NewAuthToken()
-
-	// superUser, err := beszelTests.CreateRecord(hub, core.CollectionNameSuperusers, map[string]any{
-	// 	"email":    "superuser@example.com",
-	// 	"password": "password123",
-	// })
-	// require.NoError(t, err, "Failed to create superuser")
-
-	userToken, err := user.NewAuthToken()
-	require.NoError(t, err, "Failed to create auth token")
-
-	// Create test system for user-alerts endpoints
-	system, err := beszelTests.CreateRecord(hub, "systems", map[string]any{
-		"name":  "test-system",
-		"users": []string{user.Id},
-		"host":  "127.0.0.1",
-	})
-	require.NoError(t, err, "Failed to create test system")
-
-	testAppFactory := func(t testing.TB) *pbTests.TestApp {
-		return hub.TestApp
-	}
-
-	scenarios := []beszelTests.ApiScenario{
-		// Auth Protected Routes - Should require authentication
-		{
-			Name:            "POST /test-notification - no auth should fail",
-			Method:          http.MethodPost,
-			URL:             "/api/beszel/test-notification",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"url": "generic://127.0.0.1",
-			}),
-		},
-		{
-			Name:           "POST /test-notification - with auth should succeed",
-			Method:         http.MethodPost,
-			URL:            "/api/beszel/test-notification",
-			TestAppFactory: testAppFactory,
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			Body: jsonReader(map[string]any{
-				"url": "generic://127.0.0.1",
-			}),
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"sending message"},
-		},
-		{
-			Name:            "GET /config-yaml - no auth should fail",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/config-yaml",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /config-yaml - with user auth should fail",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/config-yaml",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  403,
-			ExpectedContent: []string{"Requires admin"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /config-yaml - with admin auth should succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/config-yaml",
-			Headers: map[string]string{
-				"Authorization": adminUserToken,
-			},
-			ExpectedStatus:  200,
-			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,
-			URL:             "/api/beszel/universal-token",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /universal-token - with auth should succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/universal-token",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"active", "token", "permanent"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /universal-token - enable permanent should succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/universal-token?enable=1&permanent=1&token=permanent-token-123",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"permanent\":true", "permanent-token-123"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:            "POST /user-alerts - no auth should fail",
-			Method:          http.MethodPost,
-			URL:             "/api/beszel/user-alerts",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"value":   80,
-				"min":     10,
-				"systems": []string{system.Id},
-			}),
-		},
-		{
-			Name:   "POST /user-alerts - with auth should succeed",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"value":   80,
-				"min":     10,
-				"systems": []string{system.Id},
-			}),
-		},
-		{
-			Name:            "DELETE /user-alerts - no auth should fail",
-			Method:          http.MethodDelete,
-			URL:             "/api/beszel/user-alerts",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"systems": []string{system.Id},
-			}),
-		},
-		{
-			Name:   "DELETE /user-alerts - with auth should succeed",
-			Method: http.MethodDelete,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"success\":true"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"systems": []string{system.Id},
-			}),
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				// Create an alert to delete
-				beszelTests.CreateRecord(app, "alerts", map[string]any{
-					"name":   "CPU",
-					"system": system.Id,
-					"user":   user.Id,
-					"value":  80,
-					"min":    10,
-				})
-			},
-		},
-		{
-			Name:            "GET /containers/logs - no auth should fail",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/containers/logs?system=test-system&container=test-container",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /containers/logs - with auth but missing system param should fail",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/containers/logs?container=test-container",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  400,
-			ExpectedContent: []string{"system and container parameters are required"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /containers/logs - with auth but missing container param should fail",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/containers/logs?system=test-system",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  400,
-			ExpectedContent: []string{"system and container parameters are required"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /containers/logs - with auth but invalid system should fail",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/containers/logs?system=invalid-system&container=0123456789ab",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  404,
-			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
-		{
-			Name:            "GET /getkey - no auth should fail",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/getkey",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /getkey - with auth should also succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/getkey",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"key\":", "\"v\":"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:            "GET /first-run - no auth should succeed",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/first-run",
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"firstRun\":false"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /first-run - with auth should also succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/first-run",
-			Headers: map[string]string{
-				"Authorization": userToken,
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"firstRun\":false"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:            "GET /agent-connect - no auth should succeed (websocket upgrade fails but route is accessible)",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/agent-connect",
-			ExpectedStatus:  400,
-			ExpectedContent: []string{},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "POST /test-notification - invalid auth token should fail",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/test-notification",
-			Body: jsonReader(map[string]any{
-				"url": "generic://127.0.0.1",
-			}),
-			Headers: map[string]string{
-				"Authorization": "invalid-token",
-			},
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "POST /user-alerts - invalid auth token should fail",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/user-alerts",
-			Headers: map[string]string{
-				"Authorization": "invalid-token",
-			},
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-			Body: jsonReader(map[string]any{
-				"name":    "CPU",
-				"value":   80,
-				"min":     10,
-				"systems": []string{system.Id},
-			}),
-		},
-	}
-
-	for _, scenario := range scenarios {
-		scenario.Test(t)
-	}
-}
-
-func TestFirstUserCreation(t *testing.T) {
-	t.Run("CreateUserEndpoint available when no users exist", func(t *testing.T) {
+func TestAppUrl(t *testing.T) {
+	t.Run("no APP_URL does't change app url", func(t *testing.T) {
 		hub, _ := beszelTests.NewTestHub(t.TempDir())
 		defer hub.Cleanup()
 
-		hub.StartHub()
-
-		testAppFactoryExisting := func(t testing.TB) *pbTests.TestApp {
-			return hub.TestApp
-		}
-
-		scenarios := []beszelTests.ApiScenario{
-			{
-				Name:   "POST /create-user - should be available when no users exist",
-				Method: http.MethodPost,
-				URL:    "/api/beszel/create-user",
-				Body: jsonReader(map[string]any{
-					"email":    "firstuser@example.com",
-					"password": "password123",
-				}),
-				ExpectedStatus:  200,
-				ExpectedContent: []string{"User created"},
-				TestAppFactory:  testAppFactoryExisting,
-				BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-					userCount, err := hub.CountRecords("users")
-					require.NoError(t, err)
-					require.Zero(t, userCount, "Should start with no users")
-					superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
-					require.NoError(t, err)
-					require.EqualValues(t, 1, len(superusers), "Should start with one temporary superuser")
-					require.EqualValues(t, migrations.TempAdminEmail, superusers[0].GetString("email"), "Should have created one temporary superuser")
-				},
-				AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-					userCount, err := hub.CountRecords("users")
-					require.NoError(t, err)
-					require.EqualValues(t, 1, userCount, "Should have created one user")
-					superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
-					require.NoError(t, err)
-					require.EqualValues(t, 1, len(superusers), "Should have created one superuser")
-					require.EqualValues(t, "firstuser@example.com", superusers[0].GetString("email"), "Should have created one superuser")
-				},
-			},
-			{
-				Name:   "POST /create-user - should not be available when users exist",
-				Method: http.MethodPost,
-				URL:    "/api/beszel/create-user",
-				Body: jsonReader(map[string]any{
-					"email":    "firstuser@example.com",
-					"password": "password123",
-				}),
-				ExpectedStatus:  404,
-				ExpectedContent: []string{"wasn't found"},
-				TestAppFactory:  testAppFactoryExisting,
-			},
-		}
-
-		for _, scenario := range scenarios {
-			scenario.Test(t)
-		}
+		settings := hub.Settings()
+		assert.Equal(t, "http://localhost:8090", settings.Meta.AppURL)
 	})
-
-	t.Run("CreateUserEndpoint not available when USER_EMAIL, USER_PASSWORD are set", func(t *testing.T) {
-		os.Setenv("BESZEL_HUB_USER_EMAIL", "me@example.com")
-		os.Setenv("BESZEL_HUB_USER_PASSWORD", "password123")
-		defer os.Unsetenv("BESZEL_HUB_USER_EMAIL")
-		defer os.Unsetenv("BESZEL_HUB_USER_PASSWORD")
-
+	t.Run("APP_URL changes app url", func(t *testing.T) {
+		t.Setenv("APP_URL", "http://example.com/app")
 		hub, _ := beszelTests.NewTestHub(t.TempDir())
 		defer hub.Cleanup()
 
