Co-ordinator backup failure causes crash

What happened?

Zigbee2MQTT terminated when its scheduled coordinator backup timed out while querying a network-connected SMLIGHT Z-Stack coordinator.

Immediately before the failure, Z2M published a normal bridge health response with:

[2026-07-21 10:22:03] info: 	z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/bridge/health', payload '{"response_time":1784625723481,"os":{"load_average":[0.06,0.06,0.01],"memory_used_mb":2329.21,"memory_percent":61.656},"process":{"uptime_sec":518409,"memory_used_mb":160.5,"memory_percent":4.2484},"mqtt":{"connected":true,"queued":0,"published":6162059,"received":904},"devices": { ... } }'

Then the failure - with a stack that shows it's a backup process:

Error: SRSP - SYS - version after 6000ms
    at Object.start (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/utils/waitress.ts:68:23)
    at /app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/znp/znp.ts:256:45
    at AsyncMutex.run (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/utils/async-mutex.ts:17:26)
    at Znp.request (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/znp/znp.ts:249:27)
    at Znp.requestWithReply (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/znp/znp.ts:228:34)
    at AdapterBackup.getAdapterVersion (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/adapter/adapter-backup.ts:437:48)
    at AdapterBackup.createBackup (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/adapter/adapter-backup.ts:58:48)
    at ZStackAdapter.backup (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/adapter/zStackAdapter.ts:1021:49)
    at Controller.backup (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/controller/controller.ts:482:47)

And subsequently every automated restart failed with:

[2026-07-21 10:22:31] error: 	z2m: Error: Failed to connect to the adapter (Error: SRSP - SYS - ping after 6000ms)
    at ZStackAdapter.start (/app/node_modules/.pnpm/zigbee-herdsman@10.6.1/node_modules/zigbee-herdsman/src/adapter/z-stack/adapter/zStackAdapter.ts:95:27)

The uptime reported in the health payload 518409 is six days, nine seconds, which made me think this failure is probably something that happens on a daily schedule. Surre enough the backup timer is hard coded to daily:

        // Set backup timer to 1 day.
        this.backupTimer = setInterval(() => this.backup(), 86400000);

The SMLIGHT network interface and web UI remained reachable. The coordinator later recovered after the SMLIGHT/radio was restarted.

So... the underlying radio hang might be an SMLIGHT/firmware problem. But the scheduled call currently appears to discard the promise returned by backup(). So when the backup rejects, the rejection is not handled and terminates the Z2M process.

Environment:

• Zigbee2MQTT version: 2.12.1 (Home Assistant add-on)
• Herdsman: 10.6.1
• Adapter: SLZB-MR3
• Radio: CC2674P10

An AI search found several related historical reports, and reported that none appears to cover the current unhandled scheduled-backup rejection, though I haven't verified that info:

• https://github.com/Koenkk/zigbee2mqtt/issues/8711 — same timeout/unhandled rejection in 2021; resolved by updating old coordinator firmware
• https://github.com/Koenkk/zigbee2mqtt/issues/8452 — daily CC26x2 crash without a backup stack
• https://github.com/Koenkk/zigbee2mqtt/issues/31927 — backup failure during shutdown after an existing USB/serial failure
• https://github.com/Koenkk/zigbee2mqtt/issues/29116 — recurring SMLIGHT freezes without the daily-backup correlation

What did you expect to happen?

An unexpected adapter response during a backup should not crash the process.

How to reproduce it (minimal and precise)

I imagine you may be able to reproduce the problem by mocking adapter.backup() to reject when the scheduled callback runs.

Zigbee2MQTT version

2.12.1

Adapter firmware version

ZStack3x0 0x00212effff05da74 Revision: 20260310

Adapter

SLZB-MR3 - Radio: CC2674P10

Setup

Add-on in Home Assistant OS

Device database.db entry

No response

Debug log

No response

Notes

No response

Start in zigbee-herdsman/src/controller/controller.ts at the daily backup timer and trace how the scheduled backup promise is handled. Reproduce the case by mocking adapter.backup() to reject when the callback runs, then verify the rejection no longer terminates the process and that the existing restart behavior remains intact.