[Bug]: ForegroundServiceStartNotAllowedException in LightningNodeService.onCreate (dataSync quota, Android 15+)

[piotr-iohk]

What happened?

Google Play reports a production crash when LightningNodeService is created and calls startForeground() in onCreate.

Cause from stack:

android.app.ForegroundServiceStartNotAllowedException: Time limit already exhausted for foreground service type dataSync

Exception java.lang.RuntimeException:
  at android.app.ActivityThread.handleCreateService (ActivityThread.java:5929)
  ...
Caused by android.app.ForegroundServiceStartNotAllowedException: Time limit already exhausted for foreground service type dataSync
  ...
  at android.app.Service.startForeground (Service.java:776)
  at to.bitkit.androidServices.LightningNodeService.onCreate (LightningNodeService.kt:64)
  at android.app.ActivityThread.handleCreateService (ActivityThread.java:5916)

• Service type in manifest: foregroundServiceType="dataSync".
• On Android 15+ (API 35+), dataSync foreground services are limited to 6 hours total per 24-hour window; after that, starting another dataSync FGS throws this exception.
• Started from MainActivity via LaunchedEffect when wallet exists and notifications are granted → startForegroundService(LightningNodeService).
• Service uses START_STICKY — system may recreate the service after kill.
• Play Vitals (60-day): ~23% of top crash events (~36 events, ~14 users); mostly version 181 (2.2.0); Android 15/16 only; mostly foreground.

Expected behavior

App must not crash when the user opens Bitkit or the system restarts LightningNodeService after the dataSync FGS time budget is exhausted. Defer FGS start until allowed, use WakeNodeWorker where appropriate, or comply with Android 15+ dataSync limits.

Steps to Reproduce

1. Android 15 or 16 device, mainnet wallet, notifications on.
2. Keep LightningNodeService running in background until dataSync 6h / 24h budget is exhausted.
3. Open app — MainActivity starts service → crash in onCreate at startForeground.
4. Alternate: START_STICKY restart while quota exhausted.
5. Logcat: confirm Time limit already exhausted for foreground service type dataSync.

Related repro: may follow #980 (slow teardown after onTimeout).

Logs / Screenshots / Recordings

Exception java.lang.RuntimeException:
  at android.app.ActivityThread.handleCreateService (ActivityThread.java:5929)
  at android.app.ActivityThread.-$$Nest$mhandleCreateService (Unknown Source)
  at android.app.ActivityThread$H.handleMessage (ActivityThread.java:2924)
  at android.os.Handler.dispatchMessage (Handler.java:110)
  at android.os.Looper.loopOnce (Looper.java:273)
  at android.os.Looper.loop (Looper.java:363)
  at android.app.ActivityThread.main (ActivityThread.java:10060)
  at java.lang.reflect.Method.invoke
  at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run (RuntimeInit.java:632)
  at com.android.internal.os.ZygoteInit.main (ZygoteInit.java:975)
Caused by android.app.ForegroundServiceStartNotAllowedException: Time limit already exhausted for foreground service type dataSync
  at android.app.ForegroundServiceStartNotAllowedException$1.createFromParcel (ForegroundServiceStartNotAllowedException.java:54)
  at android.app.ForegroundServiceStartNotAllowedException$1.createFromParcel (ForegroundServiceStartNotAllowedException.java:50)
  at android.os.Parcel.readParcelableInternal (Parcel.java:5229)
  at android.os.Parcel.readParcelable (Parcel.java:5211)
  at android.os.Parcel.createExceptionOrNull (Parcel.java:3349)
  at android.os.Parcel.createException (Parcel.java:3338)
  at android.os.Parcel.readException (Parcel.java:3321)
  at android.os.Parcel.readException (Parcel.java:3263)
  at android.app.IActivityManager$Stub$Proxy.setServiceForeground (IActivityManager.java:7539)
  at android.app.Service.startForeground (Service.java:776)
  at to.bitkit.androidServices.LightningNodeService.onCreate (LightningNodeService.kt:64)
  at android.app.ActivityThread.handleCreateService (ActivityThread.java:5916)

Play Console: Vitals → ForegroundServiceStartNotAllowedException → LightningNodeService.onCreate.

Play report: https://play.google.com/console/u/1/developers/5690237613698939049/app/4973032799696771282/vitals/crashes/7cb781591f83cb407a22509c31455ef9/details?isUserPerceived=true&days=60

Bitkit Version

2.2.0 (version code 181), mainnet to.bitkit. Shipped tag v2.2.0 (2026-04-07), targetSdk = 36.

Device / OS

Android 15 (SDK 35), Android 16 Beta (SDK 36). Example devices: Samsung A16, A17x, PA3Q, others.

Reproducibility

Sometimes (<50%)

Additional context

Likely chain with #980: long-running dataSync FGS → onTimeout → slow stop → quota issues → restart/open app → this crash.

