Losing Ikea Timmerflotte sensors every few days

[phaze75]

Version

matter-server/0.5.7 (matter.js/0.17.0-alpha.0-20260312-920b3a4ec)

Node.js Version

?

Operating System

Home Assistant OS

Issue Description

I have four Ikea Timmerflotte sensors in my grid that have been rock solid since using matterjs-server before version 0.5.0, though I’m not sure of the exact version. Lately, they work flawlessly for a few days without errors, but then I regularly lose them - apparently during session renewal, judging by the logs. Once that happens, they stay offline until I reset them by removing the batteries, after which they reconnect instantly and work perfectly again for a few days. Three of these sensors are bound via Matter binding to three Eve Thermo 4th Gen radiator thermostats, and that Matter binding has also been rock solid for several months. All four Timmerflotte sensors are affected by the issue, so Matter binding doesn't seem decisive.

Today, I lost two Ikea Timmerflotte sensors (@1:b and @1:d) in quick succession, so I managed to capture both incidents in a 5000-line Debug mode log:

core_matter_server_2026-03-21T11-14-58.715Z.log

This is the log showing the reset/recovery of the two Timmerflotte sensors:

core_matter_server_2026-03-21T17-32-57.112Z.log

Steps to Reproduce

There are no concrete steps to reproduce the issue.

Server Logs

Server log attached to the incident description.

Device Information

VendorName:IKEA of Sweden
ProductName:TIMMERFLOTTE temp/hmd sensor
Commissioned:2025-12-10T13:11:47.170000
Last interviewed:2026-03-19T17:02:59.773000
Is bridge:false
Serialnumber:
Matter version:1.3.0
All device types:Ota Requestor / Power Source / Root Node / Temperature Sensor / Humidity Sensor

Additional Context

No response

Pre-submission Checklist

• I am running an Alpha/Beta version of the Matter.js server
• I have included the complete server logs (not truncated)
• I have checked existing issues to avoid duplicates

[phaze75]

Just now, another Timmerflotte sensor went offline. Same issue, same pattern. This is the only one without Matter binding. So this is not relevant.

[Apollon77]

Hi, in fact if they go down and stay offline then this is a device or network issue. How many active ecosystems you use aside HA?

[Apollon77]

Ok for @1🅱️

• 11:47:35.054 Subscriptioin timed out, we try to reach the device but not working
• 11:48:42.074 we declare device as offline, MDNS still knows the IP but does not respond, we start to try to reach the device further, max every 2min
• 11:58:39.025 MDNS records expire! We have no IPS anymore, we inform HA about unreachable

Thats it ... nothing we can do ... device went offline.

Effectively as info. There is nothing like a "session renewal" ... unless the device stops sending data and we need to re subscribe!

[phaze75]

Hi, in fact if they go down and stay offline then this is a device or network issue. How many active ecosystems you use aside HA?

Good morning, I use only one ecosystem - HA Matter Server Beta, HA OTBR Beta with ZBT-1 Beta. Which is stable with all other devices - 5 Myggbett, 1 Myggspray, 2 Alpstuga, 2 Bilresa, 3 Eve Thermo, 7 Eve Energy, 1 Eve Weather and last but not least not so much anymore with 4 Timmerflottes. The other sensors are in the same rooms as the Timmerflottes and in the same distance to the routing end devices, the Eve Energys and Alpstugas.

Since December the Timmerflottes have been rock solid and now out of a sudden they all become flaky?

[phaze75]

Regarding a potential networking issue with Thread, specifically the HA OTBR Beta App, there’s still no way to properly investigate this in the OTBR log due to the lack of useful timestamps in its entries. This makes it impossible to correlate anything with the Matter.js log. I first flagged the issue in early February on Discord (#otbr-beta) and have since opened a related issue: home-assistant/addons#4476

Given the inseparable end-to-end dependency of Matter and Thread in “Matter over Thread” devices, this is, in my opinion, a significant roadblock to diagnosing and understanding the root causes of device failures with Matter and Thread alike. I hate to be that guy, but this seems to be an issue with OHF’s handling of the OTBR Beta App’s workstream - the ability to properly work with log files is the very foundation of a Beta test.

[Apollon77]

Ok, as said ... Here btw the check for the second log:
You see that we try every 2min to reach the device - look for "4eb596deff1c37cc" like

�[2m2026-03-21 18:30:43.683 DEBUG �[0;1;90mMessageChannel �[0;90mMessage » �[34mfor: �[2;39mSC/Sigma1 �[0;34mretrans#: �[2;39m207 �[0;34mbackoff: �[2;39m2m �[0;34mid: �[35m•unsecured#4eb596deff1c37cc⇵bc3b✉0e21af2e ...

And 18:30:47.193 we got answer and directly started to re-establish the connection and were online again at 18:30:52.559

All worked as expected. The device went offline somehow.

Did you tried to replace the battery? maybe worth a try

[tannerdsilva]

I have not seen any issues with the Timmerflotte, I am operating nearly 10 of them.

[phaze75]

Ok, as said ... Here btw the check for the second log: You see that we try every 2min to reach the device - look for "4eb596deff1c37cc" like

All worked as expected. The device went offline somehow.

Yeah, I see. Thanks for checking!

Did you tried to replace the battery? maybe worth a try

No, but I can easily try, although the battery levels are around 80%. So this shouldn't be an issue.

As mentioned, the logical next step would be to move from the application layer down to the network layer and check the Thread network. If the network layer is inconclusive, the only thing left to investigate is the device itself, which might be crashing.

[tannerdsilva]

Timmerflotte is not the only IKEA sleepy I’ve seen that will dump its entire battery in an instant. Others will do it too. I’m also unsure what is causing that, but I know it’s nothing in my immediate control (it’s not my MLD control plane or anything like that).

[phaze75]

If a battery dump were the cause, it would be easy to spot. I’ve already checked all Timmerflotte battery voltages, and the lowest is at 2,6V, meaning 1,3V per cell - so still well above the nominal 1,2V for NiMH LADDAs. What I can confirm is that after a Matter.js App restart from time to time there was a straggler among my Ikea sleepies that needed several minutes more to get back online. After that I could notice that within those few minutes its battery level had dropped noticeably - but I never noticed a battery dump.