-		hub.StartHub()
-
-		testAppFactory := func(t testing.TB) *pbTests.TestApp {
-			return hub.TestApp
-		}
-
-		scenario := beszelTests.ApiScenario{
-			Name:            "POST /create-user - should not be available when USER_EMAIL, USER_PASSWORD are set",
-			Method:          http.MethodPost,
-			URL:             "/api/beszel/create-user",
-			ExpectedStatus:  404,
-			ExpectedContent: []string{"wasn't found"},
-			TestAppFactory:  testAppFactory,
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				users, err := hub.FindAllRecords("users")
-				require.NoError(t, err)
-				require.EqualValues(t, 1, len(users), "Should start with one user")
-				require.EqualValues(t, "me@example.com", users[0].GetString("email"), "Should have created one user")
-				superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
-				require.NoError(t, err)
-				require.EqualValues(t, 1, len(superusers), "Should start with one superuser")
-				require.EqualValues(t, "me@example.com", superusers[0].GetString("email"), "Should have created one superuser")
-			},
-			AfterTestFunc: func(t testing.TB, app *pbTests.TestApp, res *http.Response) {
-				users, err := hub.FindAllRecords("users")
-				require.NoError(t, err)
-				require.EqualValues(t, 1, len(users), "Should still have one user")
-				require.EqualValues(t, "me@example.com", users[0].GetString("email"), "Should have created one user")
-				superusers, err := hub.FindAllRecords(core.CollectionNameSuperusers)
-				require.NoError(t, err)
-				require.EqualValues(t, 1, len(superusers), "Should still have one superuser")
-				require.EqualValues(t, "me@example.com", superusers[0].GetString("email"), "Should have created one superuser")
-			},
-		}
-
-		scenario.Test(t)
+		settings := hub.Settings()
+		assert.Equal(t, "http://example.com/app", settings.Meta.AppURL)
 	})
 }
-
-func TestCreateUserEndpointAvailability(t *testing.T) {
-	t.Run("CreateUserEndpoint available when no users exist", func(t *testing.T) {
-		hub, _ := beszelTests.NewTestHub(t.TempDir())
-		defer hub.Cleanup()
-
-		// Ensure no users exist
-		userCount, err := hub.CountRecords("users")
-		require.NoError(t, err)
-		require.Zero(t, userCount, "Should start with no users")
-
-		hub.StartHub()
-
-		testAppFactory := func(t testing.TB) *pbTests.TestApp {
-			return hub.TestApp
-		}
-
-		scenario := beszelTests.ApiScenario{
-			Name:   "POST /create-user - should be available when no users exist",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/create-user",
-			Body: jsonReader(map[string]any{
-				"email":    "firstuser@example.com",
-				"password": "password123",
-			}),
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"User created"},
-			TestAppFactory:  testAppFactory,
-		}
-
-		scenario.Test(t)
-
-		// Verify user was created
-		userCount, err = hub.CountRecords("users")
-		require.NoError(t, err)
-		require.EqualValues(t, 1, userCount, "Should have created one user")
-	})
-
-	t.Run("CreateUserEndpoint not available when users exist", func(t *testing.T) {
-		hub, _ := beszelTests.NewTestHub(t.TempDir())
-		defer hub.Cleanup()
-
-		// Create a user first
-		_, err := beszelTests.CreateUser(hub, "existing@example.com", "password")
-		require.NoError(t, err)
-
-		hub.StartHub()
-
-		testAppFactory := func(t testing.TB) *pbTests.TestApp {
-			return hub.TestApp
-		}
-
-		scenario := beszelTests.ApiScenario{
-			Name:   "POST /create-user - should not be available when users exist",
-			Method: http.MethodPost,
-			URL:    "/api/beszel/create-user",
-			Body: jsonReader(map[string]any{
-				"email":    "another@example.com",
-				"password": "password123",
-			}),
-			ExpectedStatus:  404,
-			ExpectedContent: []string{"wasn't found"},
-			TestAppFactory:  testAppFactory,
-		}
-
-		scenario.Test(t)
-	})
-}
-
-func TestAutoLoginMiddleware(t *testing.T) {
-	var hubs []*beszelTests.TestHub
-
-	defer func() {
-		defer os.Unsetenv("AUTO_LOGIN")
-		for _, hub := range hubs {
-			hub.Cleanup()
-		}
-	}()
-
-	os.Setenv("AUTO_LOGIN", "user@test.com")
-
-	testAppFactory := func(t testing.TB) *pbTests.TestApp {
-		hub, _ := beszelTests.NewTestHub(t.TempDir())
-		hubs = append(hubs, hub)
-		hub.StartHub()
-		return hub.TestApp
-	}
-
-	scenarios := []beszelTests.ApiScenario{
-		{
-			Name:            "GET /getkey - without auto login should fail",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/getkey",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:            "GET /getkey - with auto login should fail if no matching user",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/getkey",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:            "GET /getkey - with auto login should succeed",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/getkey",
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"key\":", "\"v\":"},
-			TestAppFactory:  testAppFactory,
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.CreateUser(app, "user@test.com", "password123")
-			},
-		},
-	}
-
-	for _, scenario := range scenarios {
-		scenario.Test(t)
-	}
-}
-
-func TestTrustedHeaderMiddleware(t *testing.T) {
-	var hubs []*beszelTests.TestHub
-
-	defer func() {
-		defer os.Unsetenv("TRUSTED_AUTH_HEADER")
-		for _, hub := range hubs {
-			hub.Cleanup()
-		}
-	}()
-
-	os.Setenv("TRUSTED_AUTH_HEADER", "X-Beszel-Trusted")
-
-	testAppFactory := func(t testing.TB) *pbTests.TestApp {
-		hub, _ := beszelTests.NewTestHub(t.TempDir())
-		hubs = append(hubs, hub)
-		hub.StartHub()
-		return hub.TestApp
-	}
-
-	scenarios := []beszelTests.ApiScenario{
-		{
-			Name:            "GET /getkey - without trusted header should fail",
-			Method:          http.MethodGet,
-			URL:             "/api/beszel/getkey",
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /getkey - with trusted header should fail if no matching user",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/getkey",
-			Headers: map[string]string{
-				"X-Beszel-Trusted": "user@test.com",
-			},
-			ExpectedStatus:  401,
-			ExpectedContent: []string{"requires valid"},
-			TestAppFactory:  testAppFactory,
-		},
-		{
-			Name:   "GET /getkey - with trusted header should succeed",
-			Method: http.MethodGet,
-			URL:    "/api/beszel/getkey",
-			Headers: map[string]string{
-				"X-Beszel-Trusted": "user@test.com",
-			},
-			ExpectedStatus:  200,
-			ExpectedContent: []string{"\"key\":", "\"v\":"},
-			TestAppFactory:  testAppFactory,
-			BeforeTestFunc: func(t testing.TB, app *pbTests.TestApp, e *core.ServeEvent) {
-				beszelTests.CreateUser(app, "user@test.com", "password123")
-			},
-		},
-	}
-
-	for _, scenario := range scenarios {
-		scenario.Test(t)
-	}
-}

