[client] Android debug bundle support (#5888)

Add Android debug bundle support with Troubleshoot UI

Makefile

@@ -5,7 +5,7 @@ GOLANGCI_LINT := $(shell pwd)/bin/golangci-lint
 $(GOLANGCI_LINT):
 	@echo "Installing golangci-lint..."
 	@mkdir -p ./bin
-	@GOBIN=$(shell pwd)/bin go install github.com/golangci/golangci-lint/cmd/golangci-lint@latest
+	@GOBIN=$(shell pwd)/bin go install github.com/golangci/golangci-lint/v2/cmd/golangci-lint@latest
 
 # Lint only changed files (fast, for pre-push)
 lint: $(GOLANGCI_LINT)

client/android/client.go

@@ -8,6 +8,7 @@ import (
 	"os"
 	"slices"
 	"sync"
+	"time"
 
 	"golang.org/x/exp/maps"
 
@@ -15,6 +16,7 @@ import (
 
 	"github.com/netbirdio/netbird/client/iface/device"
 	"github.com/netbirdio/netbird/client/internal"
+	"github.com/netbirdio/netbird/client/internal/debug"
 	"github.com/netbirdio/netbird/client/internal/dns"
 	"github.com/netbirdio/netbird/client/internal/listener"
 	"github.com/netbirdio/netbird/client/internal/peer"
@@ -26,6 +28,7 @@ import (
 	"github.com/netbirdio/netbird/formatter"
 	"github.com/netbirdio/netbird/route"
 	"github.com/netbirdio/netbird/shared/management/domain"
+	types "github.com/netbirdio/netbird/upload-server/types"
 )
 
 // ConnectionListener export internal Listener for mobile
@@ -68,7 +71,30 @@ type Client struct {
 	uiVersion             string
 	networkChangeListener listener.NetworkChangeListener
 
+	stateMu       sync.RWMutex
 	connectClient *internal.ConnectClient
+	config        *profilemanager.Config
+	cacheDir      string
+}
+
+func (c *Client) setState(cfg *profilemanager.Config, cacheDir string, cc *internal.ConnectClient) {
+	c.stateMu.Lock()
+	defer c.stateMu.Unlock()
+	c.config = cfg
+	c.cacheDir = cacheDir
+	c.connectClient = cc
+}
+
+func (c *Client) stateSnapshot() (*profilemanager.Config, string, *internal.ConnectClient) {
+	c.stateMu.RLock()
+	defer c.stateMu.RUnlock()
+	return c.config, c.cacheDir, c.connectClient
+}
+
+func (c *Client) getConnectClient() *internal.ConnectClient {
+	c.stateMu.RLock()
+	defer c.stateMu.RUnlock()
+	return c.connectClient
 }
 
 // NewClient instantiate a new Client
@@ -93,6 +119,7 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 
 	cfgFile := platformFiles.ConfigurationFilePath()
 	stateFile := platformFiles.StateFilePath()
+	cacheDir := platformFiles.CacheDir()
 
 	log.Infof("Starting client with config: %s, state: %s", cfgFile, stateFile)
 
@@ -124,8 +151,9 @@ func (c *Client) Run(platformFiles PlatformFiles, urlOpener URLOpener, isAndroid
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder)
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	return c.connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile)
+	c.setState(cfg, cacheDir, connectClient)
+	return connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile, cacheDir)
 }
 
 // RunWithoutLogin we apply this type of run function when the backed has been started without UI (i.e. after reboot).
@@ -135,6 +163,7 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 
 	cfgFile := platformFiles.ConfigurationFilePath()
 	stateFile := platformFiles.StateFilePath()
+	cacheDir := platformFiles.CacheDir()
 
 	log.Infof("Starting client without login with config: %s, state: %s", cfgFile, stateFile)
 
@@ -157,8 +186,9 @@ func (c *Client) RunWithoutLogin(platformFiles PlatformFiles, dns *DNSList, dnsR
 
