SLZB-06M TCP connection drops after ~30s (ECONNRESET) with Zigbee2MQTT ember adapter

[dangerbones]

Description of the issue

Hey all! :)

I am running Zigbee2MQTT 2.9.1 with the ember adapter using an SLZB-06M (EFR32MG21) over LAN.

The connection successfully initializes, Zigbee2MQTT loads all devices, and the Zigbee network starts correctly.
However after ~30 seconds the connection is reset and Zigbee2MQTT stops with:

Port Error: read ECONNRESET
ERROR_SERIAL_INIT
Adapter disconnected

The SLZB console simultaneously reports the client disconnecting.

This appears to be a stable startup followed by a dropped TCP socket, not a startup failure.

---

Environment

Zigbee2MQTT:
2.9.1

zigbee-herdsman:
9.0.9

Coordinator:
SLZB-06M

Chip:
EFR32MG21

Firmware:
EmberZNet 8.0.2 [GA]
build 397
EZSP protocol 14

Home Assistant OS
Raspberry Pi5

Zigbee2MQTT config
serial:
port: socket://192.168.0.173:6638
adapter: ember

MQTT:
mqtt:
server: mqtt://core-mosquitto:1883

---

Network test
From Home Assistant terminal:
nc -vz 192.168.0.173 6638

Result:
192.168.0.173:6638 open

Connectivity is confirmed.

---

Zigbee2MQTT logs

Startup works correctly:
zigbee-herdsman started (resumed)
Coordinator firmware version: EmberZNet 8.0.2
Currently 35 devices are joined
Connected to MQTT server
Zigbee2MQTT started

After ~30 seconds:
Port Error: read ECONNRESET
Fatal error, status=ERROR_SERIAL_INIT
Adapter disconnected, stopping

---

SLZB console logs:

SocketServer | [EFR32MG21] New client, id: 0
SocketServer | [EFR32MG21] Client disconnected, id: 0

---

Behaviour
1. Z2M starts successfully
2. Zigbee network resumes
3. Devices are loaded
4. ~30 seconds later the TCP socket is closed
5. Zigbee2MQTT shuts down

⸻

Expected behaviour

Connection to the SLZB socket should remain stable.

⸻

Additional info

USB mode was also tested but the device does not enumerate as a serial device (/dev/ttyACM* does not appear).

Multiple USB cables were tested.

Thanks all!

Addon version

Current version: 2.9.1-1

Platform

Core - 2026.3.1
Supervisor - 2026.03.0
OS - 17.1

Logs of the issue (if applicable)

No response

[dangerbones]

Additional testing:

Repeated testing from the HA terminal.

nc -vz 192.168.0.173 6638; echo $?

Sometimes returns:

192.168.0.173:6638 open
0

Other times returns:

1

This suggests the SLZB socket server intermittently stops listening on port 6638.

[dangerbones]

Quick update:

I now have a stronger reproduction case.

Zigbee2MQTT can sometimes start successfully with this config:

serial:
port: socket://192.168.0.173:6638
adapter: ember

When it does start, it:
• completes ASH/EZSP startup
• resumes the Zigbee network
• loads all 35 devices
• connects to MQTT
• then drops after ~30–35 seconds

Error at failure:

Port Error: read ECONNRESET
Fatal error, status=ERROR_SERIAL_INIT
Adapter disconnected, stopping

ASH counters at failure are clean:

CRC errors: 0
Comm errors: 0
ACK timeouts: 0

This suggests it is not a Zigbee RF / mesh issue.

Also, from the HA terminal, repeated checks to 192.168.0.173:6638 sometimes succeed and sometimes fail, which matches Zigbee2MQTT alternating between:
• ECONNREFUSED during startup
• ECONNRESET after successful startup

So at this point it looks like the SLZB LAN socket/server is intermittently unavailable or dropping the client session.

Current environment:
• Zigbee2MQTT 2.9.1
• zigbee-herdsman 9.0.9
• SLZB-06M
• EFR32MG21
• EmberZNet 8.0.2 [GA], build 397, EZSP 14
• Home Assistant OS on Raspberry Pi 5

Thanks! :)

[dangerbones]

Another update: I changed the Ethernet cable and thought stability improved, but the issue still reproduced. Zigbee2MQTT stayed up longer, passed normal startup, and exchanged normal traffic (RX=230, TX=290) before disconnecting with read ECONNRESET. The last EZSP frame was SEND_UNICAST, and ASH counters still showed CRC errors: 0, Comm errors: 0, ACK timeouts: 0. This suggests the session is being dropped during normal operation, not only during startup.

[ebinks123]

I am having the same issue with SLZB-06U

[sodhiperry]

I am having same issue with SLZB-MR2, so I bought a new SLZB-06U thinking it was a faulty coordinator. However, having same issues. I have tried literally everything from changing firmwares, to reinstalling z2m, repairing all devices, to opening all ports on my wireless modem. It was working prior to version 2.9.1-1.

It will work on restarting the coordinator for about 30 seconds, then fails everything after, can't even add new devices or anything.

How do you go about checking and seeing these errors? I would like to be able to debug and figure out where the issue lays.

[sodhiperry]

Also to add on, I have tried channel 11,15, and 20. Same issues.

[dangerbones]

@ebinks123 @sodhiperry any update?
I still can't get it working. Not sure if you've had any luck? :)

Thanks!

[dangerbones]

Maintainers - hey :)

I rolled back to Zigbee2MQTT 2.8.0-1 and the disconnect still occurs with the same failure pattern:

zh:ember:uart:ash: Port Error: read ECONNRESET
zh:ember:ezsp: Fatal error, status=ERROR_SERIAL_INIT. Last Frame: [FRAME: ID=38:"GET_EUI64" Seq=33 Len=13]

ASH counters remain clean at failure:

CRC errors: 0
Comm errors: 0
ACK timeouts: 0

So rollback to 2.8.0-1 did not resolve the issue, and the disconnect still appears to be the transport/socket session dropping rather than Zigbee RF/mesh corruption.

Thanks all!

[sodhiperry]

It may be a home assistant issue, as HA had an update also? It would be beyond my knowledge to see how HA would see connections and send information to the apps and network connected devices? Does this occur with USB or just an issue with network? as I can just connect the coordinator directly. Not what I want to do but maybe an option?

I'll go ahead and connect through usb and see if it improves anything, will let you know.

[sodhiperry]

Just to give an update, USB does work, but had to repair all my lights and devices. HOWEVER.....just to be sure, i had restarted the SLZB-MR01U coordinator and HA to make sure the process is not a temp fix......it took an unusual amount of time (10 minutes+), but it did respond and the lights have been working well and responding quickly.

I'll give another update in the next day or two on how that is going and/if there is any unusual activity.

[dangerbones]

@sodhiperry I did the same thing - moved to USB and it's fine.
Can't get it to start over network which is disappointing - but will keep trying and hope there's a HA update etc to see if it fixes it.

Thanks!

[sodhiperry]

There is a new HA update, however at this point I am not going to test that, as USB is working and i'm just using my other coordinator as a router to help push the signal.

[github-actions]

This issue is stale because it has been open 30 days with no activity. Remove stale label or comment or this will be closed in 7 days