Error: No running instances of com.amazon.dev.shell.service found

:backhand_index_pointing_right: Bug Description

1. Summary

Vega CLI device operations (vega device install-app, vega device launch-app, vega device uninstall-app, and any vega exec vda shell command) frequently fail with:

Vega operation FAILED due to an irrecoverable error: Incomplete: Error: No running instances of com.amazon.dev.shell.service found

vega exec vda devices still reports the Fire TV Stick as connected over TCP/IP. The CLI then treats the missing Developer Mode shell as irrecoverable and aborts, so we cannot install, launch, or debug the app until the shell service comes back (usually after reconnect/reboot). This blocks Vega development on a physical device.

App Name: Internal React Native for Vega development build (not published on the Vega Appstore)

App Link on Amazon Appstore: N/A

Bug Severity: Blocks current development

2. Steps to Reproduce

  1. Enable Developer Mode on a physical Fire TV Stick 4K Select (armv7l) and connect over TCP/IP:

    vega exec vda connect 192.168.1.222

  2. Confirm the device is listed:

    vega exec vda devices # 192.168.1.222:5555    device
    
  3. From the Vega app project directory, run a normal device workflow:

    vega device install-app --dir . -b Debug
    
    vega device launch-app --dir .
    
  4. Repeat install/launch over a development session (minutes to hours). The failure is intermittent. It also occurs immediately after a successful launch, with no app/code change.

  5. A reboot of the device clears the error for a period of time.

No specific React Native component or method call is involved. The failure happens in KeplerCLI DeviceManagerHandler before the app package is pushed or launched, on the first vda shell probe:

vda shell 'vdcm get com.amazon.devconf/system/device-info/device-type-id && uname -m && ...'

3. Observed Behavior

  • vega exec vda devices reports the device as connected.
  • The next vega exec vda shell (device-info, vpm install, vlcm, cat /etc/os-release, etc.) fails immediately with Error: No running instances of com.amazon.dev.shell.service found.
  • KeplerCLI wraps that as Incomplete and reports an irrecoverable failure. Example from the log:
2026-08-13 15:20:40 error [DeviceManagerHandler]: Error installing app from  (buildType: Debug) on device : Incomplete: Error: No running instances of com.amazon.dev.shell.service found
  • The same error also fails vega device launch-app and vega device uninstall-app.
  • It happened 44 times in vega-2026-08-13.log (install Debug/Release and launch), from ~10:09 through ~15:20.
  • Earlier the same day, the same commands succeeded (Command succeeded, architecture armv7l, Debug armv7 .vpkg installed and launched).
  • Reproduced live while filling this report:
    • vega exec vda devices192.168.1.222:5555 device
    • vega exec vda -s 192.168.1.222:5555 shell cat /etc/os-releaseError: No running instances of com.amazon.dev.shell.service found

4. Expected Behavior

If VDA lists the device as connected, vega exec vda shell should reach Developer Mode Shell (com.amazon.dev.shell.service) so install/launch/uninstall can proceed.

If the shell service is down, the CLI should:

  • report a recoverable/actionable error (restart shell service, reconnect, reboot), and/or
  • retry / wait for the service, rather than treating it as irrecoverable Incomplete.

This error should also be documented in VDA connection troubleshooting.

4.a Possible Root Cause & Temporary Workaround

A reboot of the Vega FireTV stick resolves the issue for a period of time.

Not a reliable workaround. Device reboot / re-enable Developer Mode sometimes helps, but the failure returns during a normal session.

5. Logs or crash report

Representative excerpts:

2026-08-13 10:09:42 debug [DeviceManagerHandler]: Error when running command: vda shell 'vdcm get ...'. Error: Error: No running instances of com.amazon.dev.shell.service found

2026-08-13 10:09:42 error [DeviceManagerHandler]: Error installing app from  (buildType: Debug) on device : Incomplete: Error: No running instances of com.amazon.dev.shell.service found

2026-08-13 13:01:41 error [DeviceManagerHandler]: Error launching app '' from . on device : Incomplete: Error: No running instances of com.amazon.dev.shell.service found

2026-08-13 13:38:18 error [DeviceManagerHandler]: Error installing app from  (buildType: Release) on device : Incomplete: Error: No running instances of com.amazon.dev.shell.service found

6. Environment

SDK Version: 0.23.8358 (from vega --version using .../0.23.8358/bin/vega and KEPLER_SDK_PATH). KeplerCLI 2.0.23038.0 / keplerCliVersion=0.200.0. Host also has 0.22.5600 on PATH.

