New translations en.po (Japanese)

[ci skip]

agent/docker.go

@@ -68,10 +68,11 @@ type dockerManager struct {
 	excludeContainers    []string                    // Patterns to exclude containers by name
 	usingPodman          bool                        // Whether the Docker Engine API is running on Podman
 
-	registryClient      *http.Client                  // Client for registry requests; nil uses a client with a 10-second timeout
-	imageUpdatesMutex   sync.RWMutex                  // Protects imageUpdates, its entries, and imageUpdatesRunning
-	imageUpdates        map[string]*imageUpdateStatus // Shared update status keyed by normalized image reference
-	imageUpdatesRunning bool                          // Whether a background image-update batch is in progress
+	registryClient       *http.Client                  // Client for registry requests; nil uses a client with a 10-second timeout
+	imageUpdatesDisabled bool                          // Whether image update checks are disabled by configuration
+	imageUpdatesMutex    sync.RWMutex                  // Protects imageUpdates, its entries, and imageUpdatesRunning
+	imageUpdates         map[string]*imageUpdateStatus // Shared update status keyed by normalized image reference
+	imageUpdatesRunning  bool                          // Whether a background image-update batch is in progress
 
 	// Cache-time-aware tracking for CPU stats (similar to cpu.go)
 	// Maps cache time intervals to container-specific CPU usage tracking
@@ -688,6 +689,8 @@ func newDockerManager(agent *Agent) *dockerManager {
 		userAgent: "Docker-Client/",
 	}
 
+	dockerImageCheck, _ := utils.GetEnv("DOCKER_IMAGE_CHECK")
+
 	// Read container exclusion patterns from environment variable
 	var excludeContainers []string
 	if excludeStr, set := utils.GetEnv("EXCLUDE_CONTAINERS"); set && excludeStr != "" {
@@ -707,10 +710,11 @@ func newDockerManager(agent *Agent) *dockerManager {
 			Timeout:   timeout,
 			Transport: userAgentTransport,
 		},
-		containerStatsMap: make(map[string]*container.Stats),
-		sem:               make(chan struct{}, 5),
-		apiContainerList:  []*container.ApiInfo{},
-		excludeContainers: excludeContainers,
+		containerStatsMap:    make(map[string]*container.Stats),
+		sem:                  make(chan struct{}, 5),
+		apiContainerList:     []*container.ApiInfo{},
+		excludeContainers:    excludeContainers,
+		imageUpdatesDisabled: dockerImageCheck == "false",
 
 		// Initialize cache-time-aware tracking structures
 		lastCpuContainer:    make(map[uint16]map[string]uint64),

agent/docker_image_updates.go

@@ -31,6 +31,9 @@ func normalizedImageReference(image string) string {
 // refreshImageUpdates starts at most one background batch. Neither its network
 // work nor its completion is part of the container metrics wait group.
 func (dm *dockerManager) refreshImageUpdates(containers []*container.ApiInfo, now time.Time) {
+	if dm.imageUpdatesDisabled {
+		return
+	}
 	dm.imageUpdatesMutex.Lock()
 	defer dm.imageUpdatesMutex.Unlock()
 	if dm.imageUpdatesRunning {

agent/docker_image_updates_test.go

@@ -27,6 +27,29 @@ func waitForImageUpdates(t *testing.T, dm *dockerManager) {
 	}, time.Second*3, time.Millisecond)
 }
 
+func TestDisableDockerImageUpdateCheck(t *testing.T) {
+	t.Setenv("BESZEL_AGENT_DOCKER_IMAGE_CHECK", "false")
+	server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if r.URL.Path == "/version" {
+			fmt.Fprint(w, `{"Version":"25.0.0"}`)
+			return
+		}
+		http.NotFound(w, r)
+	}))
+	defer server.Close()
+	t.Setenv("BESZEL_AGENT_DOCKER_HOST", server.URL)
+
+	dm := newDockerManager(nil)
+	require.True(t, dm.imageUpdatesDisabled)
+	dm.registryClient = &http.Client{Transport: roundTripFunc(func(*http.Request) (*http.Response, error) {
+		t.Fatal("disabled image update check made a registry request")
+		return nil, nil
+	})}
+	dm.refreshImageUpdates([]*container.ApiInfo{{Image: "nginx", Names: []string{"/nginx"}}}, time.Now())
+	require.False(t, dm.imageUpdatesRunning)
+	require.Nil(t, dm.imageUpdates)
+}
+
 func TestImageUpdateCacheAndStats(t *testing.T) {
 	local := "sha256:" + strings.Repeat("a", 64)
 	remote := "sha256:" + strings.Repeat("b", 64)

agent/network_monitor_probe.go

@@ -8,9 +8,12 @@ import (
 	"net/http"
 	"time"
 
+	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/internal/entities/monitor"
 )
 
+const networkMonitorUserAgent = "Beszel-Agent/" + beszel.Version + " (+https://beszel.dev)"
+
 // monitorProbe performs one check. Errors are recorded as loss by the task runner.
 // Implementations must honor cancellation and bound their execution time.
 type monitorProbe func(context.Context, monitor.Config) (int64, error)
@@ -93,6 +96,7 @@ func monitorHTTP(ctx context.Context, client *http.Client, url string) (int64, e
 	if err != nil {
 		return -1, err
 	}
+	req.Header.Set("User-Agent", networkMonitorUserAgent)
 	resp, err := client.Do(req)
 	if err != nil {
 		return -1, err

agent/network_monitor_test.go

@@ -10,6 +10,7 @@ import (
 	"testing"
 	"time"
 
+	"github.com/henrygd/beszel"
 	"github.com/henrygd/beszel/internal/entities/monitor"
 	"github.com/stretchr/testify/assert"
 	"github.com/stretchr/testify/require"
@@ -240,6 +241,7 @@ func TestMonitorManagerGetRandomDelay(t *testing.T) {
 func TestMonitorHTTP(t *testing.T) {
 	t.Run("success", func(t *testing.T) {
 		server := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+			assert.Equal(t, "Beszel-Agent/"+beszel.Version+" (+https://beszel.dev)", r.Header.Get("User-Agent"))
 			w.WriteHeader(http.StatusNoContent)
 		}))
 		defer server.Close()

internal/alerts/alerts_api_test.go

@@ -11,6 +11,7 @@ import (
 	"strings"
 	"sync/atomic"
 	"testing"
+	"testing/synctest"
 
 	beszelTests "github.com/henrygd/beszel/internal/tests"
 	pbTests "github.com/pocketbase/pocketbase/tests"
@@ -533,6 +534,20 @@ func TestSendTestNotification(t *testing.T) {
 
 	for _, url := range []string{localURL, "smtp://user:pass@127.0.0.1/?fromAddress=sender@example.com&toAddresses=recipient@example.com", "mqtt://127.0.0.1/topic"} {
 		scenarios = append(scenarios, beszelTests.ApiScenario{
+			BeforeTestFunc: func(tb testing.TB, _ *pbTests.TestApp, e *core.ServeEvent) {
+				if !strings.HasPrefix(url, "mqtt://") {
+					return
+				}
+				// Keep the real MQTT rejection path, but advance its library's
+				// fixed timeout using virtual time instead of waiting 10 seconds.
+				e.Router.BindFunc(func(re *core.RequestEvent) error {
+					var err error
+					synctest.Test(tb.(*testing.T), func(t *testing.T) {
+						err = re.Next()
+					})
+					return err
+				})
+			},
 			Name:            "readonly cannot send to " + url,
 			Method:          http.MethodPost,
 			URL:             "/api/beszel/test-notification",

internal/alerts/notification_client_test.go

@@ -11,6 +11,7 @@ import (
 	"strings"
 	"sync/atomic"
 	"testing"
+	"testing/synctest"
 
