I have a large ZHA network where devices in different rooms intermittently become unavailable, reconnect by themselves, and sometimes cannot be controlled individually.
The important observation is that Zigbee group commands still work reliably. For example, two IKEA TRÅDFRI drivers forming one PAX light both respond to a Zigbee group command even when Home Assistant reports one of them as unavailable.
This suggests that multicast/group communication still works, while individually addressed unicast communication, address resolution, or routing is failing.
The logs also show persistent Zigbee short-address conflicts.
Home Assistant and coordinator hardware
- Home Assistant Core: 2026.8.2
- Home Assistant OS: 18.2
- Supervisor: 2026.08.0
- Installation: x86-64 Home Assistant OS VM on PROXMOX
- Integration: ZHA
- Coordinator: Home Assistant Connect ZBT-2
- ZBT-2 firmware:
9.1.1.0 build 0 (20260818191539)(BETA) - Its a single family home with quite some distance to neighbors
I already upgraded from a SkyConnect to a ZBT-2. That didn’t really help so I updated the Firmware to the latest Beta.
Network size and device composition
The diagnostic contains 134 entries:
-
1 coordinator
-
88 Zigbee routers
-
45 end devices
Manufacturer distribution:
-
63 IKEA of Sweden devices
-
15 eWeLink devices
-
10 SONOFF devices
-
8 devices without manufacturer information
-
6 LUMI/Aqara devices
-
5 Wing devices
-
Various additional Tuya devices
Radio channels
Current configuration:
-
Zigbee: channel 21
-
2.4 GHz Wi-Fi: channel 1, fixed at 20 MHz
Unicast and routing errors
During approximately one hour in the newest log, the coordinator reports:
-
313 ×
ZIGBEE_NO_APS_ACK -
197 ×
ZIGBEE_SEND_UNICAST_NO_ROUTE -
50 ×
ZIGBEE_SEND_UNICAST_ROUTE_DISCOVERY_UNDERWAY -
31 ×
MAC_NO_ACK_RECEIVED -
Several delivery failures during the rejoin loops
Examples among the affected devices include:
-
0x3BC0: 16 MAC no-ack failures -
0xBB31: 15 MAC no-ack failures -
0x8B5D: 10 no-APS-ack failures -
0x0307: 10 no-route failures plus repeated route discovery -
0xA27C,0xA7E4, and0xEDDC: additional no-route and route-discovery failures
The radio’s hourly counters also show:
-
8,268 APS unicast packets received
-
2,617 successful APS unicast transmissions
-
605 failed APS unicast transmissions
-
2,813 MAC unicast retries
-
875 failed MAC unicast transmissions
-
147 route discoveries
-
150 neighbors added
-
150 neighbors removed
-
84 neighbors marked stale
-
11,537 network decryption failures
Until a few weeks ago everything worked perfectly for over a year. But for the last couple of weeks the entire network is just unstable and is driving me insane. I tried to analyze the large log files with ChatGPT and Claude but they only tell me that there an issues but not how to solve them. I looked through Reddit and this community but I haven’t found anything close to my issues.