ZBT2 coordinator issues after migration – devices dropping, or unresponsive

[JackoAmmerlaan]

I recently replaced my LAUNCHXL-CC26X2R1 with a ZBT2.

The migration went smoothly, and there was no need to re-pair any devices. Initially, everything seemed fine.

While reviewing the logs, I noticed a large number of
ROUTE_ERROR_MANY_TO_ONE_ROUTE_FAILURE messages. This appears to be normal behavior, as Z-Stack does not report these errors in the same way. Therefore, I’m not sure whether these errors were also occurring with the old coordinator.

However, after running the new stick for almost a month, I’ve started to notice that some lights occasionally stop responding to commands. These failing devices are Philips HUE GU10 bulbs. In the logs, this shows up as:

error: 2m: Failed to ping 'SPOT EET RO'

The issue usually resolves itself after waiting a few minutes or by power-cycling the bulb. What’s particularly strange is that I have a cluster of four HUE GU10 lights located about 30 cm apart. Sometimes one or two of them suddenly become unresponsive, while the other two continue to work normally. Since all of them act as routers, I wouldn’t expect this behavior.

In addition, some battery-powered devices (SONOFF SNZB-03) occasionally drop off the network entirely and can only be recovered by re-pairing them. After re-pairing, they work fine for several days before losing their connection again. Not all SNZB-03 devices show this behavior—especially those located within one meter of the coordinator.

I also recently bought an eWeLink contact sensor. It pairs successfully, but drops off the network almost immediately after pairing. I’m not sure whether this is related to the issues described above, but it feels suspicious.

The ZBT2 stick is installed in the exact same location as the old coordinator. Moving it to another location in the house isn’t an option, as the PC running Zigbee2MQTT is also connected to a P1 meter. Everything is located in the meter cupboard, and there is no Wi-Fi access point nearby.

Does anyone have an idea what could be causing these issues or how to resolve them?

One theory I have is that the ZBT2 is more powerful and may be communicating directly with some devices instead of routing through others. When that direct connection drops, it might take some time for the mesh to rebuild an alternative route.

Software & Environment Overview

Zigbee2MQTT version: 2.7.1
Frontend version: 2.4.2
zigbee-herdsman-converters: 25.83.1
zigbee-herdsman: 7.0.4
Node.js version: v22.16.0
Coordinator: EmberZNet (ZBT2)
IEEE: 0x00124b001caab60b
Revision: 7.4.4
[GA]

Total number of devices: 101