 	"github.com/nicholas-fedor/shoutrrr/pkg/types"
 	"golang.org/x/net/dns/dnsmessage"
@@ -178,39 +179,45 @@ func TestPublicNotificationTCP(t *testing.T) {
 	} {
 		t.Run(rawURL, func(t *testing.T) {
 			t.Parallel()
+			// MQTT waits for a fixed library timeout even after a dial failure.
+			// Virtual time preserves the full send/cleanup path without that delay.
 			t.Run("internal destination", func(t *testing.T) {
-				err := sendPublicNotification(strings.ReplaceAll(rawURL, "HOST", "127.0.0.1"), "test")
-				if !errors.Is(err, errInternalDestination) {
-					t.Fatalf("expected blocked destination, got %v", err)
-				}
+				synctest.Test(t, func(t *testing.T) {
+					err := sendPublicNotification(strings.ReplaceAll(rawURL, "HOST", "127.0.0.1"), "test")
+					if !errors.Is(err, errInternalDestination) {
+						t.Fatalf("expected blocked destination, got %v", err)
+					}
+				})
 			})
 			t.Run("public destination uses injected dialer", func(t *testing.T) {
-				var calls atomic.Int32
-				stopped := errors.New("test dial stopped")
-				service, err := newPublicNotificationService(strings.ReplaceAll(rawURL, "HOST", "8.8.8.8"), types.SenderOptions{
-					DialContext: func(ctx context.Context, network, address string) (net.Conn, error) {
-						calls.Add(1)
-						if network != "tcp" || !strings.HasPrefix(address, "8.8.8.8:") {
-							t.Errorf("unexpected dial: %s %s", network, address)
-						}
-						if err := checkNotificationAddress(address); err != nil {
-							t.Error(err)
-						}
-						return nil, stopped
-					},
+				synctest.Test(t, func(t *testing.T) {
+					var calls atomic.Int32
+					stopped := errors.New("test dial stopped")
+					service, err := newPublicNotificationService(strings.ReplaceAll(rawURL, "HOST", "8.8.8.8"), types.SenderOptions{
+						DialContext: func(ctx context.Context, network, address string) (net.Conn, error) {
+							calls.Add(1)
+							if network != "tcp" || !strings.HasPrefix(address, "8.8.8.8:") {
+								t.Errorf("unexpected dial: %s %s", network, address)
+							}
+							if err := checkNotificationAddress(address); err != nil {
+								t.Error(err)
+							}
+							return nil, stopped
+						},
+					})
+					if err != nil {
+						t.Fatal(err)
+					}
+					if closer, ok := service.(io.Closer); ok {
+						defer closer.Close()
+					}
+					if err := service.Send("test", &types.Params{}); err == nil {
+						t.Fatal("expected dial failure")
+					}
+					if calls.Load() == 0 {
+						t.Fatal("custom dialer was not used")
+					}
 				})
-				if err != nil {
-					t.Fatal(err)
-				}
-				if closer, ok := service.(io.Closer); ok {
-					defer closer.Close()
-				}
-				if err := service.Send("test", &types.Params{}); err == nil {
-					t.Fatal("expected dial failure")
-				}
-				if calls.Load() == 0 {
-					t.Fatal("custom dialer was not used")
-				}
 			})
 		})
 	}

internal/cmd/agent/agent.go

@@ -1,9 +1,11 @@
 package main
 
 import (
+	"errors"
 	"fmt"
 	"log"
 	"os"
+	"runtime"
 	"strings"
 
 	"github.com/henrygd/beszel"
@@ -14,6 +16,12 @@ import (
 	"golang.org/x/crypto/ssh"
 )
 
+type noKeyProvidedError struct{}
+
+func (noKeyProvidedError) Error() string {
+	return "no key provided: must set -key flag, KEY env var, or KEY_FILE env var. Use 'beszel-agent help' for usage"
+}
+
 // cli options
 type cmdOptions struct {
 	key    string // key is the public key(s) for SSH authentication.
@@ -124,7 +132,7 @@ func (opts *cmdOptions) loadPublicKeys() ([]ssh.PublicKey, error) {
 	// Try key file
 	keyFile, ok := utils.GetEnv("KEY_FILE")
 	if !ok {
-		return nil, fmt.Errorf("no key provided: must set -key flag, KEY env var, or KEY_FILE env var. Use 'beszel-agent help' for usage")
+		return nil, noKeyProvidedError{}
 	}
 
 	pubKey, err := os.ReadFile(keyFile)
@@ -138,6 +146,14 @@ func (opts *cmdOptions) getAddress() string {
 	return agent.GetAddress(opts.listen)
 }
 
+func isBenignStartupError(err error, goos string) bool {
+	if goos != "windows" {
+		return false
+	}
+	var noKeyErr noKeyProvidedError
+	return errors.As(err, &noKeyErr)
+}
+
 // handleFingerprint handles the "fingerprint" command with subcommands "view" and "reset".
 func handleFingerprint() {
 	subCmd := ""
@@ -182,6 +198,12 @@ func main() {
 	var err error
 	serverConfig.Keys, err = opts.loadPublicKeys()
 	if err != nil {
+		if isBenignStartupError(err, runtime.GOOS) {
+			// WinGet launches the executable without configuration during validation.
+			// Exit successfully in that case while retaining the error on other platforms.
+			log.Print("Failed to load public keys:", err)
+			return
+		}
 		log.Fatal("Failed to load public keys:", err)
 	}
 

internal/cmd/agent/agent_test.go

@@ -2,6 +2,7 @@ package main
 
 import (
 	"crypto/ed25519"
+	"errors"
 	"os"
 	"path/filepath"
 	"testing"
@@ -187,6 +188,26 @@ func TestLoadPublicKeys(t *testing.T) {
 	}
 }
 