internal/hub/hub_test_helpers.go

@@ -2,7 +2,9 @@
 
 package hub
 
-import "github.com/henrygd/beszel/internal/hub/systems"
+import (
+	"github.com/henrygd/beszel/internal/hub/systems"
+)
 
 // TESTING ONLY: GetSystemManager returns the system manager
 func (h *Hub) GetSystemManager() *systems.SystemManager {
@@ -18,3 +20,7 @@ func (h *Hub) GetPubkey() string {
 func (h *Hub) SetPubkey(pubkey string) {
 	h.pubKey = pubkey
 }
+
+func (h *Hub) SetCollectionAuthSettings() error {
+	return setCollectionAuthSettings(h)
+}

internal/hub/server.go

@@ -0,0 +1,42 @@
+package hub
+
+import (
+	"encoding/json"
+	"net/url"
+	"strings"
+
+	"github.com/henrygd/beszel"
+	"github.com/henrygd/beszel/internal/hub/utils"
+)
+
+// PublicAppInfo defines the structure of the public app information that will be injected into the HTML
+type PublicAppInfo struct {
+	BASE_PATH           string
+	HUB_VERSION         string
+	HUB_URL             string
+	OAUTH_DISABLE_POPUP bool `json:"OAUTH_DISABLE_POPUP,omitempty"`
+}
+
+// modifyIndexHTML injects the public app information into the index.html content
+func modifyIndexHTML(hub *Hub, html []byte) string {
+	info := getPublicAppInfo(hub)
+	content, err := json.Marshal(info)
+	if err != nil {
+		return string(html)
+	}
+	htmlContent := strings.ReplaceAll(string(html), "./", info.BASE_PATH)
+	return strings.Replace(htmlContent, "\"{info}\"", string(content), 1)
+}
+
+func getPublicAppInfo(hub *Hub) PublicAppInfo {
+	parsedURL, _ := url.Parse(hub.appURL)
+	info := PublicAppInfo{
+		BASE_PATH:   strings.TrimSuffix(parsedURL.Path, "/") + "/",
+		HUB_VERSION: beszel.Version,
+		HUB_URL:     hub.appURL,
+	}
+	if val, _ := utils.GetEnv("OAUTH_DISABLE_POPUP"); val == "true" {
+		info.OAUTH_DISABLE_POPUP = true
+	}
+	return info
+}

internal/hub/server_development.go

@@ -5,14 +5,11 @@ package hub
 import (
 	"fmt"
 	"io"
-	"log/slog"
 	"net/http"
 	"net/http/httputil"
 	"net/url"
 	"strings"
 
-	"github.com/henrygd/beszel"
-
 	"github.com/pocketbase/pocketbase/core"
 	"github.com/pocketbase/pocketbase/tools/osutils"
 )
@@ -39,7 +36,7 @@ func (rm *responseModifier) RoundTrip(req *http.Request) (*http.Response, error)
 	}
 	resp.Body.Close()
 	// Create a new response with the modified body
-	modifiedBody := rm.modifyHTML(string(body))
+	modifiedBody := modifyIndexHTML(rm.hub, body)
 	resp.Body = io.NopCloser(strings.NewReader(modifiedBody))
 	resp.ContentLength = int64(len(modifiedBody))
 	resp.Header.Set("Content-Length", fmt.Sprintf("%d", len(modifiedBody)))
@@ -47,22 +44,8 @@ func (rm *responseModifier) RoundTrip(req *http.Request) (*http.Response, error)
 	return resp, nil
 }
 
-func (rm *responseModifier) modifyHTML(html string) string {
-	parsedURL, err := url.Parse(rm.hub.appURL)
-	if err != nil {
-		return html
-	}
-	// fix base paths in html if using subpath
-	basePath := strings.TrimSuffix(parsedURL.Path, "/") + "/"
-	html = strings.ReplaceAll(html, "./", basePath)
-	html = strings.Replace(html, "{{V}}", beszel.Version, 1)
-	html = strings.Replace(html, "{{HUB_URL}}", rm.hub.appURL, 1)
-	return html
-}
-
 // startServer sets up the development server for Beszel
 func (h *Hub) startServer(se *core.ServeEvent) error {
-	slog.Info("starting server", "appURL", h.appURL)
 	proxy := httputil.NewSingleHostReverseProxy(&url.URL{
 		Scheme: "http",
 		Host:   "localhost:5173",

internal/hub/server_production.go

@@ -5,10 +5,9 @@ package hub
 import (
 	"io/fs"
 	"net/http"
-	"net/url"
 	"strings"
 
-	"github.com/henrygd/beszel"
+	"github.com/henrygd/beszel/internal/hub/utils"
 	"github.com/henrygd/beszel/internal/site"
 
 	"github.com/pocketbase/pocketbase/apis"
@@ -17,22 +16,13 @@ import (
 
 // startServer sets up the production server for Beszel
 func (h *Hub) startServer(se *core.ServeEvent) error {
-	// parse app url
-	parsedURL, err := url.Parse(h.appURL)
-	if err != nil {
-		return err
-	}
-	// fix base paths in html if using subpath
-	basePath := strings.TrimSuffix(parsedURL.Path, "/") + "/"
 	indexFile, _ := fs.ReadFile(site.DistDirFS, "index.html")
-	html := strings.ReplaceAll(string(indexFile), "./", basePath)
-	html = strings.Replace(html, "{{V}}", beszel.Version, 1)
-	html = strings.Replace(html, "{{HUB_URL}}", h.appURL, 1)
+	html := modifyIndexHTML(h, indexFile)
 	// set up static asset serving
 	staticPaths := [2]string{"/static/", "/assets/"}
 	serveStatic := apis.Static(site.DistDirFS, false)
 	// get CSP configuration
-	csp, cspExists := GetEnv("CSP")
+	csp, cspExists := utils.GetEnv("CSP")
 	// add route
 	se.Router.GET("/{path...}", func(e *core.RequestEvent) error {
 		// serve static assets if path is in staticPaths

