Describe the issue you are experiencing
Matter Server 9.2.0 repeatedly exhausts the available memory on a supported Home Assistant Yellow (2 GB) installation. The Linux OOM killer terminates the Matter Server container with signal 9 / exit code 137. Supervisor's watchdog then restarts it, eventually creating a rapid crash loop and making Home Assistant unavailable.
This installation recorded 102 Matter Server exit-137 events between 2026-08-22 and 2026-08-23. During the worst period, Matter Server restarted every 20–30 seconds. The host later rebooted without a clean shutdown trail.
Stopping Matter Server immediately restores stability. This closely resembles #4715 and #4695, but is still reproducible on Matter Server 9.2.0 on a Home Assistant Yellow.
What type of installation are you running?
Home Assistant OS
Which operating system are you running on?
Home Assistant Operating System
Which app are you reporting an issue with?
Matter Server
What is the version of the app?
9.2.0
Steps to reproduce the issue
- Run Matter Server 9.2.0 on Home Assistant Yellow with the normal Matter integration and existing Matter devices.
- Leave the app running for several hours.
- Available host memory is eventually exhausted.
- The kernel kills
app_core_matter_server; Supervisor reports exit code 137 and restarts it.
- Repeated restarts increase system pressure and Home Assistant becomes intermittently unavailable or the host reboots.
- Stop Matter Server and disable its watchdog; Home Assistant remains available and current-boot OOM events stop.
The failure is intermittent rather than an immediate startup crash. It is not the SIGSEGV behavior in #4766 and the container image starts normally, so it does not match the image-corruption report in #4720.
System Health information
version | core-2026.8.3
installation_type | Home Assistant OS
container_arch | aarch64
host_os | Home Assistant OS 18.2
board | yellow
supervisor_version | supervisor-2026.07.5
Matter Server | 9.2.0
Matter Server latest | 9.2.0
host/container memory limit | 1,977,737,216 bytes (~1.84 GiB)
Whisper and Smartbed MQTT were already stopped with automatic startup and watchdog disabled before this Matter Server crash loop was diagnosed.
Anything in the Supervisor logs that might be useful for us?
2026-08-23 02:42:54.146 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 02:43:24.074 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 02:43:58.636 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:19:30.574 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:19:46.493 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:20:04.869 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:20:30.587 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:20:56.369 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:21:17.507 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:21:39.073 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:21:59.060 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:22:22.402 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:22:46.719 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
2026-08-23 03:23:12.149 ERROR (MainThread) [supervisor.apps.app] App Matter Server exited with non-zero exit code 137
Supervisor recorded 102 such Matter Server exit-137 events in the available logs.
Anything in the app logs that might be useful for us?
Add-on version: 9.2.0
System: Home Assistant OS 18.2 (aarch64 / yellow)
Home Assistant Core: 2026.8.3
Home Assistant Supervisor: 2026.07.5
[03:21:19] INFO: Starting Matter Server...
[03:21:20] INFO: Using 'end0' as primary network interface.
[03:21:20] INFO: Successfully send discovery information to Home Assistant.
[07:21:35] WARNING: matter-server service exited with code 137 (by signal 9).
[03:21:41] INFO: Starting Matter Server...
[03:21:41] INFO: Using 'end0' as primary network interface.
[03:21:41] INFO: Successfully send discovery information to Home Assistant.
[07:21:55] WARNING: matter-server service exited with code 137 (by signal 9).
(The add-on displays UTC in the exit line and local time in its startup wrapper, which accounts for the apparent timestamp mismatch.)
Additional information
Kernel evidence conclusively identifies Matter Server as the OOM victim:
kernel: oom-kill: ... task_memcg=/system.slice/docker-af7a51c97f7997ecb633a26b08df19f6ec73bfcc6848ec22df2f6a507100dea2.scope,task=MainThread,... oom_score_adj:200
kernel: Out of memory: Killed process 550065 (MainThread) ... anon-rss:233808kB ...
dockerd: ... ep=app_core_matter_server net=host ...
The sequence repeats with newly recreated Matter Server container IDs. Examples showed approximately 148–263 MB container peaks before the kernel selected Matter Server under global memory pressure. A separate long-running Python process was later killed at approximately 868 MB RSS, confirming system-wide exhaustion on the 2 GB host.
After stopping Matter Server, setting boot to manual, and disabling its watchdog:
Matter Server state: stopped
Matter Server memory: 0 MB
Home Assistant API: reachable
Current-boot OOM records after containment: 0
Home Assistant's Matter integration subsequently starts the app again even while app boot is set to manual and watchdog is false, so keeping it contained may require disabling the integration as well.
Related reports:
The Home Assistant Yellow is supported hardware and cannot be upgraded beyond its installed 2 GB RAM, so increasing RAM is not an available resolution here.
Describe the issue you are experiencing
Matter Server 9.2.0 repeatedly exhausts the available memory on a supported Home Assistant Yellow (2 GB) installation. The Linux OOM killer terminates the Matter Server container with signal 9 / exit code 137. Supervisor's watchdog then restarts it, eventually creating a rapid crash loop and making Home Assistant unavailable.
This installation recorded 102 Matter Server exit-137 events between 2026-08-22 and 2026-08-23. During the worst period, Matter Server restarted every 20–30 seconds. The host later rebooted without a clean shutdown trail.
Stopping Matter Server immediately restores stability. This closely resembles #4715 and #4695, but is still reproducible on Matter Server 9.2.0 on a Home Assistant Yellow.
What type of installation are you running?
Home Assistant OS
Which operating system are you running on?
Home Assistant Operating System
Which app are you reporting an issue with?
Matter Server
What is the version of the app?
9.2.0
Steps to reproduce the issue
app_core_matter_server; Supervisor reports exit code 137 and restarts it.The failure is intermittent rather than an immediate startup crash. It is not the SIGSEGV behavior in #4766 and the container image starts normally, so it does not match the image-corruption report in #4720.
System Health information
Whisper and Smartbed MQTT were already stopped with automatic startup and watchdog disabled before this Matter Server crash loop was diagnosed.
Anything in the Supervisor logs that might be useful for us?
Supervisor recorded 102 such Matter Server exit-137 events in the available logs.
Anything in the app logs that might be useful for us?
(The add-on displays UTC in the exit line and local time in its startup wrapper, which accounts for the apparent timestamp mismatch.)
Additional information
Kernel evidence conclusively identifies Matter Server as the OOM victim:
The sequence repeats with newly recreated Matter Server container IDs. Examples showed approximately 148–263 MB container peaks before the kernel selected Matter Server under global memory pressure. A separate long-running Python process was later killed at approximately 868 MB RSS, confirming system-wide exhaustion on the 2 GB host.
After stopping Matter Server, setting boot to manual, and disabling its watchdog:
Home Assistant's Matter integration subsequently starts the app again even while app boot is set to manual and watchdog is false, so keeping it contained may require disabling the integration as well.
Related reports:
The Home Assistant Yellow is supported hardware and cannot be upgraded beyond its installed 2 GB RAM, so increasing RAM is not an available resolution here.