+func TestIsBenignStartupError(t *testing.T) {
+	tests := []struct {
+		name string
+		err  error
+		goos string
+		want bool
+	}{
+		{name: "missing key on windows", err: noKeyProvidedError{}, goos: "windows", want: true},
+		{name: "wrapped missing key on windows", err: errors.Join(errors.New("startup failed"), noKeyProvidedError{}), goos: "windows", want: true},
+		{name: "missing key on linux", err: noKeyProvidedError{}, goos: "linux", want: false},
+		{name: "different error on windows", err: errors.New("invalid key"), goos: "windows", want: false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			assert.Equal(t, tt.want, isBenignStartupError(tt.err, tt.goos))
+		})
+	}
+}
+
 func TestGetNetwork(t *testing.T) {
 	tests := []struct {
 		name     string

internal/dockerfile_agent

@@ -12,7 +12,7 @@ RUN apk add --no-cache ca-certificates && update-ca-certificates
 
 # Build
 ARG TARGETOS TARGETARCH
-RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
+RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
 
 RUN rm -rf /tmp/*
 

internal/dockerfile_agent_alpine

@@ -10,14 +10,14 @@ COPY . ./
 
 # Build
 ARG TARGETOS TARGETARCH
-RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
+RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
 
 RUN rm -rf /tmp/*
 
 # --------------------------
 # Final image: default scratch-based agent
 # --------------------------
-FROM alpine:3.23
+FROM alpine:3.24
 COPY --from=builder /agent /agent
 
 # AMD GPU name lookup (used by agent on Linux when /usr/share/libdrm/amdgpu.ids is read)
@@ -28,4 +28,4 @@ RUN apk add --no-cache smartmontools zfs
 # Ensure data persistence across container recreations
 VOLUME ["/var/lib/beszel-agent"]
 
-ENTRYPOINT ["/agent"]
+ENTRYPOINT ["/agent"]

internal/dockerfile_agent_intel

@@ -10,13 +10,13 @@ COPY . ./
 
 # Build
 ARG TARGETOS TARGETARCH
-RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
+RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /agent ./internal/cmd/agent
 
 # --------------------------
 # Final image
 # Note: must cap_add: [CAP_PERFMON] and mount /dev/dri/ as volume
 # --------------------------
-FROM alpine:3.23
+FROM alpine:3.24
 
 COPY --from=builder /agent /agent
 

internal/dockerfile_agent_nvidia

@@ -10,7 +10,7 @@ COPY . ./
 
 # Build
 ARG TARGETOS TARGETARCH
-RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -tags glibc -ldflags "-w -s" -o /agent ./internal/cmd/agent
+RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -tags glibc -ldflags "-w -s" -o /agent ./internal/cmd/agent
 
 # --------------------------
 # Smartmontools builder stage

internal/dockerfile_agent_nvidia_slim

@@ -17,7 +17,7 @@ RUN set -eux; \
   if [ "$TARGETARCH" = "arm" ] && [ -n "$TARGETVARIANT" ]; then \
     export GOARM="${TARGETVARIANT#v}"; \
   fi; \
-  CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH \
+  CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH \
     go build -tags glibc -ldflags "-w -s" -o /agent ./internal/cmd/agent
 
 # --------------------------
@@ -70,7 +70,9 @@ RUN set -eux; \
 # --------------------------
 FROM --platform=$TARGETPLATFORM debian:bookworm-slim AS zfsutils-builder
 
-RUN apt-get update && apt-get install -y --no-install-recommends \
+# zfsutils-linux is distributed in Debian's contrib component.
+RUN sed -i 's/Components: main/Components: main contrib/' /etc/apt/sources.list.d/debian.sources \
+  && apt-get update && apt-get install -y --no-install-recommends \
   zfsutils-linux \
   && rm -rf /var/lib/apt/lists/*
 

internal/dockerfile_hub

@@ -17,7 +17,7 @@ RUN update-ca-certificates
 
 # Build
 ARG TARGETOS TARGETARCH
-RUN CGO_ENABLED=0 GOGC=75 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /beszel ./internal/cmd/hub
+RUN CGO_ENABLED=0 GOOS=$TARGETOS GOARCH=$TARGETARCH go build -ldflags "-w -s" -o /beszel ./internal/cmd/hub
 
 # ? -------------------------
 FROM scratch
@@ -31,4 +31,4 @@ VOLUME ["/beszel_data"]
 EXPOSE 8090
 
 ENTRYPOINT [ "/beszel" ]
-CMD ["serve", "--http=0.0.0.0:8090"]
+CMD ["serve", "--http=0.0.0.0:8090"]

internal/hub/agent_connect_test.go

@@ -978,11 +978,18 @@ func TestAgentWebSocketIntegration(t *testing.T) {
 				}
 			}
 
-			// Verify system status
-			updatedSystemRecord, err := testApp.FindRecordById("systems", systemRecord.Id)
-			require.NoError(t, err)
-			status := updatedSystemRecord.GetString("status")
-			assert.Equal(t, tc.expectSystemStatus, status, "System status should match expected value")
+			// A connected WebSocket does not mean the hub has finished verifying
+			// the agent and updating the system. Wait for the database state rather
+			// than assuming that work completes within a fixed sleep under load.
+			var status string
+			require.EventuallyWithT(t, func(c *assert.CollectT) {
+				updatedSystemRecord, err := testApp.FindRecordById("systems", systemRecord.Id)
+				if !assert.NoError(c, err) {
+					return
+				}
+				status = updatedSystemRecord.GetString("status")
+				assert.Equal(c, tc.expectSystemStatus, status, "System status should match expected value")
+			}, 5*time.Second, 20*time.Millisecond)
 
 			t.Logf("%s - System status: %s, Fingerprint: %s", tc.description, status, finalFingerprint)
 		})
@@ -1142,42 +1149,43 @@ func TestMultipleSystemsWithSameUniversalToken(t *testing.T) {
 
 			// Verify system creation/reuse behavior
 			if tc.expectConnection {
-				// Count systems after connection
-				systemsAfter, err := testApp.FindRecordsByFilter("systems", "users ~ {:userId}", "", -1, 0, map[string]any{"userId": userRecord.Id})
-				require.NoError(t, err)
-				systemsAfterCount := len(systemsAfter)
-
+				expectedSystemsAfter := systemsBeforeCount
 				if tc.expectNewSystem {
-					// Should have created a new system
+					expectedSystemsAfter++
 					systemCount++
-					assert.Equal(t, systemsBeforeCount+1, systemsAfterCount, "Should have created a new system")
-					assert.Equal(t, systemCount, systemsAfterCount, "Total system count should match expected")
-				} else {
-					// Should have reused existing system
-					assert.Equal(t, systemsBeforeCount, systemsAfterCount, "Should not have created a new system")
-					assert.Equal(t, systemCount, systemsAfterCount, "Total system count should remain the same")
 				}
 
-				time.Sleep(20 * time.Millisecond)
+				// WebSocket connection precedes the hub's asynchronous system
+				// setup. Re-read all database state until setup is complete.
+				var systemId, status string
+				require.EventuallyWithT(t, func(c *assert.CollectT) {
+					systemsAfter, err := testApp.FindRecordsByFilter("systems", "users ~ {:userId}", "", -1, 0, map[string]any{"userId": userRecord.Id})
+					if !assert.NoError(c, err) {
+						return
+					}
+					assert.Len(c, systemsAfter, expectedSystemsAfter, "System creation/reuse should match expected behavior")
+					assert.Len(c, systemsAfter, systemCount, "Total system count should match expected")
 
-				// Verify that a fingerprint record exists for this fingerprint
-				fingerprints, err := testApp.FindRecordsByFilter("fingerprints", "token = {:token} && fingerprint = {:fingerprint}", "", -1, 0, map[string]any{
-					"token":       universalToken,
-					"fingerprint": tc.agentFingerprint,
-				})
-				require.NoError(t, err)
-				require.Len(t, fingerprints, 1, "Should have exactly one fingerprint record for this token+fingerprint combination")
+					fingerprints, err := testApp.FindRecordsByFilter("fingerprints", "token = {:token} && fingerprint = {:fingerprint}", "", -1, 0, map[string]any{
+						"token":       universalToken,
+						"fingerprint": tc.agentFingerprint,
+					})
+					if !assert.NoError(c, err) || !assert.Len(c, fingerprints, 1, "Should have exactly one fingerprint record for this token+fingerprint combination") {
+						return
+					}
 
-				fingerprint := fingerprints[0]
-				assert.Equal(t, universalToken, fingerprint.GetString("token"), "Fingerprint should have the universal token")
-				assert.Equal(t, tc.agentFingerprint, fingerprint.GetString("fingerprint"), "Fingerprint should match agent's fingerprint")
+					fingerprint := fingerprints[0]
+					assert.Equal(c, universalToken, fingerprint.GetString("token"), "Fingerprint should have the universal token")
+					assert.Equal(c, tc.agentFingerprint, fingerprint.GetString("fingerprint"), "Fingerprint should match agent's fingerprint")
 
-				// Verify system status
-				systemId := fingerprint.GetString("system")
-				system, err := testApp.FindRecordById("systems", systemId)
-				require.NoError(t, err)
-				status := system.GetString("status")
-				assert.Equal(t, tc.expectSystemStatus, status, "System status should match expected value")
+					systemId = fingerprint.GetString("system")
+					system, err := testApp.FindRecordById("systems", systemId)
+					if !assert.NoError(c, err) {
+						return
+					}
+					status = system.GetString("status")
+					assert.Equal(c, tc.expectSystemStatus, status, "System status should match expected value")
+				}, 5*time.Second, 20*time.Millisecond)
 
 				t.Logf("%s - System ID: %s, Status: %s, New System: %v", tc.description, systemId, status, tc.expectNewSystem)
 			}

internal/hub/systems/network_monitor_ssh_test.go

@@ -0,0 +1,159 @@
+//go:build testing
+
+package systems
+
+import (
+	"context"
+	"crypto/ed25519"
+	"crypto/rand"
+	"net"
+	"sync/atomic"
+	"testing"
+	"time"
+
+	"github.com/fxamacker/cbor/v2"
+	"github.com/henrygd/beszel/internal/common"
+	"github.com/henrygd/beszel/internal/entities/monitor"
+	esystem "github.com/henrygd/beszel/internal/entities/system"
+	"github.com/henrygd/beszel/internal/hub/expirymap"
+	"github.com/pocketbase/pocketbase/core"
+	"github.com/stretchr/testify/require"
+	"golang.org/x/crypto/ssh"
+)
+
+func TestSSHNetworkMonitorReconnectSync(t *testing.T) {
+	sys, app := newTestSystemWithHub(t)
+	sys.manager.zfsFetchMap = expirymap.New[zfsFetchState](time.Hour)
+	t.Cleanup(sys.manager.zfsFetchMap.StopCleaner)
+	sys.ctx = context.Background()
+	sys.Status = up
+	_, key, err := ed25519.GenerateKey(rand.Reader)
+	require.NoError(t, err)
+	signer, err := ssh.NewSignerFromKey(key)
+	require.NoError(t, err)
+	config := &ssh.ServerConfig{NoClientAuth: true, ServerVersion: "SSH-2.0-beszel_0.20.0"}
+	config.AddHostKey(signer)
+	listener, err := net.Listen("tcp", "127.0.0.1:0")
+	require.NoError(t, err)
+	t.Cleanup(func() { _ = listener.Close() })
+	sys.Host, sys.Port, err = net.SplitHostPort(listener.Addr().String())
+	require.NoError(t, err)
+	sys.manager.sshConfig = &ssh.ClientConfig{User: "test", HostKeyCallback: ssh.InsecureIgnoreHostKey(), Timeout: time.Second}
+	t.Cleanup(sys.closeSSHConnection)
+	requests := make(chan monitor.SyncRequest, 10)
+	var failSync atomic.Bool
+	go func() {
+		for {
+			conn, err := listener.Accept()
+			if err != nil {
+				return
+			}
+			go func() {
+				server, channels, reqs, err := ssh.NewServerConn(conn, config)
+				if err != nil {
+					_ = conn.Close()
+					return
+				}
+				defer server.Close()
+				go ssh.DiscardRequests(reqs)
+				for channel := range channels {
+					ch, reqs, err := channel.Accept()
+					if err != nil {
+						return
+					}
+					go func() {
+						defer ch.Close()
+						for req := range reqs {
+							if req.Type != "shell" {
+								_ = req.Reply(false, nil)
+								continue
+							}
+							_ = req.Reply(true, nil)
+							var request common.HubRequest[cbor.RawMessage]
+							if cbor.NewDecoder(ch).Decode(&request) != nil {
+								return
+							}
+							response := common.AgentResponse{}
+							switch request.Action {
+							case common.GetData:
+								response.SystemData = &esystem.CombinedData{}
+							case common.SyncNetworkMonitors:
+								var syncReq monitor.SyncRequest
+								if cbor.Unmarshal(request.Data, &syncReq) != nil {
+									return
+								}
+								requests <- syncReq
+								if failSync.Load() {
+									response.Error = "test sync failure"
+								} else {
+									response.Data, _ = cbor.Marshal(monitor.SyncResponse{})
+								}
+							}
+							_ = cbor.NewEncoder(ch).Encode(response)
+							_, _ = ch.SendRequest("exit-status", false, ssh.Marshal(struct{ Status uint32 }{0}))
+							return
+						}
+					}()
+				}
+			}()
+		}
+	}()
+	collection, err := app.FindCachedCollectionByNameOrId("network_monitors")
+	require.NoError(t, err)
+	probe := core.NewRecord(collection)
+	probe.Load(map[string]any{"system": sys.Id, "target": "localhost", "protocol": "tcp", "port": 80, "interval": 60, "enabled": true})
+	require.NoError(t, app.SaveNoValidate(probe))
+	fetch := func() {
+		t.Helper()
+		_, err := sys.fetchDataFromAgent(common.DataRequestOptions{})
+		require.NoError(t, err, "monitor sync failure must not fail stats fetching")
+	}
+	receive := func() monitor.SyncRequest {
+		t.Helper()
+		select {
+		case req := <-requests:
+			require.Equal(t, monitor.SyncActionReplace, req.Action)
+			return req
+		case <-time.After(time.Second):
+			t.Fatal("missing full monitor sync")
+			return monitor.SyncRequest{}
+		}
+	}
+	fetch()
+	require.Equal(t, probe.Id, receive().Configs[0].ID)
+	require.False(t, sys.monitorsNeedSync.Load())
+	fetch()
+	require.Empty(t, requests, "steady-state fetch must not resync")
+
+	// Simulate loss of the agent process/connection and its in-memory monitors.
+	require.NoError(t, sys.client.Load().Close())
+	fetch()
+	require.Equal(t, probe.Id, receive().Configs[0].ID)
+	require.False(t, sys.monitorsNeedSync.Load())
+
+	// Failed replacements are retried on the next successful stats fetch.
+	require.NoError(t, sys.client.Load().Close())
+	failSync.Store(true)
+	fetch()
+	receive()
+	require.True(t, sys.monitorsNeedSync.Load())
+	failSync.Store(false)
+	fetch()
+	receive()
+	require.False(t, sys.monitorsNeedSync.Load())
+
+	probe.Set("enabled", false)
+	require.NoError(t, app.SaveNoValidate(probe))
+	require.NoError(t, sys.client.Load().Close())
+	fetch()
+	require.Empty(t, receive().Configs, "empty replacement must clear stale monitors")
+}
+
+func TestPendingNetworkMonitorSyncQueryFailure(t *testing.T) {
+	sys, app := newTestSystemWithHub(t)
+	_, err := app.DB().NewQuery("DROP TABLE network_monitors").Execute()
+	require.NoError(t, err)
+	sys.monitorsNeedSync.Store(true)
+	sys.syncPendingNetworkMonitors()
+	require.True(t, sys.monitorsNeedSync.Load())
+}