internal/hub/systems/ssh_timeout_test.go

@@ -0,0 +1,56 @@
+//go:build testing
+
+package systems
+
+import (
+	"errors"
+	"testing"
+	"testing/synctest"
+	"time"
+
+	"github.com/stretchr/testify/assert"
+)
+
+// TestRunWithTimeout covers the guard added for issue #2041: the per-system SSH
+// data exchange must never block the updater indefinitely on a dead connection.
+func TestRunWithTimeout(t *testing.T) {
+	t.Run("returns the operation result when it completes before the timeout", func(t *testing.T) {
+		synctest.Test(t, func(t *testing.T) {
+			wantErr := errors.New("boom")
+			onTimeoutCalled := false
+
+			retry, err := runWithTimeout(10*time.Second, func() (bool, error) {
+				return true, wantErr
+			}, func() { onTimeoutCalled = true })
+
+			assert.True(t, retry, "should return the operation's retry value")
+			assert.Equal(t, wantErr, err, "should return the operation's error")
+			assert.False(t, onTimeoutCalled, "onTimeout must not fire when the op completes")
+		})
+	})
+
+	t.Run("times out and tears down the connection when the op blocks", func(t *testing.T) {
+		synctest.Test(t, func(t *testing.T) {
+			// unblock simulates a half-open connection: the op is stuck reading a
+			// response that never arrives until the connection is torn down.
+			unblock := make(chan struct{})
+			onTimeoutCalled := false
+			start := time.Now()
+
+			retry, err := runWithTimeout(5*time.Second, func() (bool, error) {
+				<-unblock
+				return false, nil
+			}, func() {
+				onTimeoutCalled = true
+				close(unblock) // tearing down the connection releases the blocked read
+			})
+
+			assert.Equal(t, 5*time.Second, time.Since(start), "should return exactly at the timeout")
+			assert.True(t, retry, "a timeout should be retryable so the next tick re-dials")
+			assert.Error(t, err, "a timeout must surface an error so the system is set down")
+			assert.True(t, onTimeoutCalled, "onTimeout must fire so the dead connection is closed")
+
+			synctest.Wait() // ensure the released op goroutine exits cleanly
+		})
+	})
+}

internal/hub/systems/system.go

@@ -14,6 +14,7 @@ import (
 
 	"github.com/henrygd/beszel/internal/common"
 	"github.com/henrygd/beszel/internal/hub/transport"
+	"github.com/henrygd/beszel/internal/hub/utils"
 	"github.com/henrygd/beszel/internal/hub/ws"
 
 	"github.com/henrygd/beszel/internal/entities/container"
@@ -145,6 +146,7 @@ func (sys *System) update() error {
 		// update smart interval if it's set on the agent side
 		if data.Details.SmartInterval > 0 {
 			sys.smartInterval = data.Details.SmartInterval
+			sys.manager.hub.Logger().Info("SMART interval updated from agent details", "system", sys.Id, "interval", sys.smartInterval.String())
 			// 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)
@@ -156,11 +158,10 @@ func (sys *System) update() error {
 		if sys.smartInterval <= 0 {
 			sys.smartInterval = time.Hour
 		}
-		lastFetch, _ := sys.manager.smartFetchMap.GetOk(sys.Id)
-		if time.Since(time.UnixMilli(lastFetch-1e4)) >= sys.smartInterval && sys.smartFetching.CompareAndSwap(false, true) {
+		if sys.shouldFetchSmart() && sys.smartFetching.CompareAndSwap(false, true) {
+			sys.manager.hub.Logger().Info("SMART fetch", "system", sys.Id, "interval", sys.smartInterval.String())
 			go func() {
 				defer sys.smartFetching.Store(false)
-				sys.manager.smartFetchMap.Set(sys.Id, time.Now().UnixMilli(), sys.smartInterval+time.Minute)
 				_ = sys.FetchAndSaveSmartDevices()
 			}()
 		}
@@ -184,7 +185,7 @@ func (sys *System) handlePaused() {
 
 // createRecords updates the system record and adds system_stats and container_stats records
 func (sys *System) createRecords(data *system.CombinedData) (*core.Record, error) {
-	systemRecord, err := sys.getRecord()
+	systemRecord, err := sys.getRecord(sys.manager.hub)
 	if err != nil {
 		return nil, err
 	}
@@ -318,10 +319,11 @@ func createContainerRecords(app core.App, data []*container.Stats, systemId stri
 	valueStrings := make([]string, 0, len(data))
 	for i, container := range data {
 		suffix := fmt.Sprintf("%d", i)
-		valueStrings = append(valueStrings, fmt.Sprintf("({:id%[1]s}, {:system}, {:name%[1]s}, {:image%[1]s}, {:status%[1]s}, {:health%[1]s}, {:cpu%[1]s}, {:memory%[1]s}, {:net%[1]s}, {:updated})", suffix))
+		valueStrings = append(valueStrings, fmt.Sprintf("({:id%[1]s}, {:system}, {:name%[1]s}, {:image%[1]s}, {:ports%[1]s}, {:status%[1]s}, {:health%[1]s}, {:cpu%[1]s}, {:memory%[1]s}, {:net%[1]s}, {:updated})", suffix))
 		params["id"+suffix] = container.Id
 		params["name"+suffix] = container.Name
 		params["image"+suffix] = container.Image
+		params["ports"+suffix] = container.Ports
 		params["status"+suffix] = container.Status
 		params["health"+suffix] = container.Health
 		params["cpu"+suffix] = container.Cpu
@@ -333,7 +335,7 @@ func createContainerRecords(app core.App, data []*container.Stats, systemId stri
 		params["net"+suffix] = netBytes
 	}
 	queryString := fmt.Sprintf(
-		"INSERT INTO containers (id, system, name, image, status, health, cpu, memory, net, updated) VALUES %s ON CONFLICT(id) DO UPDATE SET system = excluded.system, name = excluded.name, image = excluded.image, status = excluded.status, health = excluded.health, cpu = excluded.cpu, memory = excluded.memory, net = excluded.net, updated = excluded.updated",
+		"INSERT INTO containers (id, system, name, image, ports, status, health, cpu, memory, net, updated) VALUES %s ON CONFLICT(id) DO UPDATE SET system = excluded.system, name = excluded.name, image = excluded.image, ports = excluded.ports, status = excluded.status, health = excluded.health, cpu = excluded.cpu, memory = excluded.memory, net = excluded.net, updated = excluded.updated",
 		strings.Join(valueStrings, ","),
 	)
 	_, err := app.DB().NewQuery(queryString).Bind(params).Execute()
@@ -342,8 +344,8 @@ func createContainerRecords(app core.App, data []*container.Stats, systemId stri
 
 // getRecord retrieves the system record from the database.
 // If the record is not found, it removes the system from the manager.
-func (sys *System) getRecord() (*core.Record, error) {
-	record, err := sys.manager.hub.FindRecordById("systems", sys.Id)
+func (sys *System) getRecord(app core.App) (*core.Record, error) {
+	record, err := app.FindRecordById("systems", sys.Id)
 	if err != nil || record == nil {
 		_ = sys.manager.RemoveSystem(sys.Id)
 		return nil, err
@@ -351,6 +353,27 @@ func (sys *System) getRecord() (*core.Record, error) {
 	return record, nil
 }
 
