#CASESession Timouts

1 messages · Page 1 of 1 (latest)

hollow pelican
#

Hi, 8 of 60 MoT devices are unavailable in Home Assistant:

  • Node ID - Device manufacturer and model - Name
  • 6 - EVE D&W - Badezimmer Fenster
  • 7 - EVE D&W - Kinderzimmer Fenster
  • 159 - EVE Thermo - Küche Thermostat
  • 222 - EVE Energy - Küche Kühlschrank
  • 270 - Nanoleaf NL54/GU10 - Abstellkammer Spot 1
  • 289 - Nanoleaf NL67/A19 - Büro Tischleuchte 2
  • 292 - EVE Thermo Control - Büro Thermo Control
  • 297 - Nanoleaf NL54/GU10 - Büro Spot 2
    All devices are pingable from the HA GUI. I already downgraded Matter Server from 6.5.1 to 6.5.0. But it didn't help. Next step could be a reboot of my HAOS (HA Yellow).
    Here is complete logfile since the last Matter server restart: https://dpaste.org/MAtNN
    Any idea?
mighty bolt
#

Please try a full reboot of your HAOS host

#

This smells like the issue we are seeing when multiple Apple border routers are present

hollow pelican
hollow pelican
mighty bolt
#

We have an open issue on Github where multiple folks are reporting issues with unavailable devices in HA (where apple home is still online) but only when there are multiple Apple border routers (homepods)

#

I've experienced this issue myself and in my case updating the homepod minis to OS 18 fixed the issues

hollow pelican
hollow pelican
mighty bolt
mighty bolt
hollow pelican
mighty bolt
hollow pelican
#

OK, didn't know that.

hollow pelican
#

Hi again, the HAOS reboot didn't help. Even though 7 of the the 8 unavailable devices are available, I now have 8 unavailable devices again:

  • Ping ok - 5 - EVE D&W - Küche Fenster
  • Ping ok - 20 - EVE Energy - Galerie PC
  • Ping ok - 158 - EVE Thermo - Badezimmer Thermostat
  • Ping ok - 159 - EVE Thermo - Küche Thermostat
  • Ping ok - 170 - EVE Thermo - Kinderzimmer Thermostat
  • Ping ok - 239 - EVE Weather - Aussen Norden
  • Ping ok - 256 - Nanoleaf NL67/A19 - Wohnbereich Tischleuchte 2
  • Ping not ok - 259 - Nanoleaf NL67/A19 - Küche Deckenleuchte 1
#

So, one of the 8 devices is not pingable. My Nanoleaf bulbs are paired to HA only, while all my EVE devices are apired to Apple Home and Home Assistant.

#

I am going to reboot my Apple Home Hubs (TBRs) as the next step.

mighty bolt
#

Yeah, please restart the homepods one by one to see what happens

#

btw the link to the log is wrong 😉

hollow pelican
mighty bolt
#

Yes, that log indeed looks exactly like what was reported on Github by others (and I experienced it as well).
communication just failing (hence all the case session timeouts) - that could be from RF issues but our current feeling is that this is a routing issue where somehow our traffic gets routed through the wrong BR. Do you still have some/all devices hooked up to Apple or just only to HA?

#

Anyways, I'd be interested to know if restarting any/all homepods would result in the issue clearing up

hollow pelican
# mighty bolt Yes, that log indeed looks exactly like what was reported on Github by others (a...

Yeah, all my EVE devices (39) are still paired to Apple Home and Home Assistant, while all my Nanoleaf devices (23) are paired to Home Assistant only. As already said, all my devices were reachable/pingable. After the HAOS reboot one device lost its connectivity to HA. It was not pingable from HA anymore. The other 7 devices/bulbs were pingable, but unavailable in HA. The not pinable device was a Nonaleaf bulb that was paired to HA only, so I couldn't see its connectivity state in Apple Home. However I can't say if good or bad Thread routes were used in my mesh. I do not think that it was an interference issue. I am using Thread channel 25, ZigBee channel 20 and WiFi 2.4Ghz channels 1 and 6 only.

hollow pelican
obsidian nova
#

i got 3 apple BR's (all on the 18 RC) but no where near as many end devices and its been stable, maybe try waiting a few hours until that drops and trying it?

hollow pelican
hollow pelican
mighty bolt
mighty bolt
mighty bolt
# hollow pelican Rebooted my 7 Apple Home Hubs. Everything is back available in Home Assistant no...

Yeah, I was afraid of that. Now question comes if the Apple Border Routers have an issue or we (still) have an issue with routing tables. HAOS host reboot didnt help so that pretty much rules out the stale routes and that leaves Apple BR. Please let me know how it goes after updating to OS18 although I must say my home network still isnt perfect - its much better with OS18 on the homepods but not super. Maybe as a test I'm going to shutdown all the homepods for a couple of days and only run on the ATV's

hollow pelican
mighty bolt
#

Ah yes, that looks MUCH better!

hollow pelican
mighty bolt
#

We're seeing too many vague issues with Apple BR's lately so until we get some tech contact with them, I'm afraid we're going to add an alert to our Matter integration that its not advised to use (multiple) Apple BR's for the time being.

hollow pelican
#

My HomePods are all connected to the 5GHz frequency band. So the Interference of channel 6 with channel 20 of my ZigBee network shouldn't harm my Thread network or the TREL communication.

#

My Philips Hue ZigBee network is rock stable. Always, never had any issue. 😉

hollow pelican
mighty bolt
#

Yeah, I know but we are out of options.

mighty bolt
mighty bolt
#

But still, when you have it on channel 20, it makes sense to disable your wifi channel 6 to make room for zigbee

#

If you only use the 2.4ghz band for IoT gear (and all good stuff is on 5ghz) then it is no problem to just set all your AP's to channel 1 for 2.4 ghz. Just send them to low transmit power so they dont hurt eachother

obsidian nova
#

much better overnight w/OS18 @hollow pelican ?

hollow pelican
#

Hi, yesterday evening/night I updated all my Apple devices to the latest OS 18.0. Had some problems with one of my HomePods. It didn't show the update. So I resetted it and could do the update. When all my tvOS (Apple HomePods and AppleTVs) were updated, I configured my primary Hub (one of my 2 AppleTV 4K 3rd Gen.). At that point the other other AppleTV was the active Hub. Nearly all devices became unavailable and then reconnected. All devices are connected and are working as expected.

#

But I see a lot of Failed to Send CHIP MessageCounterin my logfile for some devices. I already removed the Batteries or removed them from current. I did this for 3 devices:

170 - 00000000000000AA - EVE Thermo
220 - 00000000000000DC - EVE Motion
260 - 0000000000000104 - Nanoleaf A19

Here is the log: https://dpaste.org/QiGKg#L

Any idea what this MessageCounter wants to tell me?

#

Mainly the following 2 nodes produce this error message:

  • 00000000000000AA
  • 0000000000000104

Maybe I have to reset them... What do you think?

#

But as already said, all devices are connected and work as expected.

hollow pelican
hollow pelican
hollow pelican
#

Now I recognize that the CASESession Timeout issue is gone.

#

So, I have a new issue Failed to Send CHIP MessageCounter 😄

#

Maybe I have to reboot everything after all the updates to get a fresh/clean environment...

mighty bolt
# hollow pelican I have 4 APs in the house and 2 outside. I 'need' the density to have perfect 6G...

Sure, just be aware that you are hijacking the frequency for your zigbee as the 2.4ghz is very limited.
Hence my suggestion to set 2.4 ghz all APs to channel 1 (or any other channel which is not overlapping with zigbee or thread) with lowest transmit power and then use high power for your 5 (and 6) ghz. Maybe your Hue setup will greatly benefit from that as its radio is currently probably competing with your AP's on channel 6. Just a tip.

mighty bolt
hollow pelican
# mighty bolt Sure, just be aware that you are hijacking the frequency for your zigbee as the ...