internal/hub/systems/network_monitor_sync_test.go

@@ -6,6 +6,7 @@ import (
 	"net/http"
 	"net/http/httptest"
 	"strings"
+	"sync/atomic"
 	"testing"
 	"time"
 
@@ -13,6 +14,7 @@ import (
 	"github.com/fxamacker/cbor/v2"
 	"github.com/henrygd/beszel/internal/common"
 	"github.com/henrygd/beszel/internal/entities/monitor"
+	esystem "github.com/henrygd/beszel/internal/entities/system"
 	"github.com/henrygd/beszel/internal/hub/ws"
 	"github.com/lxzan/gws"
 	"github.com/pocketbase/pocketbase/core"
@@ -22,17 +24,31 @@ import (
 type monitorSyncClient struct {
 	gws.BuiltinEventHandler
 	requests chan common.HubRequest[monitor.SyncRequest]
+	failSync atomic.Bool
 }
 
 func (c *monitorSyncClient) OnMessage(conn *gws.Conn, message *gws.Message) {
 	defer message.Close()
-	var req common.HubRequest[monitor.SyncRequest]
+	var req common.HubRequest[cbor.RawMessage]
 	if err := cbor.Unmarshal(message.Bytes(), &req); err != nil {
 		return
 	}
-	c.requests <- req
-	data, _ := cbor.Marshal(monitor.SyncResponse{})
-	response, _ := cbor.Marshal(common.AgentResponse{Id: req.Id, Data: data})
+	resp := common.AgentResponse{Id: req.Id}
+	if req.Action == common.GetData {
+		resp.SystemData = &esystem.CombinedData{}
+	} else {
+		var data monitor.SyncRequest
+		if err := cbor.Unmarshal(req.Data, &data); err != nil {
+			return
+		}
+		c.requests <- common.HubRequest[monitor.SyncRequest]{Id: req.Id, Action: req.Action, Data: data}
+		if c.failSync.Load() {
+			resp.Error = "test sync failure"
+		} else {
+			resp.Data, _ = cbor.Marshal(monitor.SyncResponse{})
+		}
+	}
+	response, _ := cbor.Marshal(resp)
 	_ = conn.WriteMessage(gws.OpcodeBinary, response)
 }
 
@@ -57,7 +73,7 @@ func TestNetworkMonitorSyncSkipsOlderAgents(t *testing.T) {
 }
 
 func TestNetworkMonitorReconnectSync(t *testing.T) {
-	for _, change := range []string{"delete", "disable"} {
+	for _, change := range []string{"delete", "disable", "retry"} {
 		t.Run(change, func(t *testing.T) {
 			sys, app := newTestSystemWithHub(t)
 			record, err := app.FindRecordById("systems", sys.Id)
@@ -120,9 +136,33 @@ func TestNetworkMonitorReconnectSync(t *testing.T) {
 				}
 			}
 
+			client.failSync.Store(change == "retry")
 			initial := connect()
 			require.Len(t, initial.Configs, 1)
 			require.Equal(t, probe.Id, initial.Configs[0].ID)
+			if change == "retry" {
+				system, err := sm.GetSystem(sys.Id)
+				require.NoError(t, err)
+				require.Eventually(t, system.monitorsNeedSync.Load, time.Second, time.Millisecond)
+				// A second failed sync must not fail the stats fetch or clear pending state.
+				_, err = system.fetchDataFromAgent(common.DataRequestOptions{})
+				require.NoError(t, err)
+				require.True(t, system.monitorsNeedSync.Load())
+				require.Len(t, client.requests, 1)
+				<-client.requests
+				client.failSync.Store(false)
+				_, err = system.fetchDataFromAgent(common.DataRequestOptions{})
+				require.NoError(t, err)
+				require.False(t, system.monitorsNeedSync.Load())
+				require.Len(t, client.requests, 1)
+				retry := <-client.requests
+				require.Equal(t, monitor.SyncActionReplace, retry.Data.Action)
+				require.Equal(t, initial.Configs, retry.Data.Configs)
+				_, err = system.fetchDataFromAgent(common.DataRequestOptions{})
+				require.NoError(t, err)
+				require.Empty(t, client.requests, "successful sync must not repeat on every fetch")
+				return
+			}
 			require.NoError(t, sm.RemoveSystem(sys.Id))
 			if change == "delete" {
 				require.NoError(t, app.Delete(probe))

internal/hub/systems/network_monitors.go

@@ -2,6 +2,7 @@ package systems
 
 import (
 	"context"
+	"fmt"
 	"time"
 
 	"github.com/henrygd/beszel"
@@ -9,6 +10,27 @@ import (
 	"github.com/henrygd/beszel/internal/entities/monitor"
 )
 