+// HasUser checks if the given user is in the system's users list.
+// Returns true if SHARE_ALL_SYSTEMS is enabled (any authenticated user can access any system).
+func (sys *System) HasUser(app core.App, user *core.Record) bool {
+	if user == nil {
+		return false
+	}
+	if v, _ := utils.GetEnv("SHARE_ALL_SYSTEMS"); v == "true" {
+		return true
+	}
+	var recordData = struct {
+		Users string
+	}{}
+	err := app.DB().NewQuery("SELECT users FROM systems WHERE id={:id}").
+		Bind(dbx.Params{"id": sys.Id}).
+		One(&recordData)
+	if err != nil || recordData.Users == "" {
+		return false
+	}
+	return strings.Contains(recordData.Users, user.Id)
+}
+
 // setDown marks a system as down in the database.
 // It takes the original error that caused the system to go down and returns any error
 // encountered during the process of updating the system status.
@@ -358,13 +381,14 @@ func (sys *System) setDown(originalError error) error {
 	if sys.Status == down || sys.Status == paused {
 		return nil
 	}
-	record, err := sys.getRecord()
+	record, err := sys.getRecord(sys.manager.hub)
 	if err != nil {
 		return err
 	}
 	if originalError != nil {
 		sys.manager.hub.Logger().Error("System down", "system", record.GetString("name"), "err", originalError)
 	}
+	sys.detailsFetched.Store(false)
 	record.Set("status", down)
 	return sys.manager.hub.SaveNoValidate(record)
 }
@@ -508,11 +532,16 @@ func (sys *System) FetchSystemdInfoFromAgent(serviceName string) (systemd.Servic
 	return result, err
 }
 
-// FetchSmartDataFromAgent fetches SMART data from the agent
-func (sys *System) FetchSmartDataFromAgent() (map[string]smart.SmartData, error) {
+// FetchSmartDataFromAgent fetches SMART data from the agent.
+func (sys *System) FetchSmartDataFromAgent() (smart.SmartDataResponse, error) {
 	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
 	defer cancel()
-	var result map[string]smart.SmartData
+	if sys.agentVersion.LT(beszel.MinVersionAgentResponse) {
+		var data map[string]smart.SmartData
+		err := sys.request(ctx, common.GetSmartData, nil, &data)
+		return smart.SmartDataResponse{Data: data}, err
+	}
+	var result smart.SmartDataResponse
 	err := sys.request(ctx, common.GetSmartData, nil, &result)
 	return result, err
 }
@@ -600,10 +629,17 @@ func (sys *System) runSSHOperation(timeout time.Duration, retries int, operation
 			continue
 		}
 
-		retry, opErr := func() (bool, error) {
+		// Bound the whole operation. A half-open TCP connection (a dead peer that
+		// never sends RST/FIN) or a wedged agent that accepts the session but
+		// never writes a response would otherwise block the read forever. Because
+		// StartUpdater runs update() synchronously on its ticker, that stalls the
+		// per-system updater indefinitely with no error and no re-dial until the
+		// hub is restarted (issue #2041). On timeout we tear down the connection
+		// so the blocked read unwinds and the system is re-dialed on the next tick.
+		retry, opErr := runWithTimeout(sshOperationTimeout, func() (bool, error) {
 			defer session.Close()
 			return operation(session)
-		}()
+		}, sys.closeSSHConnection)
 
 		if opErr == nil {
 			return nil
@@ -622,6 +658,43 @@ func (sys *System) runSSHOperation(timeout time.Duration, retries int, operation
 	return fmt.Errorf("ssh operation failed")
 }
 
+// sshOperationTimeout bounds a single SSH data exchange (send request, read
+// response, wait for the remote command to exit). It is more generous than the
+// session-creation timeout to tolerate briefly slow agents, but is kept well
+// under the collection interval so a stalled connection is detected and
+// re-dialed within one cycle (see issue #2041).
+const sshOperationTimeout = 20 * time.Second
+
+// runWithTimeout runs op in a goroutine and returns its result, or, if op does
+// not finish within timeout, calls onTimeout (used to tear down the connection
+// so a blocked op can unwind) and returns a retryable timeout error. This
+// guarantees the caller can never block indefinitely on a dead SSH connection.
+func runWithTimeout(timeout time.Duration, op func() (bool, error), onTimeout func()) (retry bool, err error) {
+	type opResult struct {
+		retry bool
+		err   error
+	}
+	// Buffered so the op goroutine never leaks even when we return on timeout.
+	done := make(chan opResult, 1)
+	go func() {
+		r, e := op()
+		done <- opResult{retry: r, err: e}
+	}()
+
+	timer := time.NewTimer(timeout)
+	defer timer.Stop()
+
+	select {
+	case res := <-done:
+		return res.retry, res.err
+	case <-timer.C:
+		if onTimeout != nil {
+			onTimeout()
+		}
+		return true, fmt.Errorf("ssh operation timed out after %s", timeout)
+	}
+}
+
 // createSSHClient creates a new SSH client for the system
 func (s *System) createSSHClient() error {
 	if s.manager.sshConfig == nil {
@@ -637,14 +710,43 @@ func (s *System) createSSHClient() error {
 		host = net.JoinHostPort(host, s.Port)
 	}
 	var err error
-	s.client, err = ssh.Dial(network, host, s.manager.sshConfig)
+	s.client, err = dialSSHWithKeepAlive(network, host, s.manager.sshConfig)
 	if err != nil {
 		return err
 	}
 	s.agentVersion, _ = extractAgentVersion(string(s.client.Conn.ServerVersion()))
+	s.manager.resetFailedSmartFetchState(s.Id)
 	return nil
 }
 
+// sshKeepAliveInterval is the TCP keep-alive idle interval for SSH connections
+// to agents. Enabling OS-level keep-alives lets the hub eventually detect a
+// dead peer on an otherwise idle connection instead of trusting it forever.
+// This is a backstop for genuine network death; an application-level wedge
+// (agent process hung while its kernel keeps ACKing) is caught by the
+// per-operation timeout in runSSHOperation instead (see issue #2041).
+const sshKeepAliveInterval = 30 * time.Second
+
+// dialSSHWithKeepAlive dials an SSH connection like ssh.Dial, but enables TCP
+// keep-alive on the underlying connection so half-open connections are
+// eventually detected by the operating system.
+func dialSSHWithKeepAlive(network, addr string, config *ssh.ClientConfig) (*ssh.Client, error) {
+	dialer := net.Dialer{
+		Timeout:   config.Timeout,
+		KeepAlive: sshKeepAliveInterval,
+	}
+	conn, err := dialer.Dial(network, addr)
+	if err != nil {
+		return nil, err
+	}
+	sshConn, chans, reqs, err := ssh.NewClientConn(conn, addr, config)
+	if err != nil {
+		_ = conn.Close()
+		return nil, err
+	}
+	return ssh.NewClient(sshConn, chans, reqs), nil
+}
+
 // createSessionWithTimeout creates a new SSH session with a timeout to avoid hanging
 // in case of network issues
 func (sys *System) createSessionWithTimeout(timeout time.Duration) (*ssh.Session, error) {

internal/hub/systems/system_manager.go

@@ -41,10 +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
-	smartFetchMap *expirymap.ExpiryMap[int64]   // Stores last SMART fetch time per system ID
+	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[smartFetchState] // Stores last SMART fetch time/result; TTL is only for cleanup
 }
 