Suggested fix: do not call startForeground unconditionally in onCreate on API 35+ without checking allowance; fix onTimeout/stop paths in #980; revisit START_STICKY and architecture (#443).

References:

• https://developer.android.com/develop/background-work/services/fgs/timeout
• https://developer.android.com/about/versions/15/behavior-changes-15
• Optimise Node Foreground Service #443, [Bug]: ForegroundServiceDidNotStopInTimeException on LightningNodeService (Android 15+, 2.2.0) #980

[ovitrif]

(ported over from wrong issue where I shared this 1st), anyways, the fix in #978 addresses both crashes

Initial investigation context:

---

Bitkit Android FGS Crash Handoff

Repo: bitkit-android

Crash

Pixel 10 Pro logs show Bitkit dev crashing while creating LightningNodeService:

FATAL EXCEPTION: main
java.lang.RuntimeException: Unable to create service to.bitkit.androidServices.LightningNodeService
Caused by: android.app.ForegroundServiceStartNotAllowedException:
Service.startForeground() not allowed due to mAllowStartForeground false:
service to.bitkit.dev/to.bitkit.androidServices.LightningNodeService
at to.bitkit.androidServices.LightningNodeService.onCreate(LightningNodeService.kt:64)

Files Inspected

• app/src/main/java/to/bitkit/androidServices/LightningNodeService.kt
• app/src/main/java/to/bitkit/ui/MainActivity.kt
• app/src/main/AndroidManifest.xml
• app/build.gradle.kts
• app/src/main/java/to/bitkit/fcm/WakeNodeWorker.kt
• app/src/main/java/to/bitkit/fcm/FcmService.kt
• app/src/test/java/to/bitkit/androidServices/LightningNodeServiceTest.kt

Diagnosis

This is not an LDK/Rust crash. Android rejects the foreground-service promotion.

Important details:

• LightningNodeService.onCreate() calls startForeground(...) immediately.
• LightningNodeService.onStartCommand() returns START_STICKY.
• MainActivity.tryStartForegroundService() calls startForegroundService(...).
• Manifest declares FOREGROUND_SERVICE, FOREGROUND_SERVICE_DATA_SYNC, and android:foregroundServiceType="dataSync".
• App targets SDK 36.

Likely failure path:

1. Bitkit starts LightningNodeService.
2. The service calls startForeground(...) in onCreate().
3. Because the service returns START_STICKY, Android may later recreate it after process/service death.
4. If that recreation happens while the app is backgrounded, the recreated service calls startForeground(...).
5. Android rejects it with ForegroundServiceStartNotAllowedException.

The runCatching around startForegroundService(...) in MainActivity does not catch this because the crash happens later inside service creation.

Relevant Android Docs

• https://developer.android.com/develop/background-work/services/fgs/restrictions-bg-start
• https://developer.android.com/develop/background-work/services/fgs/troubleshooting
• https://developer.android.com/develop/background-work/services/fgs/timeout

Android 15+ also limits dataSync foreground services to 6 hours of background runtime per 24h window.

Proposed Fix

1. Move startForeground(...) out of onCreate().
2. Add an explicit ACTION_START_SERVICE.
3. Start the service from MainActivity with that action.
4. In onStartCommand(), only promote to foreground for the explicit start action.
5. For null or unexpected intents, do not promote; call stopSelf(startId) and return START_NOT_STICKY.
6. Replace normal START_STICKY behavior with START_NOT_STICKY.
7. Wrap foreground promotion in a helper that catches:
   • ForegroundServiceStartNotAllowedException
   • SecurityException
   • narrowly scoped RuntimeException
8. Only call setupService() / lightningRepo.start(...) after foreground promotion succeeds.
9. Keep onTimeout() stopping lightningRepo and the service.

Tests To Add Or Update

Update LightningNodeServiceTest to cover:

• explicit start action promotes service and calls lightningRepo.start
• stop action calls lightningRepo.stop and returns START_NOT_STICKY
• null intent does not call lightningRepo.start
• unexpected action does not call lightningRepo.start
• foreground promotion failure does not call lightningRepo.start

Validation

Run:

./gradlew compileDevDebugKotlin
./gradlew testDevDebugUnitTest
./gradlew detekt

Focused iteration command:

./gradlew testDevDebugUnitTest --tests LightningNodeServiceTest

Manual QA

• Install dev build on Pixel / Android 15+ or 16+.
• Enable notifications/background payments.
• Open app with wallet present; confirm node service starts.
• Background app; confirm no crash.
• Kill process/service while backgrounded; confirm no sticky recreation crash.
• Open app again; confirm service starts cleanly.
• Optional: shorten dataSync timeout with adb and verify onTimeout() stops cleanly.

Suggested Commit Messages

• fix: guard node service startup — 90%
• fix: prevent sticky node crash — 85%
• fix: harden foreground service start — 80%