+// syncPendingNetworkMonitors runs on WebSocket connect and after successful stats
+// fetches. Failed syncs retry on the next update without taking the system down.
+func (sys *System) syncPendingNetworkMonitors() {
+	if !sys.monitorsNeedSync.Swap(false) {
+		return
+	}
+	if err := sys.syncAllNetworkMonitors(); err != nil {
+		sys.monitorsNeedSync.Store(true)
+		sys.manager.hub.Logger().Warn("failed to sync monitors to agent", "system", sys.Id, "err", err)
+	}
+}
+
+func (sys *System) syncAllNetworkMonitors() error {
+	configs, err := sys.manager.GetMonitorConfigsForSystem(sys.Id)
+	if err != nil {
+		return fmt.Errorf("failed to load monitors: %w", err)
+	}
+	// An empty set must also replace probes retained across a disconnect.
+	return sys.SyncNetworkMonitors(configs)
+}
+
 // SyncNetworkMonitors sends monitor configurations to the agent.
 func (sys *System) SyncNetworkMonitors(configs []monitor.Config) error {
 	_, err := sys.syncNetworkMonitors(monitor.SyncRequest{Action: monitor.SyncActionReplace, Configs: configs})

internal/hub/systems/system.go

@@ -56,6 +56,9 @@ type System struct {
 	smartInterval  time.Duration              // Interval for periodic SMART data updates
 	zfsFetching    atomic.Bool                // True if ZFS pools are currently being fetched
 	zfsInterval    time.Duration              // Interval for periodic ZFS detail data updates
+
+	// A fresh connection needs a full monitor configuration sync.
+	monitorsNeedSync atomic.Bool
 	// Serialize persistence from scheduled updates and resumes through commit.
 	recordsMu sync.Mutex
 	// Protected by recordsMu; realtime reads don't consume probes.
@@ -630,7 +633,10 @@ func (sys *System) request(ctx context.Context, action common.WebSocketAction, r
 	err := sys.sshTransport.RequestWithRetry(ctx, action, req, dest, 1)
 	// Keep legacy SSH client/version fields in sync for other code paths.
 	if sys.sshTransport != nil {
-		sys.client.Store(sys.sshTransport.GetClient())
+		client := sys.sshTransport.GetClient()
+		if previous := sys.client.Swap(client); client != nil && client != previous {
+			sys.monitorsNeedSync.Store(true)
+		}
 		sys.agentVersion = sys.sshTransport.GetAgentVersion()
 	}
 	return err
@@ -688,6 +694,7 @@ func (sys *System) fetchDataFromAgent(options common.DataRequestOptions) (*syste
 	if sys.WsConn != nil && sys.WsConn.IsConnected() {
 		wsData, err := sys.fetchDataViaWebSocket(options)
 		if err == nil {
+			sys.syncPendingNetworkMonitors()
 			return wsData, nil
 		}
 		// close the WebSocket connection if error and try SSH
@@ -698,6 +705,7 @@ func (sys *System) fetchDataFromAgent(options common.DataRequestOptions) (*syste
 	if err != nil {
 		return nil, err
 	}
+	sys.syncPendingNetworkMonitors()
 	return sshData, nil
 }
 
@@ -932,6 +940,7 @@ func (s *System) createSSHClient() error {
 		return err
 	}
 	s.agentVersion, _ = extractAgentVersion(string(client.Conn.ServerVersion()))
+	s.monitorsNeedSync.Store(true)
 	s.manager.resetFailedSmartFetchState(s.Id)
 	s.manager.resetFailedZfsFetchState(s.Id)
 	return nil

internal/hub/systems/system_manager.go

@@ -349,23 +349,14 @@ func (sm *SystemManager) AddWebSocketSystem(systemId string, agentVersion semver
 	system := sm.NewSystem(systemId)
 	system.WsConn = wsConn
 	system.agentVersion = agentVersion
+	system.monitorsNeedSync.Store(true)
 
 	if err := sm.AddRecord(systemRecord, system); err != nil {
 		return err
 	}
 
 	// Sync network monitors to the newly connected agent
-	go func() {
-		configs, err := sm.GetMonitorConfigsForSystem(systemId)
-		if err != nil {
-			sm.hub.Logger().Warn("failed to load monitors for agent", "system", systemId, "err", err)
-			return
-		}
-		// An empty set must also replace any probes retained across a disconnect.
-		if err := system.SyncNetworkMonitors(configs); err != nil {
-			sm.hub.Logger().Warn("failed to sync monitors to agent", "system", systemId, "err", err)
-		}
-	}()
+	go system.syncPendingNetworkMonitors()
 
 	return nil
 }

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

@@ -59,7 +59,11 @@ export const containerChartCols: ColumnDef<ContainerRecord>[] = [
 			const allSystems = $allSystemsById.get()
 			const systemNameA = allSystems[a.original.system]?.name ?? ""
 			const systemNameB = allSystems[b.original.system]?.name ?? ""
-			return systemNameA.localeCompare(systemNameB)
+			const primary = systemNameA.localeCompare(systemNameB)
+			if (primary !== 0) {
+				return primary
+			}
+			return a.original.name.localeCompare(b.original.name)
 		},
 		header: ({ column }) => <HeaderButton column={column} name={t`System`} Icon={ServerIcon} />,
 		cell: ({ getValue }) => {

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

@@ -211,7 +211,6 @@ export default memo(function SystemDetail({ id }: { id: string }) {
 						<TemperatureChart {...coreProps} setPageBottomExtraMargin={setPageBottomExtraMargin} />
 						<FanChart {...coreProps} />
 						<BatteryChart system={system} {...coreProps} />
-						<SwapChart chartData={chartData} grid={grid} dataEmpty={dataEmpty} systemStats={systemStats} />
 						{pageBottomExtraMargin > 0 && <div style={{ marginBottom: pageBottomExtraMargin }}></div>}
 					</div>
 				</TabsContent>

internal/site/src/components/routes/system/charts/monitors-charts.tsx

@@ -27,6 +27,8 @@ type MonitorChartBaseProps = MonitorChartProps & {
 	tickFormatter: (value: number) => string
 	contentFormatter: ({ value }: { value: number | string }) => string | number
 	domain?: [number | "auto", number | "auto"]
+	/** Overrides the per-monitor line colors (e.g. a fixed color for single-monitor charts). */
+	color?: string
 }
 
 function MonitorChart({
@@ -41,6 +43,7 @@ function MonitorChart({
 	tickFormatter,
 	contentFormatter,
 	domain,
+	color,
 	showFilter = monitors.length > 1,
 }: MonitorChartBaseProps) {
 	const storedFilter = useStore($monitorFilter)
@@ -67,11 +70,12 @@ function MonitorChart({
 				label,
 				dataKey: (record: NetworkMonitorStatsRecord) => record.stats?.[p.id]?.[metric] ?? null,
 				dot,
-				color: count <= 5 ? i + 1 : `hsl(${(i * 360) / count}, var(--chart-saturation), var(--chart-lightness))`,
+				color:
+					color ?? (count <= 5 ? i + 1 : `hsl(${(i * 360) / count}, var(--chart-saturation), var(--chart-lightness))`),
 			})
 		}
 		return { dataPoints: points, visibleKeys: visibleIDs }
-	}, [monitors, filter, metric, chartData.chartTime])
+	}, [monitors, filter, metric, chartData.chartTime, color])
 