 	// todo do not throw error in case of cancelled context
 	ctx = internal.CtxInitState(ctx)
-	c.connectClient = internal.NewConnectClient(ctx, cfg, c.recorder)
+	connectClient := internal.NewConnectClient(ctx, cfg, c.recorder)
-	return c.connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile)
+	c.setState(cfg, cacheDir, connectClient)
+	return connectClient.RunOnAndroid(c.tunAdapter, c.iFaceDiscover, c.networkChangeListener, slices.Clone(dns.items), dnsReadyListener, stateFile, cacheDir)
 }
 
 // Stop the internal client and free the resources
@@ -173,11 +203,12 @@ func (c *Client) Stop() {
 }
 
 func (c *Client) RenewTun(fd int) error {
-	if c.connectClient == nil {
+	cc := c.getConnectClient()
+	if cc == nil {
 		return fmt.Errorf("engine not running")
 	}
 
-	e := c.connectClient.Engine()
+	e := cc.Engine()
 	if e == nil {
 		return fmt.Errorf("engine not initialized")
 	}
@@ -185,6 +216,73 @@ func (c *Client) RenewTun(fd int) error {
 	return e.RenewTun(fd)
 }
 
+// DebugBundle generates a debug bundle, uploads it, and returns the upload key.
+// It works both with and without a running engine.
+func (c *Client) DebugBundle(platformFiles PlatformFiles, anonymize bool) (string, error) {
+	cfg, cacheDir, cc := c.stateSnapshot()
+
+	// If the engine hasn't been started, load config from disk
+	if cfg == nil {
+		var err error
+		cfg, err = profilemanager.UpdateOrCreateConfig(profilemanager.ConfigInput{
+			ConfigPath: platformFiles.ConfigurationFilePath(),
+		})
+		if err != nil {
+			return "", fmt.Errorf("load config: %w", err)
+		}
+		cacheDir = platformFiles.CacheDir()
+	}
+
+	deps := debug.GeneratorDependencies{
+		InternalConfig: cfg,
+		StatusRecorder: c.recorder,
+		TempDir:        cacheDir,
+	}
+
+	if cc != nil {
+		resp, err := cc.GetLatestSyncResponse()
+		if err != nil {
+			log.Warnf("get latest sync response: %v", err)
+		}
+		deps.SyncResponse = resp
+
+		if e := cc.Engine(); e != nil {
+			if cm := e.GetClientMetrics(); cm != nil {
+				deps.ClientMetrics = cm
+			}
+		}
+	}
+
+	bundleGenerator := debug.NewBundleGenerator(
+		deps,
+		debug.BundleConfig{
+			Anonymize:         anonymize,
+			IncludeSystemInfo: true,
+		},
+	)
+
+	path, err := bundleGenerator.Generate()
+	if err != nil {
+		return "", fmt.Errorf("generate debug bundle: %w", err)
+	}
+	defer func() {
+		if err := os.Remove(path); err != nil {
+			log.Errorf("failed to remove debug bundle file: %v", err)
+		}
+	}()
+
+	uploadCtx, cancel := context.WithTimeout(context.Background(), 2*time.Minute)
+	defer cancel()
+
+	key, err := debug.UploadDebugBundle(uploadCtx, types.DefaultBundleURL, cfg.ManagementURL.String(), path)
+	if err != nil {
+		return "", fmt.Errorf("upload debug bundle: %w", err)
+	}
+
+	log.Infof("debug bundle uploaded with key %s", key)
+	return key, nil
+}
+
 // SetTraceLogLevel configure the logger to trace level
 func (c *Client) SetTraceLogLevel() {
 	log.SetLevel(log.TraceLevel)
@@ -214,12 +312,13 @@ func (c *Client) PeersList() *PeerInfoArray {
 }
 
 func (c *Client) Networks() *NetworkArray {
-	if c.connectClient == nil {
+	cc := c.getConnectClient()
+	if cc == nil {
 		log.Error("not connected")
 		return nil
 	}
 
-	engine := c.connectClient.Engine()
+	engine := cc.Engine()
 	if engine == nil {
 		log.Error("could not get engine")
 		return nil
@@ -300,7 +399,7 @@ func (c *Client) toggleRoute(command routeCommand) error {
 }
 
 func (c *Client) getRouteManager() (routemanager.Manager, error) {
-	client := c.connectClient
+	client := c.getConnectClient()
 	if client == nil {
 		return nil, fmt.Errorf("not connected")
 	}

client/android/platform_files.go

@@ -7,4 +7,5 @@ package android
 type PlatformFiles interface {
 	ConfigurationFilePath() string
 	StateFilePath() string
+	CacheDir() string
 }

client/internal/connect.go

@@ -94,6 +94,7 @@ func (c *ConnectClient) RunOnAndroid(
 	dnsAddresses []netip.AddrPort,
 	dnsReadyListener dns.ReadyListener,
 	stateFilePath string,
+	cacheDir string,
 ) error {
 	// in case of non Android os these variables will be nil
 	mobileDependency := MobileDependency{
@@ -103,6 +104,7 @@ func (c *ConnectClient) RunOnAndroid(
 		HostDNSAddresses:      dnsAddresses,
 		DnsReadyListener:      dnsReadyListener,
 		StateFilePath:         stateFilePath,
+		TempDir:               cacheDir,
 	}
 	return c.run(mobileDependency, nil, "")
 }
@@ -338,6 +340,7 @@ func (c *ConnectClient) run(mobileDependency MobileDependency, runningChan chan
 			log.Error(err)
 			return wrapErr(err)
 		}
+		engineConfig.TempDir = mobileDependency.TempDir
 
 		relayManager := relayClient.NewManager(engineCtx, relayURLs, myPrivateKey.PublicKey().String(), engineConfig.MTU)
 		c.statusRecorder.SetRelayMgr(relayManager)

client/internal/debug/debug.go

@@ -16,7 +16,6 @@ import (
 	"path/filepath"
 	"runtime"
 	"runtime/pprof"
-	"slices"
 	"sort"
 	"strings"
 	"time"
@@ -31,7 +30,6 @@ import (
 	"github.com/netbirdio/netbird/client/internal/updater/installer"
 	nbstatus "github.com/netbirdio/netbird/client/status"
 	mgmProto "github.com/netbirdio/netbird/shared/management/proto"
-	"github.com/netbirdio/netbird/util"
 )
 
 const readmeContent = `Netbird debug bundle
@@ -234,6 +232,7 @@ type BundleGenerator struct {
 	statusRecorder *peer.Status
 	syncResponse   *mgmProto.SyncResponse
 	logPath        string
+	tempDir        string
 	cpuProfile     []byte
 	refreshStatus  func() // Optional callback to refresh status before bundle generation
 	clientMetrics  MetricsExporter
@@ -256,6 +255,7 @@ type GeneratorDependencies struct {
 	StatusRecorder *peer.Status
 	SyncResponse   *mgmProto.SyncResponse
 	LogPath        string
+	TempDir        string // Directory for temporary bundle zip files. If empty, os.TempDir() is used.
 	CPUProfile     []byte
 	RefreshStatus  func() // Optional callback to refresh status before bundle generation
 	ClientMetrics  MetricsExporter
@@ -275,6 +275,7 @@ func NewBundleGenerator(deps GeneratorDependencies, cfg BundleConfig) *BundleGen
 		statusRecorder: deps.StatusRecorder,
 		syncResponse:   deps.SyncResponse,
 		logPath:        deps.LogPath,
+		tempDir:        deps.TempDir,
 		cpuProfile:     deps.CPUProfile,
 		refreshStatus:  deps.RefreshStatus,
 		clientMetrics:  deps.ClientMetrics,
@@ -287,7 +288,7 @@ func NewBundleGenerator(deps GeneratorDependencies, cfg BundleConfig) *BundleGen
 
 // Generate creates a debug bundle and returns the location.
 func (g *BundleGenerator) Generate() (resp string, err error) {
-	bundlePath, err := os.CreateTemp("", "netbird.debug.*.zip")
+	bundlePath, err := os.CreateTemp(g.tempDir, "netbird.debug.*.zip")
 	if err != nil {
 		return "", fmt.Errorf("create zip file: %w", err)
 	}
@@ -373,15 +374,8 @@ func (g *BundleGenerator) createArchive() error {
 		log.Errorf("failed to add wg show output: %v", err)
 	}
 
