Bug Description
1. Summary
com.amazon.map.service (component com.amazon.map.async.service.main) crashes in an infinite loop on Vega Virtual Device (VVD) — the Lifecycle Manager kills each newly-launched instance with reason AppNotResponding under 1 millisecond after launch, then immediately relaunches it, which then gets killed again, forever. This appears to starve the app’s JS thread (React Native / Hermes) of CPU, causing the app to freeze entirely (React stops committing UI updates) with no error surfaced to the app.
App Name: [App Name]
App Link on Amazon Appstore: N/A (app not yet published; found during in-development testing on VVD)
Bug Severity:
- Impacts operation of app
- Blocks current development
- Improvement suggestion
- Issue with documentation
- Other
2. Steps to Reproduce
- Build a React Native app with Kepler SDK
0.23.8128and install it on a Vega Virtual Device. - Launch the app normally (no special component or method call triggers this — it starts immediately on app launch).
- Start streaming device logs:
vega device start-log-stream. - Observe
com.amazon.map.service/com.amazon.map.async.service.maincrash-relaunch-crash continuously in the log stream from the moment the app launches. - While this loop is running, use the app normally (e.g. play a video) and observe the UI/JS thread eventually freezing.
3. Observed Behavior
Immediately after app launch, com.amazon.map.async.service.main is launched, then force-terminated by the Lifecycle Manager for AppNotResponding within under 1 millisecond:
07:41:56.656492 lcm-server:[Lcm.cpp:5104] appInst[514](com.amazon.map.async.service.main): Launched in pid 40962
07:41:56.657327 lcm-server:[Lcm.cpp:6616] Forcefully terminating process 40962 with reason AppNotResponding
07:41:56.663097 acr_core_dump: Process 'com.amazon.map.service:map_service_library' (PID=40962) crashed with signal 3
07:41:56.663279 acr_core_dump: Hit crash throttle, crash report skipped for process com.amazon.map.service (PID=40962). Please wait 60 seconds before observing new acrs of the same fingerprint.
This cycle repeats with a new PID every ~200-500ms, continuously, for as long as the app runs — thousands of occurrences were recorded in a single ~5 minute log capture.
Correlated with this, the app’s own React Native rendering appears to freeze entirely. MountingManager::SurfaceTelemetryLogger (emitted periodically by the RN renderer) reports the exact same stats repeatedly for 2+ minutes straight (e.g. an unchanging mutations counter), meaning React stopped committing any UI updates at all during this period. During this same window, video playback in the app also gets stuck (buffered position stops advancing entirely, no further playback events fire).
4. Expected Behavior
com.amazon.map.service should launch and run normally without being immediately flagged as unresponsive and killed. No real process can become “unresponsive” within under 1ms of being spawned, so this looks like a false-positive ANR detection / resource coordination bug in the Lifecycle Manager, not a genuine hang in the map service itself.
4.a Possible Root Cause & Temporary Workaround
Possible root cause: This looks like it may be a regression of a previously-fixed defect. The Vega SDK 0.21 release notes describe fixing a very similar-sounding issue:
Fixed the music playback issues by resolving a resource management conflict. Previously, map service was incorrectly terminated by LCM. Improved coordination between Resource Manager and Lifecycle Manager to prevent premature service termination.
This is being seen on SDK 0.23.8128, which is newer than 0.21 — so either this is a regression of that fix, or an unresolved variant of the same underlying Resource Manager / Lifecycle Manager coordination issue.
Workaround: Per the documented VVD troubleshooting guide (vega/0.23/kvd-issues.html, “The VVD service crashes, causing your app to freeze and become unresponsive”), running:
vega virtual-device stop
vega virtual-device start
vega device launch-app <app>
temporarily resolves it, but the crash loop recurs after continued use.
5. Logs or crash report
App/Device Logs: captured via vega device start-log-stream; excerpts included above. Full log file available on request.
Crash Logs / Crash Report: acr_core_dump entries are being throttled (“Hit crash throttle, crash report skipped”) for nearly all occurrences, so a full ACR was only generated for the very first crash in the sequence. Can pull additional detail via vega device copy-logs if needed.
6. Environment
SDK Version:
$ vega --version
Active SDK Version: 0.23.8128
Vega CLI Version: 1.2.18
App State: Foreground
OS Information (cat /etc/os-release via vega device run-cmd --command 'cat /etc/os-release'):
NAME="OS"
OE_VERSION="4.0.0"
OS_MAJOR_VERSION="1"
OS_MINOR_VERSION="1"
RELEASE_ID="19"
OS_VERSION="1.1"
BRANCH_CODE="TV Ship"
BUILD_VARIANT="user"
BUILD_TAGS="dev-keys"
BUILD_DATE="Thu May 28 17:38:10 UTC 2026"
7. Example Code Snippet / Screenshots / Screengrabs
N/A — this is not triggered by any specific component or API call in the app; it is a platform system service crash loop present from app launch onward, independent of the app’s own code.
Playback Issues
This is not a player-SDK-specific bug (it is a platform/OS-level service crash loop), but since it manifests as stalled video playback, for context:
Player SDK: HLS-based video player using standard IMA ad integration
Audio Codecs: AAC
Video Codecs: H.264
Manifest Types: HLS (m3u8)
Q: If applicable, please provide your media/content url — N/A (internal test content, not publicly accessible)
Q: Are there any special headers required to reproduce the issue you are facing? — N/A
Additional Context
This crash loop was discovered while investigating an intermittent “video does not resume after ads” bug in the app. That separate issue was traced to and fixed in the app’s own ad-integration code, but this com.amazon.map.service crash loop is a distinct, deeper issue that persists independent of ad playback and appears platform-level. It’s possible this crash loop is also a contributing factor to other intermittent instability/freezes observed on this VVD instance throughout development.