React Native Version: RN 0.72

App State: N/A — failure is on the host CLI / device shell service, before or independently of the app. When it hits, install/launch cannot run.

OS Information: Could not collect via vega exec vda shell cat /etc/os-release because of this bug (Error: No running instances of com.amazon.dev.shell.service found) even though the device was listed as connected. From Settings → My Fire TV → About at the time of the issue (Aug 13, 2026, 3:25 PM MDT):

Device Type:        Amazon Fire TV Stick 4K Select
Software Version:   OS 1.2 (2101020054720)
Fire TV Home:       1.72.0.297062.0
Architecture:       armv7l (from earlier successful vda shell uname -m)
Connection:         TCP/IP 192.168.1.222:5555

Host: macOS 26.6.1 (Build 25G76), Darwin 25.6.0, ARM64 (Apple M3)

7. Example Code Snippet / Screenshots / Screengrabs

N/A for app code. Commands that hit this:

vega exec vda connect 192.168.1.222
vega exec vda devices
vega device install-app --dir . -b Debug
vega device install-app --dir . -b Release
vega device launch-app --dir .
vega device uninstall-app
vega exec vda shell cat /etc/os-release

:backhand_index_pointing_right: Playback Issues

N/A — not a playback issue.

Player SDK / Version / Codecs / Manifest / Content URL / Headers: N/A

:backhand_index_pointing_right: Additional Context

  • This is a new issue, it has not been present until today.
  • Vega SDK 0.23.8358 on macOS; Fire TV Stick 4K Select used all day over Wi‑Fi VDA (vega exec vda connect 192.168.1.222 plus vega exec vda reverse for Metro 8081 and 8097).
  • Failure is intermittent in the same session: install/launch can succeed, then fail minutes later with the device still listed.
  • CLI error text is incomplete (Error installing app from (buildType: Debug) on device, empty app name on launch), which makes the failure look like a bad project path even though the project is valid.
  • Reviewed Fix VDA Connection and Power Issues | Vega Troubleshooting , Resolve Fire TV Stick Issues | Vega Troubleshooting , and Resolve Vega Extension and CLI Issues | Vega Troubleshooting — none document this error or a recovery for a dead com.amazon.dev.shell.service while VDA still shows the device.

Hi @Jon_Page,

Thank you for the detailed bug report - the logs, environment details, and reproduction steps are very helpful.

We understand this is blocking your development workflow. We’re investigating the com.amazon.dev.shell.service crash on Fire TV Stick 4K Select running OS 1.2.

A few questions to help our investigation:

  1. Is this still happening consistently, or was it isolated to that one session?
  2. Are you running any background processes on the device (e.g., headless services, Metro bundler reverse ports) that might be contributing to memory pressure?
  3. Could you share the full vega-2026-08-13.log file? The 44 occurrences and timing pattern will help us narrow down the trigger.

In the meantime, when the shell service goes down you could try disconnecting and reconnecting the device:

vega exec vda disconnect 192.168.1.222:5555
vega exec vda connect 192.168.1.222

We’ve noted your feedback about the CLI error handling and the documentation gap - we’ve flagged both internally with the team.

Warm regards,
Aishwarya

It was happening all day, and yes, we are using the headless playback and I was working on our CC track implementation. Rebooting the device was the only fix I knew about, but I could try the disconnect

Hi @Jon_Page,

Thank you for the quick follow-up and for confirming the headless playback usage.

We’ve been investigating this on our end. Could you please share full device-side logs so we can confirm the root cause?

We’ll keep you posted. Feel free to reach back to me if you have any questions!

Thanks,
Aishwarya

I have shared logs with my amazon contact :+1: