i am experiencing the issue that sometimes, randomly, matter devices get “unavailable”. I don’t know when this happens, but if it happens, one or more devices get unavailable at the exact same timestamp. It can happen throughout the day and in the last 4 weeks, this happend 3 times. The following devices were affected:
Aqara FP300
Ikea Timmerflotte
I also use Eve devices at home, they weren’t affected yet, but as i said, it seems random. I have 33 matter devices in my HA:
Ikea Timmerflotte (9)
Ikea Door/Window (7)
Eve Thermo (7)
Eve Energy (4)
Eve Motion (1)
Aqara FP 300 (4)
Last time 2 devices went unavailable, this time it was 5. Usually FP300 and Ikea Timmerflotte are affected.
From reddit a random guy said to restart Matter Server 2-3 times, it works for him but not for me. I take out the battery, wait 10s, put it back in and it works again. HA restart does also nothing for me.
Is that the right place to report this bug, or should i head over to github? Which logs might help to narrow down the issue?
I was having similar problems with random devices going unavailable, 4 or 5 at a time, several times a day. It looked like one router device would go unavailable, and several more would drop in the next second or so.
Thing is, the devices were scattered around the house, not all in the same area.
I ended up removing power from all the light bulbs, and taking the batteries out of the sensors and switches.
Using the Beta version of the Matter Server, I repeatedly “interviewed” the smart plugs, until all but 2 of them changed roles to become Routers.
I then gradually powered up the lights room by room over the course of a whole day, allowing each room to stabilse.
Finally, I put the batteries back in the sensors and switches, one by one.
I now have a pretty stable mesh, with just one device a day going unavailable in one room.
Seems to me that the “logic” used for creating the routing in Thread is not very good, with some devices refusing to see their closest neighbours, settling on weak links to devices much further away.
The devices that are still failing for me are the ones that won’t link to the smart plug Router 10 feet away, choosing instead a couple of light bulbs 25 feet away.
“Orange” links, which then fail.
The devices which get disconnected can be “far away” from the TBR or even in the same room. I also have a SLZB-MR1 but its not in my thread network currenlty. I was thinking about using OpenThread Border Router to make HA my TBR.
After 4 days of 100% availability, I had four Ikea GU10 lightbulbs go unavailable last night. All the devices were “routers”, and all had two routes back to the border router. All had one route with 100% availability.
Looking at the network map and the individual device routing, there was no single common point of failure (apart from the time that they went unavailable).
I have checked the history of every mains-powered Ikea device.
10 devices went unavailable yesterday at some point (5 different times).
The ikea plugs that went unavailable all recovered after a few minutes.
The E27 lights also revovered.
The GU10 lights didn’t recover, but all came back online after power-cycling.
Posting in case it helps others chasing FP300s that keep dropping offline.
Symptom: Both my FP300s would go unavailable together and stay down for hours/days. Battery 100%, both on the latest firmware, rest of my Matter fleet fine. Waking the sensor (button/motion) and restarting the Matter Server add-on did nothing.
What I found: Pulled the node data from the Matter Server and decoded the Thread mesh-local addresses (RLOC16 = routerID<<10 | childID). Turned out both FP300s were attached as children of one of my new IKEA KAJPLATS bulbs, which was acting as their Thread router/parent. The timeline confirmed it — the bulb dropped off Thread first, then both FP300 children went unavailable a few minutes later. They're sleepy end devices, so when their parent vanished, they were orphaned. (The "Matter can't resolve it / no address" errors are just a downstream symptom — orphaned devices stop renewing their SRP/mDNS records.)
The kicker: the bulb had power the whole time — it just fell off Thread while powered (a flaky router). Power-cycling the bulb brought it back, and both FP300s re-attached.
Hypothesis (where I'd love other input): This shouldn't strand a device for a day. Thread has parent re-selection — an end device that loses its parent should find a new one. I checked my mesh: it's router-rich (17 routers; the bulb alone hears 14, several at −16 to −32 dBm right next to the sensors). So backup parents were clearly in range the entire outage and the FP300 never used them. My conclusion is that the FP300's firmware just isn't aggressive about re-acquiring a parent — battery-tuned long poll interval, periodic parent-search likely disabled, stretched reattach backoff. The mesh redundancy exists; the device doesn't exercise it. So likely it's an Aqara config/firmware issue.
Takeaways:
If your FP300s keep dropping, check what Thread router they parent off — if it's a smart bulb (especially a switchable/flaky one), that's your real problem, not the sensor.
Adding more routers won't help if your mesh already has coverage — this is a device failover behavior, not a coverage gap.
Short-term fixes: keep the parent router rock-solid (don't let a bulb be critical Thread infra), or power-cycle the sensor to force a fresh attach.
Has anyone gotten Aqara to confirm/tune the FP300's parent-search behavior, or seen newer firmware improve reattach?
I think maybe your Thread network is not robust enough, with not enough devices that can route.
In the ideal case, you would first install AC mains-powered Thread devices. At least one in every room but maybe more. Every one of these would have multiple, redundant paths to a Border router. Then you would have at least two border routers. Every powered Thread device should have about three or four other mains-powered devices connected to it. Only the mains-powered device can route packets in the network.
ONLY AFTER this is set up would you add battery-powered Thread devices. These can’t route and are only end devices.
But most of us don’t build Thread networks that way and our networks have single points of failure with no alternative routing, and they are fragile. Maybe your network is like that.
My experience thus far with IKEA, is that they do fall off the Thread network after oh say around 2 to 3 weeks; and once detached, and don't make any attempt to reattach.
As for Aqara, I don't have FP300, but do have FP2 and not had any issue with those.