Yeah, thanks for the tip. But as already said, my Philips Hue ZigBee setup is rock stable and my HomePods only use 5GHz. 6dBm is the lowest possible transmit power you can configure in a Unifi setup. I can set my APs to use channel 1 and channel 5. But I do not see any reason to do this in regards to Matter over Thread. My Thread channel is clean. 😉 Do you see any reason in regards to MoT?

mighty bolt
hollow pelican
# mighty bolt Nice, that means the network stabilized - now just keep an eye on the log. An in...

My Thread mesh seems to work fine. At least I do not see any downsides/issues at the moment. I rebooted my HAOS. Now I see the MessageCounter for some other devices. This behavior is definitely new. I never saw so many MessageCounter messages in my logfile before OS 18.0.

All devices came back within some minutes. Here is a new logfile:

https://dpaste.org/bPEMA

This time the following devices produce the MessageCounter messages:

  • 000000000000010A - Nanoleaf A19
  • 00000000000000DE - EVE Energy
  • 00000000000000AA - EVE Thermo

The EVE Thermo was already producing the issues before the HAOS reboot. Maybe a concidence. I will reboot all Apple Home Hubs and see what happens.

hollow pelican
#

OK, guys... Yesterday (nearly 24 hours ago), I ’rebooted’ all my Apple Home Hubs (TBRs) and half an hour later I also rebooted HAOS again. The meshing procedure completed round about 3 p.m. (15:00 o‘clock). Since that point the logfile got silent. Now when I look at the last 12 hours in my logfile (now it’s 12:30 o‘clock), its rock solid. There is only one CHIP ERROR message Subscription Liveness timeout with SubscriptionID, but some seconds later the Re-Subscription succeeded.

Here is my logfile: https://dpaste.org/RKbNb

So, ‘Rebooting‘ all Apple Home Hubs was the key after updating to tvOS 18.0. My procedure was as follows:

  1. Remove all Apple Home Hubs from current
  2. Disable the Matter integration
  3. Wait until Apple Home reports that no Home Hub is reachable
  4. Start the primary Apple Home Hub (hardwired AppleTV 4K 3rd Gen)
  5. Wait until Apple Home reports that a Home Hub is available
  6. Start the secondary Apple Home Hub (another hardwired AppleTV 4K 3rd Gen)
  7. Wait until you see that Apple Home Hub in the Apple Home settings
  8. Wait half an hour, until the meshing procedure of my 60 MoT devices nearly completed
  9. Start all Apple HomePods (4 HomePod Minis and 1 HomePod v2)
  10. Enable the Matter integration in HAOS

In the log file you some issues (CHIP_ERRORs) after enabling the Matter integration in HAOS. IMO these are related to the meshing procedure, that was still at work.

This is definitely something that needs to changed in future versions of Thread. It really takes too much time to establish the complete mesh. My expectation is that this happens within some seconds (maybe 5 minutes, but not much more). My Philips Hue system establishes the ZigBee Mesh within seconds. Let’s hope that this gets more stable and faster over time.

Thanks for your help.

mighty bolt
#

Nice! So my hunch is right that OS18 does indeed fix some Thread issues we've seen before but it just needed an additional restart ? Now if they can also fix the issue that OTA is not working.....
Please keep a good eye on it. My (also apple based) Thread network is still unstable at times. It runs smooth for a while and then goes into crazy mode again. I still suspect this to be some sort of routing (or routing table) issue where we contact a device through a suboptimal route

hollow pelican
#

After I had around 30 MoT devices, I established the routine described above to get my thread mesh up and running:

#1285150097209950249 message

#

This always worked in the past. When I got into the 50 MoT device range, I had to disconnect some of the MoT devices from Apple Home so that they are only paired with Home Assistant (40 MoT devices are paired with AH and HA, 20 MoT devices are only paired with HA). However, I still use the established method when the Thread Mesh network gets crazy.

#

Now with OS 18 I could test whether it also works if I simply restart all Apple TBRs at once and what it looks like when HA is also running in parallel during meshing.

#

Until now, I had the feeling that the mesh network quickly suffers from overload if HA is running at the same time as the meshing process. Do you see a relationship there? I could be wrong.