-	if g.logPath != "" && !slices.Contains(util.SpecialLogs, g.logPath) {
+	if err := g.addPlatformLog(); err != nil {
-		if err := g.addLogfile(); err != nil {
+		log.Errorf("failed to add logs to debug bundle: %v", err)
-			log.Errorf("failed to add log file to debug bundle: %v", err)
-			if err := g.trySystemdLogFallback(); err != nil {
-				log.Errorf("failed to add systemd logs as fallback: %v", err)
-			}
-		}
-	} else if err := g.trySystemdLogFallback(); err != nil {
-		log.Errorf("failed to add systemd logs: %v", err)
 	}
 
 	if err := g.addUpdateLogs(); err != nil {

client/internal/debug/debug_android.go

@@ -0,0 +1,41 @@
+//go:build android
+
+package debug
+
+import (
+	"fmt"
+	"io"
+	"os/exec"
+
+	log "github.com/sirupsen/logrus"
+)
+
+func (g *BundleGenerator) addPlatformLog() error {
+	cmd := exec.Command("/system/bin/logcat", "-d")
+	stdout, err := cmd.StdoutPipe()
+	if err != nil {
+		return fmt.Errorf("logcat stdout pipe: %w", err)
+	}
+
+	if err := cmd.Start(); err != nil {
+		return fmt.Errorf("start logcat: %w", err)
+	}
+
+	var logReader io.Reader = stdout
+	if g.anonymize {
+		var pw *io.PipeWriter
+		logReader, pw = io.Pipe()
+		go anonymizeLog(stdout, pw, g.anonymizer)
+	}
+
+	if err := g.addFileToZip(logReader, "logcat.txt"); err != nil {
+		return fmt.Errorf("add logcat to zip: %w", err)
+	}
+
+	if err := cmd.Wait(); err != nil {
+		return fmt.Errorf("wait logcat: %w", err)
+	}
+
+	log.Debug("added logcat output to debug bundle")
+	return nil
+}

client/internal/debug/debug_nonandroid.go

@@ -0,0 +1,25 @@
+//go:build !android
+
+package debug
+
+import (
+	"slices"
+
+	log "github.com/sirupsen/logrus"
+
+	"github.com/netbirdio/netbird/util"
+)
+
+func (g *BundleGenerator) addPlatformLog() error {
+	if g.logPath != "" && !slices.Contains(util.SpecialLogs, g.logPath) {
+		if err := g.addLogfile(); err != nil {
+			log.Errorf("failed to add log file to debug bundle: %v", err)
+			if err := g.trySystemdLogFallback(); err != nil {
+				return err
+			}
+		}
+	} else if err := g.trySystemdLogFallback(); err != nil {
+		return err
+	}
+	return nil
+}

client/internal/engine.go

@@ -140,6 +140,7 @@ type EngineConfig struct {
 	ProfileConfig *profilemanager.Config
 
 	LogPath string
+	TempDir string
 }
 
 // EngineServices holds the external service dependencies required by the Engine.
@@ -1095,6 +1096,7 @@ func (e *Engine) handleBundle(params *mgmProto.BundleParameters) (*mgmProto.JobR
 		StatusRecorder: e.statusRecorder,
 		SyncResponse:   syncResponse,
 		LogPath:        e.config.LogPath,
+		TempDir:        e.config.TempDir,
 		ClientMetrics:  e.clientMetrics,
 		RefreshStatus: func() {
 			e.RunHealthProbes(true)

client/internal/mobile_dependency.go

@@ -22,4 +22,8 @@ type MobileDependency struct {
 	DnsManager     dns.IosDnsManager
 	FileDescriptor int32
 	StateFilePath  string
+
+	// TempDir is a writable directory for temporary files (e.g., debug bundle zip).
+	// On Android, this should be set to the app's cache directory.
+	TempDir string
 }