 	const filteredMonitorStats = useMemo(() => {
 		if (!visibleKeys.length) return monitorStats
@@ -200,6 +204,7 @@ export function LossChart({ monitorStats, grid, monitors, chartData, empty, titl
 			title={title}
 			description={t`Packet loss (%)`}
 			domain={[0, 100]}
+			color="var(--destructive)"
 			tickFormatter={(value) => `${toFixedFloat(value, value >= 10 ? 0 : 1)}%`}
 			contentFormatter={({ value }) => {
 				if (typeof value !== "number") {

internal/site/src/locales/ja/ja.po

@@ -8,7 +8,7 @@ msgstr ""
 "Language: ja\n"
 "Project-Id-Version: beszel\n"
 "Report-Msgid-Bugs-To: \n"
-"PO-Revision-Date: 2026-09-18 19:21\n"
+"PO-Revision-Date: 2026-09-22 03:34\n"
 "Last-Translator: \n"
 "Language-Team: Japanese\n"
 "Plural-Forms: nplurals=1; plural=0;\n"
@@ -2305,3 +2305,4 @@ msgstr "はい"
 #: src/components/routes/settings/layout.tsx
 msgid "Your user settings have been updated."
 msgstr "ユーザー設定が更新されました。"
+

readme.md

@@ -15,9 +15,10 @@ It has a friendly web interface, simple configuration, and is ready to use out o
 
 - **Lightweight**: Smaller and less resource-intensive than leading solutions.
 - **Simple**: Easy setup with little manual configuration required.
+- **Alerts**: Configurable alerts for most metrics. Supports many notification services.
 - **Docker stats**: Tracks CPU, memory, and network usage history for each container.
-- **ZFS**: Tracks pool capacity, health, and I/O, plus per-dataset usage.
-- **Alerts**: Configurable alerts for CPU, memory, disk, bandwidth, temperature, fan speed, load average, and status.
+- **Network monitoring**: Monitor response time and interruptions directly from agents.
+- **S.M.A.R.T.**: Disk health data and notifications on drive failure.
 - **Multi-user**: Users manage their own systems. Admins can share systems across users.
 - **OAuth / OIDC**: Supports many OAuth2 providers. Password auth can be disabled.
 - **Automatic backups**: Save to and restore from disk or S3-compatible storage.
@@ -51,7 +52,7 @@ The [quick start guide](https://beszel.dev/guide/getting-started) and other docu
 - **Temperature** - Host system sensors.
 - **Fan speed** - Host system sensors (Linux, via `/sys/class/hwmon`).
 - **GPU usage / power draw** - Nvidia, AMD, and Intel.
-- **Battery** - Host system battery charge.
+- **Battery charge** - Host system and some peripherals.
 - **Containers** - Status and metrics of all running Docker / Podman containers.
 - **S.M.A.R.T.** - Host system disk health (includes eMMC wear/EOL and Linux mdraid array health via sysfs when available).
 - **ZFS** - Pool capacity, usage, health, I/O throughput, scrub status, and per-dataset usage.

supplemental/helm/beszel-agent/Chart.yaml

@@ -2,9 +2,9 @@ apiVersion: v1
 description: Installs beszel-agent in kubernetes
 home: https://github.com/henrygd/beszel/tree/main/supplemental/helm/beszel-agent
 name: beszel-agent
-appVersion: "0.19.0"
+appVersion: "0.20.0"
 # Bump this version when publishing chart changes.
-version: 0.1.6
+version: 0.1.7
 sources:
   - https://github.com/henrygd/beszel/tree/main/supplemental/helm/beszel-agent
   - https://www.beszel.dev/

supplemental/helm/beszel-agent/README.md

@@ -80,7 +80,7 @@ Essential parameters to configure:
 | `secret.sshKey` | `ssh-key` | Key name in the secret for the SSH public key |
 | `secret.tokenKey` | `token` | Key name in the secret for the authentication token |
 | `image.repository` | `henrygd/beszel-agent` | Container image |
-| `image.tag` | Chart AppVersion (0.19.0) | Image version |
+| `image.tag` | Chart AppVersion (0.20.0) | Image version |
 | `hostNetwork` | `false` | Use host network for network monitoring |
 | `tolerations` | Allows all taints | Tolerations for running on tainted nodes |
 
@@ -385,7 +385,7 @@ helm upgrade beszel-agent ./beszel-agent \
 
 # Change image version
 helm upgrade beszel-agent ./beszel-agent \
-  --set image.tag="0.19.0"
+  --set image.tag="0.20.0"
 ```
 
 ### Restart All Agents
@@ -522,7 +522,7 @@ kubectl get secret beszel-agent -o jsonpath='{.data.ssh-key}' | base64 -d
 ## Chart Information
 
 - **Chart Version**: 0.1.0
-- **App Version**: 0.19.0
+- **App Version**: 0.20.0
 - **Kubernetes Version**: 1.19+
 - **Maintainer**: cloudwithdan (nikoloskid@pm.me)
 

supplemental/helm/beszel-hub/Chart.yaml

@@ -2,9 +2,9 @@ apiVersion: v1
 description: Installs beszel-hub in kubernetes
 home: https://github.com/henrygd/beszel/tree/main/supplemental/helm/beszel-hub
 name: beszel-hub
-appVersion: "0.19.0"
+appVersion: "0.20.0"
 # Bump this version when publishing chart changes.
-version: 0.1.6
+version: 0.1.7
 sources:
   - https://github.com/henrygd/beszel/tree/main/supplemental/helm/beszel-hub
   - https://www.beszel.dev/

supplemental/helm/beszel-hub/README.md

@@ -47,7 +47,7 @@ Key configuration options in `values.yaml`:
 |-----------|---------|-------------|
 | `replicaCount` | `1` | Number of Beszel Hub replicas |
 | `image.repository` | `henrygd/beszel` | Container image repository |
-| `image.tag` | Chart AppVersion (0.19.0) | Container image tag |
+| `image.tag` | Chart AppVersion (0.20.0) | Container image tag |
 | `image.pullPolicy` | `IfNotPresent` | Image pull policy |
 | `service.port` | `8090` | Service port |
 | `persistentVolumeClaim.enabled` | `true` | Enable persistent volume |
@@ -169,7 +169,7 @@ tolerations:
 ```yaml
 replicaCount: 3
 image:
-  tag: "0.19.0"
+  tag: "0.20.0"
 service:
   type: LoadBalancer
 ingress:
@@ -330,7 +330,7 @@ By default, Beszel Hub uses a PersistentVolumeClaim for data storage. Ensure you
 ## Chart Information
 