 // hubLike defines the interface requirements for the hub dependency.
@@ -54,6 +54,7 @@ type hubLike interface {
 	GetSSHKey(dataDir string) (ssh.Signer, error)
 	HandleSystemAlerts(systemRecord *core.Record, data *system.CombinedData) error
 	HandleStatusAlerts(status string, systemRecord *core.Record) error
+	CancelPendingStatusAlerts(systemID string)
 }
 
 // NewSystemManager creates a new SystemManager instance with the provided hub.
@@ -62,7 +63,7 @@ func NewSystemManager(hub hubLike) *SystemManager {
 	return &SystemManager{
 		systems:       store.New(map[string]*System{}),
 		hub:           hub,
-		smartFetchMap: expirymap.New[int64](time.Hour),
+		smartFetchMap: expirymap.New[smartFetchState](time.Hour),
 	}
 }
 
@@ -189,6 +190,7 @@ func (sm *SystemManager) onRecordAfterUpdateSuccess(e *core.RecordEvent) error {
 			system.closeSSHConnection()
 		}
 		_ = deactivateAlerts(e.App, e.Record.Id)
+		sm.hub.CancelPendingStatusAlerts(e.Record.Id)
 		return e.Next()
 	case pending:
 		// Resume monitoring, preferring existing WebSocket connection
@@ -306,6 +308,7 @@ func (sm *SystemManager) AddWebSocketSystem(systemId string, agentVersion semver
 	if err != nil {
 		return err
 	}
+	sm.resetFailedSmartFetchState(systemId)
 
 	system := sm.NewSystem(systemId)
 	system.WsConn = wsConn
@@ -317,6 +320,15 @@ func (sm *SystemManager) AddWebSocketSystem(systemId string, agentVersion semver
 	return nil
 }
 
+// resetFailedSmartFetchState clears only failed SMART cooldown entries so a fresh
+// agent reconnect retries SMART discovery immediately after configuration changes.
+func (sm *SystemManager) resetFailedSmartFetchState(systemID string) {
+	state, ok := sm.smartFetchMap.GetOk(systemID)
+	if ok && !state.Successful {
+		sm.smartFetchMap.Remove(systemID)
+	}
+}
+
 // createSSHClientConfig initializes the SSH client configuration for connecting to an agent's server
 func (sm *SystemManager) createSSHClientConfig() error {
 	privateKey, err := sm.hub.GetSSHKey("")

internal/hub/systems/system_realtime.go

@@ -19,7 +19,6 @@ type subscriptionInfo struct {
 var (
 	activeSubscriptions = make(map[string]*subscriptionInfo)
 	workerRunning       bool
-	realtimeTicker      *time.Ticker
 	tickerStopChan      chan struct{}
 	realtimeMutex       sync.Mutex
 )
@@ -70,7 +69,7 @@ func (sm *SystemManager) onRealtimeSubscribeRequest(e *core.RealtimeSubscribeReq
 }
 
 // onRealtimeSubscriptionAdded initializes or starts the realtime worker when the first subscription is added.
-// It ensures only one worker runs at a time and creates the ticker for periodic data fetching.
+// It ensures only one worker runs at a time.
 func (sm *SystemManager) onRealtimeSubscriptionAdded() {
 	realtimeMutex.Lock()
 	defer realtimeMutex.Unlock()
@@ -82,11 +81,6 @@ func (sm *SystemManager) onRealtimeSubscriptionAdded() {
 		tickerStopChan = make(chan struct{})
 		go sm.startRealtimeWorker()
 	}
-
-	// If no ticker exists, create one
-	if realtimeTicker == nil {
-		realtimeTicker = time.NewTicker(1 * time.Second)
-	}
 }
 
 // checkSubscriptions stops the realtime worker when there are no active subscriptions.
@@ -107,11 +101,6 @@ func (sm *SystemManager) checkSubscriptions() {
 		}
 	}
 
-	if realtimeTicker != nil {
-		realtimeTicker.Stop()
-		realtimeTicker = nil
-	}
-
 	// Mark worker as stopped (will be reset when next subscription comes in)
 	workerRunning = false
 }
@@ -135,17 +124,16 @@ func (sm *SystemManager) removeRealtimeSubscription(subscription string, options
 // It continuously fetches system data and broadcasts it to subscribed clients via WebSocket.
 func (sm *SystemManager) startRealtimeWorker() {
 	sm.fetchRealtimeDataAndNotify()
+	tick := time.Tick(1 * time.Second)
 
 	for {
 		select {
 		case <-tickerStopChan:
 			return
-		case <-realtimeTicker.C:
-			// Check if ticker is still valid (might have been stopped)
-			if realtimeTicker == nil || len(activeSubscriptions) == 0 {
+		case <-tick:
+			if len(activeSubscriptions) == 0 {
 				return
 			}
-			// slog.Debug("activeSubscriptions", "count", len(activeSubscriptions))
 			sm.fetchRealtimeDataAndNotify()
 		}
 	}

internal/hub/systems/system_smart.go

@@ -4,22 +4,67 @@ import (
 	"database/sql"
 	"errors"
 	"strings"
+	"time"
 
 	"github.com/henrygd/beszel/internal/entities/smart"
+	"github.com/pocketbase/dbx"
 	"github.com/pocketbase/pocketbase/core"
 )
 
-// FetchAndSaveSmartDevices fetches SMART data from the agent and saves it to the database
-func (sys *System) FetchAndSaveSmartDevices() error {
-	smartData, err := sys.FetchSmartDataFromAgent()
-	if err != nil || len(smartData) == 0 {
-		return err
-	}
-	return sys.saveSmartDevices(smartData)
+type smartFetchState struct {
+	LastAttempt int64
+	Successful  bool
 }
 
