Devices keep dropping

[easybeat]

Does the issue really belong here?

• [X] I definitively want to report a bug within deCONZ or its REST-API

Is there already an existing issue for this?

• [X] I have searched the existing issues and there is none for the bug at hand

Describe the bug

As in another handful bugs described here:

https://github.com/dresden-elektronik/deconz-rest-plugin/issues/2710 https://github.com/dresden-elektronik/deconz-rest-plugin/issues/6201

I'm facing the same problem since I'm using Deconz for YEARS

5 Mins. after restart

15 Mins. after restart

As @dawiinci mentioned here: https://github.com/dresden-elektronik/deconz-rest-plugin/issues/6201 is this a bug in the firmware?

Steps to reproduce the behavior

1. start deconz-gui
2. wait 5 Mins. and check connections
3. wait 15 Mins. and check connections again

Expected behavior

No connections should be dropped

Screenshots

No response

Environment

• Host system:NAS with Proxmox 8.1
• Running method: Ubuntu Desktop 20.04 as a Proxmox VM
• Firmware version: 26780700
• deCONZ version: 2.22.00 / 19.9.2022
• Device: ConBee II
• Do you use an USB extension cable: yes
• Is there any other USB or serial devices connected to the host system? If so: USB3 ext. powered Hub + ZWave Stick

deCONZ Logs

No response

Additional context

No response

[easybeat]

Here my log files deconz log droping lights.txt

[Mimiix]

The discussion about closing issues is not going to be held here. If you want to have that, open a post on the forums.

For the latter, others are more suitable to answer it.

[manup]

Does this happen with the latest release too? There were quite a bunch of fixes in regards to poll and keep devices alive.

[easybeat]

Yes it also happens with 2.24.2 ...did an update yesterday on the proxmox machine but was exactly the same....and it continues to lose more and more lamps!!

Very very anoying!

[easybeat]

@manup And just to make clear I have 2 different systems which I have the same symptoms...just one gone completely by now. And this was running since about 3 years but always had this symptoms but whenever I power cycled the lights they came back on. But not now! Powercycle doesn't help.

I have a backup back from June which I try to restore now because there everything was ok.

[manup]

Currently I wouldn't recommend to switch between versions, but get to the cause where it fails. Can you please check if DDFs are enabled via Menu -> Panels -> Control, should look like:

The logs above don't show hard networks errors as most/all commands get confirmed but there also appears to be mostly legacy traffic.

With DEV and DDF logging enabled in Help -> Debug view we should see more. It would be helpful to then pick one device which goes unreachable and check what happens there.

[easybeat]

Yes DDF is on

DDF error log.txt here the ddf log

[Mimiix]

Yes DDF is on

DDF error log.txt here the ddf log

Are you sure the DEV log is enabled on that log ?

[easybeat]

sorry missed dev DDF error log.txt

[manup]

14:08:44:311 DEV Poll Busy /devices/0x00178801011445B8 ZCL response seq: 59, status: 0x86, cluster: 0x0008

Now this looks interesting, which modelid does the 0x00178801011445B8 have – can be seen in the Node Info Panel (it's a Philips / Signify device)? The 0x86 means Unsupported Attribute, perhaps the DDF tries to read attributes which are only available in newer lights firmware, have to check with the other devs.

[easybeat]

It is a Hue GO

[easybeat]

and it shows all ok with it

[manup]

And this Hue Go does drop off after a while as well? When you power-cycle a light which has dropped off does it come back and can be controlled?

[easybeat]

No the Hue GO always works it's the lights WAND

or now even some IKEA Range Extenders

all different manufactorers failing now

[manup]

Also can you please provide the firmware version for the Hue Go as shown in the Basic Cluster: When you double click on the node and then double click on Basic (first item in the list)

Then press "read" in the attributes table:

[manup]

No the Hue GO always works it's the lights WAND

or now even some IKEA Range Extenders

all different manufactorers failing now

Hmm can't see these in the logs (perhaps too short), but this is still the old deCONZ version? It would be really helpful to make the tests with the recent version.

[easybeat]

[easybeat]

No the Hue GO always works it's the lights WAND or now even some IKEA Range Extenders all different manufactorers failing now

Hmm can't see these in the logs (perhaps too short), but this is still the old deCONZ version? It would be really helpful to make the tests with the recent version.

well I can do the update to 2.24.2 if you want me too

[easybeat]

which logs do you want from 2.24.2? DDF and DEV? on startup?