 - **Chart Version**: 0.1.0
-- **App Version**: 0.19.0
+- **App Version**: 0.20.0
 - **Kubernetes Version**: 1.19+
 - **Maintainer**: cloudwithdan (nikoloskid@pm.me)
 

supplemental/scripts/install-agent.sh

@@ -298,6 +298,94 @@ warn() {
   echo "Warning: $*" >&2
 }
 
+# Keep IDs within 16 bits, including on older OpenWrt versions whose account
+# helpers start automatic allocation at 65536.
+openwrt_unused_id() {
+  awk -F: '
+    { used[$3] = 1 }
+    END {
+      for (id = 32768; id < 65534; id++) {
+        if (!(id in used)) { print id; exit }
+      }
+      exit 1
+    }
+  ' "$1"
+}
+
+validate_openwrt_account() {
+  # Duplicate numeric IDs share permissions even when the names differ.
+  for account_file in /etc/passwd /etc/group; do
+    if ! awk -F: '
+      $1 == "beszel" { id = $3; count++ }
+      { ids[$3]++ }
+      END { if (count && (count != 1 || id == 0 || ids[id] != 1)) exit 1 }
+    ' "$account_file"; then
+      fail "The beszel account has a duplicate or root ID in $account_file. Stop beszel-agent and assign beszel an unused UID/GID before reinstalling. Update ownership only in $AGENT_DIR; do not change all files owned by the shared ID."
+    fi
+  done
+  if grep -q '^beszel:' /etc/passwd; then
+    account_gid=$(awk -F: '$1 == "beszel" { print $4 }' /etc/passwd)
+    group_gid=$(awk -F: '$1 == "beszel" { print $3 }' /etc/group)
+    [ "$account_gid" = "$group_gid" ] || fail "The beszel user's primary group is not the dedicated beszel group. Repair the account before reinstalling."
+  fi
+}
+
+configure_openwrt_account() (
+  validate_openwrt_account
+  [ -r /lib/functions.sh ] || fail "OpenWrt account helpers (/lib/functions.sh) are required."
+  # OpenWrt's library expects unset variables and defines generic functions.
+  # Source it in a subshell so neither affects the rest of the installer.
+  set +u
+  . /lib/functions.sh
+  IPKG_INSTROOT=""
+
+  if ! grep -q '^beszel:' /etc/group; then
+    account_gid=$(openwrt_unused_id /etc/group) || fail "No unused service GID available."
+    group_add beszel "$account_gid" || fail "Could not create the beszel group."
+  fi
+  account_gid=$(awk -F: '$1 == "beszel" { print $3 }' /etc/group)
+  [ -n "$account_gid" ] || fail "The beszel group was not created."
+  if ! grep -q '^beszel:' /etc/passwd; then
+    account_uid=$(openwrt_unused_id /etc/passwd) || fail "No unused service UID available."
+    user_add beszel "$account_uid" "$account_gid" "Beszel agent" /nonexistent /bin/false || fail "Could not create the beszel user."
+  fi
+  grep -q '^beszel:' /etc/passwd || fail "The beszel account is incomplete."
+  validate_openwrt_account
+  # Previous installers omitted the shadow entry. Repair it with a locked
+  # password, without rewriting an existing account or changing its IDs.
+  if ! grep -q '^beszel:' /etc/shadow; then
+    lock /var/lock/passwd || fail "Could not lock the account database."
+    shadow_status=0
+    if ! grep -q '^beszel:' /etc/shadow; then
+      printf 'beszel:!:0:0:99999:7:::\n' >> /etc/shadow || shadow_status=$?
+    fi
+    lock -u /var/lock/passwd || fail "Could not unlock the account database."
+    [ "$shadow_status" -eq 0 ] || fail "Could not create the beszel shadow entry."
+  fi
+
+  if grep -q '^docker:' /etc/group; then
+    # Match complete member names and avoid a leading comma for empty groups.
+    if ! awk -F: '$1 == "docker" { n = split($4, members, ","); for (i = 1; i <= n; i++) if (members[i] == "beszel") found = 1 } END { exit !found }' /etc/group; then
+      echo "Adding beszel to docker group"
+      if grep -q '^docker:[^:]*:[^:]*:$' /etc/group; then
+        sed -i '/^docker:/s/$/beszel/' /etc/group
+      else
+        sed -i '/^docker:/s/$/,beszel/' /etc/group
+      fi
+    fi
+  fi
+)
+
+remove_openwrt_account() {
+  if command -v userdel >/dev/null 2>&1; then
+    userdel beszel || fail "Could not remove the beszel user."
+  elif command -v deluser >/dev/null 2>&1; then
+    deluser beszel || fail "Could not remove the beszel user."
+  else
+    warn "Neither userdel nor deluser is available; the beszel account has been retained."
+  fi
+}
+
 require_value() {
   [ "$#" -ge 2 ] && [ -n "$2" ] || fail "Option $1 requires a value."
 }
@@ -672,7 +760,9 @@ if [ "$UNINSTALL" = true ]; then
   echo "Removing the dedicated user for the agent service..."
   killall beszel-agent 2>/dev/null || true # Usually already stopped by the service manager.
   if id -u beszel >/dev/null 2>&1; then
-    if is_alpine || is_openwrt; then
+    if is_openwrt; then
+      remove_openwrt_account
+    elif is_alpine; then
       deluser beszel || fail "Could not remove the beszel user."
     elif is_freebsd; then
       pw user del beszel || fail "Could not remove the beszel user."
@@ -772,32 +862,7 @@ if is_alpine; then
   fi
   
 elif is_openwrt; then
-  # Create beszel group first if it doesn't exist (check /etc/group directly)
-  if ! grep -q "^beszel:" /etc/group >/dev/null 2>&1; then
-    echo "beszel:x:999:" >> /etc/group
-  fi
-  
-  # Create beszel user if it doesn't exist (double-check to prevent duplicates)
-  if ! id -u beszel >/dev/null 2>&1 && ! grep -q "^beszel:" /etc/passwd >/dev/null 2>&1; then
-    echo "beszel:x:999:999::/nonexistent:/bin/false" >> /etc/passwd
-  fi
-  
-  # Add the user to the docker group if docker group exists and user is not already in it
-  if grep -q "^docker:" /etc/group >/dev/null 2>&1; then
-    echo "Adding beszel to docker group"
-    # Check if beszel is already in docker group
-    if ! grep "^docker:" /etc/group | grep -q "beszel"; then
-      # Add beszel to docker group by modifying /etc/group
-      # Handle both cases: group with existing members and group without members
-      if grep "^docker:" /etc/group | grep -q ":.*:.*$"; then
-        # Group has existing members, append with comma
-        sed -i 's/^docker:\([^:]*:[^:]*:\)\(.*\)$/docker:\1\2,beszel/' /etc/group
-      else
-        # Group has no members, just append
-        sed -i 's/^docker:\([^:]*:[^:]*:\)$/docker:\1beszel/' /etc/group
-      fi
-    fi
-  fi
+  configure_openwrt_account
 
 elif is_freebsd; then
   if is_opnsense || is_pfsense; then