-// saveSmartDevices saves SMART device data to the smart_devices collection
-func (sys *System) saveSmartDevices(smartData map[string]smart.SmartData) error {
+// FetchAndSaveSmartDevices fetches SMART data from the agent and saves it to the database
+func (sys *System) FetchAndSaveSmartDevices() error {
+	response, err := sys.FetchSmartDataFromAgent()
+	if err != nil {
+		sys.recordSmartFetchResult(err, 0)
+		return err
+	}
+	err = sys.saveSmartDevices(response.Data, response.Complete)
+	sys.recordSmartFetchResult(err, len(response.Data))
+	return err
+}
+
+// recordSmartFetchResult stores a cooldown entry for the SMART interval and marks
+// whether the last fetch produced any devices, so failed setup can retry on reconnect.
+func (sys *System) recordSmartFetchResult(err error, deviceCount int) {
+	if sys.manager == nil {
+		return
+	}
+	interval := sys.smartFetchInterval()
+	success := err == nil && deviceCount > 0
+	if sys.manager.hub != nil {
+		sys.manager.hub.Logger().Info("SMART fetch result", "system", sys.Id, "success", success, "devices", deviceCount, "interval", interval.String(), "err", err)
+	}
+	sys.manager.smartFetchMap.Set(sys.Id, smartFetchState{LastAttempt: time.Now().UnixMilli(), Successful: success}, interval+time.Minute)
+}
+
+// shouldFetchSmart returns true when there is no active SMART cooldown entry for this system.
+func (sys *System) shouldFetchSmart() bool {
+	if sys.manager == nil {
+		return true
+	}
+	state, ok := sys.manager.smartFetchMap.GetOk(sys.Id)
+	if !ok {
+		return true
+	}
+	return !time.UnixMilli(state.LastAttempt).Add(sys.smartFetchInterval()).After(time.Now())
+}
+
+// smartFetchInterval returns the agent-provided SMART interval or the default when unset.
+func (sys *System) smartFetchInterval() time.Duration {
+	if sys.smartInterval > 0 {
+		return sys.smartInterval
+	}
+	return time.Hour
+}
+
+// saveSmartDevices saves SMART device data and, after a complete refresh,
+// removes rows for devices that are no longer reported.
+func (sys *System) saveSmartDevices(smartData map[string]smart.SmartData, complete bool) error {
 	if len(smartData) == 0 {
 		return nil
 	}
@@ -30,20 +75,49 @@ func (sys *System) saveSmartDevices(smartData map[string]smart.SmartData) error
 		return err
 	}
 
-	for deviceKey, device := range smartData {
-		if err := sys.upsertSmartDeviceRecord(collection, deviceKey, device); err != nil {
-			return err
-		}
+	currentIDs := make(map[string]struct{}, len(smartData))
+	for deviceKey := range smartData {
+		currentIDs[makeStableHashId(sys.Id, deviceKey)] = struct{}{}
 	}
 
-	return nil
+	err = hub.RunInTransaction(func(txApp core.App) error {
+		if complete {
+			existing, err := txApp.FindRecordsByFilter(
+				collection,
+				"system = {:system}",
+				"",
+				0,
+				0,
+				dbx.Params{"system": sys.Id},
+			)
+			if err != nil {
+				return err
+			}
+			for _, record := range existing {
+				if _, ok := currentIDs[record.Id]; ok {
+					continue
+				}
+				if err := txApp.Delete(record); err != nil {
+					return err
+				}
+			}
+		}
+
+		for deviceKey, device := range smartData {
+			if err := sys.upsertSmartDeviceRecord(txApp, collection, deviceKey, device); err != nil {
+				return err
+			}
+		}
+
+		return nil
+	})
+	return err
 }
 
-func (sys *System) upsertSmartDeviceRecord(collection *core.Collection, deviceKey string, device smart.SmartData) error {
-	hub := sys.manager.hub
+func (sys *System) upsertSmartDeviceRecord(app core.App, collection *core.Collection, deviceKey string, device smart.SmartData) error {
 	recordID := makeStableHashId(sys.Id, deviceKey)
 
-	record, err := hub.FindRecordById(collection, recordID)
+	record, err := app.FindRecordById(collection, recordID)
 	if err != nil {
 		if !errors.Is(err, sql.ErrNoRows) {
 			return err
@@ -71,7 +145,7 @@ func (sys *System) upsertSmartDeviceRecord(collection *core.Collection, deviceKe
 	record.Set("cycles", powerCycles)
 	record.Set("attributes", device.Attributes)
 
-	return hub.SaveNoValidate(record)
+	return app.SaveNoValidate(record)
 }
 
 // extractPowerMetrics extracts power on hours and power cycles from SMART attributes