[manup]

Ideally DDF + DEV APS, ZDP and INFO over 10-20 minutes. That's quite noisy but should show more what's going on

[easybeat]

ok...tell me how to get a 20 mins. log from Deconz-GUI...it seems only to copy 3 minutes

[easybeat]

on 2.24.2 looks the same after a few minutes they seem to connect

but if I go into Phoscon the lights show all greyed out

[easybeat]

Deconz log 1504-1507.txt first 3 minutes

[manup]

Oh sorry I forgot the UI is too limited here. Longer logs can be done via command line, for example start deCONZ via:

deCONZ --http-port=80 --dbg-aps=1 --dbg-info=1 --dbg-zdp=1 --dbg-dev=1 --dbg-ddf=1 | tee log.txt

[easybeat]

Deconz log 1504-1510.txt somehow it seems to work

[easybeat]

Deconz log 1507-1513.txt 6 Mins. steps I can do...hope that's ok

[easybeat]

Deconz log 1513-1519.txt another 6 Mins.

[easybeat]

and by the way the Ikea Range Extenders have not been recognized anymore

[easybeat]

Deconz log 1519-1525.txt that's 21 Mins. do you need more?

[manup]

Hmm interesting I'm looking after the 0x00158D0000B1EF40 (Wand04 ?) This one is seen by many devices but no attempt yet to contact it (.. Deconz.log.1507-1513.txt).

This one doesn't have a DDF yet nor the Paul Neuhaus lights, so there seems to be something off with the legacy code :thinking:

The Ikea repeaters however should work once they are queried (this may take a bit the first time they are used via DDF).

[easybeat]

the wand lights are all paul neuhaus lights...

but should this allow to crash the whole system?

[easybeat]

Deconz log 1525-1531.txt I think in this 6 minutes light wand 03 lost the green line

[easybeat]

Hmm interesting I'm looking after the 0x00158D0000B1EF40 (Wand04 ?) This one is seen by many devices but no attempt yet to contact it (.. Deconz.log.1507-1513.txt).

This one doesn't have a DDF yet nor the Paul Neuhaus lights, so there seems to be something off with the legacy code 🤔

The Ikea repeaters however should work once they are queried (this may take a bit the first time they are used via DDF).

Any more ideas how to get this system back to work...my friends are not very happy that they can't operate their lights :-(

[easybeat]

also the Ikea Range Extender are still not recognized

[easybeat]

@manup Any more help today?

[manup]

I'd propose we make a live session perhaps with Teamviewer tomorrow or Wednesday? In addition if you can send me the zll.db file (or a backup) to support@dresden-elektronik.de I can try to make DDFs for the missing devices to check if that solves the issue.

[easybeat]

Thanks Manu.

yes live session tomorrow would be helpful. I'm available from 08.00 - 11.00 GMT+1

I expect you are in Germany? I'm from Switzerland...so we can then also change to German...

I've sent the backup files to you support e-mail.

[manup]

Thanks for the backup that helped to debug where the loading stopped.

Unfortunately I can't manage to be available before 14/15.00. However I've compiled a test version which loads the Ikea repeater directly (there was also another related fix that's scheduled for v2.25.0) and also added a DDF for the CREEBLE JZ-RGBW-Z01 light. Can you please try if these two are working better and don't fall off the network.

If they work I'd like to create DDFs for the remaining devices, should max. 2-3 devices. There is still an open issue with the legacy code handling, but moving the old device integrations to DDF is good in any case. deconz_2.24.2-debian_armhf.zip (test version)

[easybeat]

Thank you for the new version. But how do I have to install this?

I run it on an Ubuntu 20.04 desktop and get this error message:

[easybeat]

Unfortunately I can't manage to be available before 14/15.00.

And yes I could also do these times...let me know

[manup]

Thank you for the new version. But how do I have to install this? I run it on an Ubuntu 20.04 desktop and get this error message:

Oh damn my bad, thought you're running on a Raspberry Pi. Here is the version for Desktop AMD64

deconz_2.24.2-debian_amd64.zip

[easybeat]

np...I'll try to install now

[easybeat]

It says now that a never version is already installed...should I try it on the machine with the 2.22.0 version?

[manup]

I guess that's the Ubuntu UI installer thing.. Better make this on the command line:

sudo dpkg -i deconz_2.24.2-debian_amd64.deb

[easybeat]

ok installed it from command line over the 2.22.0 installation...can also try on the box with 2.24.2 already installed

It looks not much different to me

in phoscon still all greyed out

here is the log from first 6 Mins. Deconz log 20231212 1455-1501.txt

[easybeat]

Deconz log 20231212 1501-1505.txt next 5 Mins.

[manup]

Cool thanks, the Ikea range extender and Greeble lights are loaded via DDF now (for Ikea repeater the actual name is shown which wasn't before so it did pick up the code changes here). I think next it makes sense to let it run for a while to see if it settles on these devices, I'll check the logs

[easybeat]

I don't think it works as it has lost a lot of green lines already to Ikea and Wand-Lights

[easybeat]

Deconz log 20231212 1508-1511.txt

[easybeat]

[Uploading Deconz log 20231212 1512-1515.txt…]()