internal/hub/systems/system_smart_test.go

@@ -0,0 +1,229 @@
+//go:build testing
+
+package systems
+
+import (
+	"errors"
+	"testing"
+	"time"
+
+	"github.com/henrygd/beszel/internal/entities/smart"
+	esystem "github.com/henrygd/beszel/internal/entities/system"
+	"github.com/henrygd/beszel/internal/hub/expirymap"
+	_ "github.com/henrygd/beszel/internal/migrations"
+
+	"github.com/pocketbase/pocketbase/core"
+	pbtests "github.com/pocketbase/pocketbase/tests"
+	"github.com/stretchr/testify/assert"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+// stubHub implements hubLike using a plain pocketbase test app, so
+// smart-device DB tests can run in-package without an import cycle to
+// internal/hub (which imports this package).
+type stubHub struct{ core.App }
+
+func (stubHub) GetSSHKey(dataDir string) (ssh.Signer, error) { return nil, nil }
+func (stubHub) HandleSystemAlerts(systemRecord *core.Record, data *esystem.CombinedData) error {
+	return nil
+}
+func (stubHub) HandleStatusAlerts(status string, systemRecord *core.Record) error { return nil }
+func (stubHub) CancelPendingStatusAlerts(systemID string)                         {}
+
+// newTestSystemWithHub creates a System backed by a real (temp) database, along
+// with a matching "systems" record, for tests that need to exercise DB reads/writes.
+func newTestSystemWithHub(t *testing.T) (*System, *pbtests.TestApp) {
+	t.Helper()
+	testApp, err := pbtests.NewTestApp(t.TempDir())
+	require.NoError(t, err)
+	t.Cleanup(testApp.Cleanup)
+
+	sm := &SystemManager{hub: stubHub{testApp}, smartFetchMap: expirymap.New[smartFetchState](time.Hour)}
+	t.Cleanup(sm.smartFetchMap.StopCleaner)
+
+	col, err := testApp.FindCachedCollectionByNameOrId("systems")
+	require.NoError(t, err)
+	systemRecord := core.NewRecord(col)
+	systemRecord.Set("name", "test-system")
+	systemRecord.Set("host", "127.0.0.1")
+	systemRecord.Set("port", "45876")
+	require.NoError(t, testApp.SaveNoValidate(systemRecord))
+
+	sys := &System{Id: systemRecord.Id, manager: sm}
+	return sys, testApp
+}
+
+func TestRecordSmartFetchResult(t *testing.T) {
+	sm := &SystemManager{smartFetchMap: expirymap.New[smartFetchState](time.Hour)}
+	t.Cleanup(sm.smartFetchMap.StopCleaner)
+
+	sys := &System{
+		Id:            "system-1",
+		manager:       sm,
+		smartInterval: time.Hour,
+	}
+
+	// Successful fetch with devices
+	sys.recordSmartFetchResult(nil, 5)
+	state, ok := sm.smartFetchMap.GetOk(sys.Id)
+	assert.True(t, ok, "expected smart fetch result to be stored")
+	assert.True(t, state.Successful, "expected successful fetch state to be recorded")
+
+	// Failed fetch
+	sys.recordSmartFetchResult(errors.New("failed"), 0)
+	state, ok = sm.smartFetchMap.GetOk(sys.Id)
+	assert.True(t, ok, "expected failed smart fetch state to be stored")
+	assert.False(t, state.Successful, "expected failed smart fetch state to be marked unsuccessful")
+
+	// Successful fetch but no devices
+	sys.recordSmartFetchResult(nil, 0)
+	state, ok = sm.smartFetchMap.GetOk(sys.Id)
+	assert.True(t, ok, "expected fetch with zero devices to be stored")
+	assert.False(t, state.Successful, "expected fetch with zero devices to be marked unsuccessful")
+}
+
+func TestShouldFetchSmart(t *testing.T) {
+	sm := &SystemManager{smartFetchMap: expirymap.New[smartFetchState](time.Hour)}
+	t.Cleanup(sm.smartFetchMap.StopCleaner)
+
+	sys := &System{
+		Id:            "system-1",
+		manager:       sm,
+		smartInterval: time.Hour,
+	}
+
+	assert.True(t, sys.shouldFetchSmart(), "expected initial smart fetch to be allowed")
+
+	sys.recordSmartFetchResult(errors.New("failed"), 0)
+	assert.False(t, sys.shouldFetchSmart(), "expected smart fetch to be blocked while interval entry exists")
+
+	sm.smartFetchMap.Remove(sys.Id)
+	assert.True(t, sys.shouldFetchSmart(), "expected smart fetch to be allowed after interval entry is cleared")
+}
+
+func TestShouldFetchSmart_IgnoresExtendedTTLWhenFetchIsDue(t *testing.T) {
+	sm := &SystemManager{smartFetchMap: expirymap.New[smartFetchState](time.Hour)}
+	t.Cleanup(sm.smartFetchMap.StopCleaner)
+
+	sys := &System{
+		Id:            "system-1",
+		manager:       sm,
+		smartInterval: time.Hour,
+	}
+
+	sm.smartFetchMap.Set(sys.Id, smartFetchState{
+		LastAttempt: time.Now().Add(-2 * time.Hour).UnixMilli(),
+		Successful:  true,
+	}, 10*time.Minute)
+	sm.smartFetchMap.UpdateExpiration(sys.Id, 3*time.Hour)
+
+	assert.True(t, sys.shouldFetchSmart(), "expected fetch time to take precedence over updated TTL")
+}
+
+func TestResetFailedSmartFetchState(t *testing.T) {
+	sm := &SystemManager{smartFetchMap: expirymap.New[smartFetchState](time.Hour)}
+	t.Cleanup(sm.smartFetchMap.StopCleaner)
+
+	sm.smartFetchMap.Set("system-1", smartFetchState{LastAttempt: time.Now().UnixMilli(), Successful: false}, time.Hour)
+	sm.resetFailedSmartFetchState("system-1")
+	_, ok := sm.smartFetchMap.GetOk("system-1")
+	assert.False(t, ok, "expected failed smart fetch state to be cleared on reconnect")
+
+	sm.smartFetchMap.Set("system-1", smartFetchState{LastAttempt: time.Now().UnixMilli(), Successful: true}, time.Hour)
+	sm.resetFailedSmartFetchState("system-1")
+	_, ok = sm.smartFetchMap.GetOk("system-1")
+	assert.True(t, ok, "expected successful smart fetch state to be preserved")
+}
+
+// countSmartDeviceRecords returns the number of smart_devices rows for the given system.
+func countSmartDeviceRecords(t *testing.T, app core.App, systemID string) []*core.Record {
+	t.Helper()
+	records, err := app.FindAllRecords("smart_devices", nil)
+	require.NoError(t, err)
+	var forSystem []*core.Record
+	for _, r := range records {
+		if r.GetString("system") == systemID {
+			forSystem = append(forSystem, r)
+		}
+	}
+	return forSystem
+}
+
+func TestSaveSmartDevices_RemovesStaleDevices(t *testing.T) {
+	sys, testApp := newTestSystemWithHub(t)
+
+	// first fetch reports two devices: sda (serial AAA) and sdb (serial BBB)
+	err := sys.saveSmartDevices(map[string]smart.SmartData{
+		"AAA": {SerialNumber: "AAA", DiskName: "sda", ModelName: "Disk A"},
+		"BBB": {SerialNumber: "BBB", DiskName: "sdb", ModelName: "Disk B"},
+	}, true)
+	require.NoError(t, err)
+
+	records := countSmartDeviceRecords(t, testApp, sys.Id)
+	require.Len(t, records, 2, "expected both devices to be saved")
+
+	var recordA *core.Record
+	for _, r := range records {
+		if r.GetString("serial") == "AAA" {
+			recordA = r
+		}
+	}
+	require.NotNil(t, recordA, "expected to find device AAA")
+	originalID := recordA.Id
+
+	deleteEvents := 0
+	testApp.OnRecordAfterDeleteSuccess("smart_devices").BindFunc(func(e *core.RecordEvent) error {
+		deleteEvents++
+		return e.Next()
+	})
+
+	// A complete refresh confirms that BBB is gone, so remove it through
+	// PocketBase and notify realtime subscribers.
+	err = sys.saveSmartDevices(map[string]smart.SmartData{
+		"AAA": {SerialNumber: "AAA", DiskName: "sda", ModelName: "Disk A", Temperature: 42},
+	}, true)
+	require.NoError(t, err)
+
+	records = countSmartDeviceRecords(t, testApp, sys.Id)
+	require.Len(t, records, 1, "expected stale device BBB to be removed")
+	assert.Equal(t, "AAA", records[0].GetString("serial"))
+	assert.Equal(t, originalID, records[0].Id, "expected existing device to be updated in place, not recreated")
+	assert.EqualValues(t, 42, records[0].GetInt("temp"))
+	assert.Equal(t, 1, deleteEvents, "expected PocketBase delete hooks to run")
+}
+
+func TestSaveSmartDevices_IncompleteDataDoesNotRemoveDevices(t *testing.T) {
+	sys, testApp := newTestSystemWithHub(t)
+
+	require.NoError(t, sys.saveSmartDevices(map[string]smart.SmartData{
+		"AAA": {SerialNumber: "AAA", DiskName: "sda"},
+		"BBB": {SerialNumber: "BBB", DiskName: "sdb"},
+	}, true))
+
+	// AAA was collected but BBB failed. The response is useful for updating AAA,
+	// but it is not authoritative enough to remove BBB.
+	require.NoError(t, sys.saveSmartDevices(map[string]smart.SmartData{
+		"AAA": {SerialNumber: "AAA", DiskName: "sda", Temperature: 42},
+	}, false))
+
+	assert.Len(t, countSmartDeviceRecords(t, testApp, sys.Id), 2)
+	recordA, err := testApp.FindRecordById("smart_devices", makeStableHashId(sys.Id, "AAA"))
+	require.NoError(t, err)
+	assert.EqualValues(t, 42, recordA.GetInt("temp"))
+}
+
+func TestSaveSmartDevices_EmptyDataIsNoop(t *testing.T) {
+	sys, testApp := newTestSystemWithHub(t)
+
+	err := sys.saveSmartDevices(map[string]smart.SmartData{
+		"AAA": {SerialNumber: "AAA", DiskName: "sda"},
+	}, true)
+	require.NoError(t, err)
+
+	err = sys.saveSmartDevices(map[string]smart.SmartData{}, true)
+	require.NoError(t, err)
+
+	records := countSmartDeviceRecords(t, testApp, sys.Id)
+	assert.Len(t, records, 1, "empty fetch result should not delete existing devices")
+}