deCONZ does loop after upgrade to Trixie

deCONZ General Support
1

Hi there,

have upgraded my “Raspberry PI 3” based Phoscon appliance for my Raspbee II to “Raspian OS 13 (Trixie)”, overcoming all challenges, accept the deconz service(s) …

I have installed deconz-qt6 package out of beta repo and either one deconz-gui.service or deconz.service does start and restart constantly. Per journalctl I do not find much, mainly something like that:

...
An ExecStart= process belonging to unit deconz-gui.service has exited.

The process' exit code is 'killed' and its exit status is 11.
Jan 11 19:22:13 phoscon systemd[1]: deconz-gui.service: Failed with result 'signal'.
Subject: Unit failed
Defined-By: systemd
...

ExecStart is either deconz-gui.service:

...
ExecStart=/usr/bin/deCONZ --http-port=80
...

or deconz.service:

...
ExecStart=/usr/bin/deCONZ -platform minimal --http-port=80
...

For deconz-gui.service it doesn’t matter if desktop is Xorg or Xwayland based, same issue. On desktop I see for some seconds deconz GUI and my few devices are shown up, drawing their connections, before disappearing again and coming back again and again …

Before I start to restore to my old Bookworm setup I wanted to give it a try to fix that.

Any ideas here? Any input welcome, also by @deCONZ :slight_smile:

2

Can be a Segmentation fault ? “Signal” can be lot of thing no ?

Can you try to run it with command line ?
Just

/usr/bin/deCONZ -platform minimal --http-port=80

Or the other if you have the GUI ?
What is your deconz version, there is at least 2 “Segmentation fault” on version 2.31.2.

3

Yes, can be a lot things, but don’t find as said much in logs or per journalctl.

Version has to be deconz-qt6 - 2.32.2-beta for Raspberry OS 13 (Trixie).

Yes, I tried that also, headless & GUI, it remains running. But it doesn’t show up with my (paired) devices, just with the Raspbee II and openHAB does register / realize it as a complete new different deCONZ bridge. So looks like any internal number changed here, what doesn’t help either.

Restored the Phoscon appliance on Bookworm to a new SD card, bigger and faster. So, I would be able to engage the updated Trixie SD card for any investigations …

4

Because, if I remember correctly, the error message is more explicit if you run it from the command line.

You can run only 1 deconz instance, because the first one “lock” the gateway for it. So the second one can’t detect the gateway.
Strange you can run a second instance wihout your devices … Device database is stored on the host, so all deconz instances headless or desktop will use same. I think to have that need to run another OS instance completely with new user.

Another pach for the 2.31.2 to correct “crash” Fix raise to cleanup open TCP sockets and potential SEGV by manup · Pull Request #8455 · dresden-elektronik/deconz-rest-plugin · GitHub

5

Hi, that is a mistunderstanding, I never tried to run 2 instances parallel.

I have tried to run manually:

/usr/bin/deCONZ -platform minimal --http-port=80

as also:

/usr/bin/deCONZ --http-port=80

After each other with properly stopped deCONZ systemd services.

In both case it has not read in my stored configuration obviously. Best guess was, someting in env is missing while those tries.

Where is the configuration stored on Phoscon / deCONZ appliances? So the paired devices etc.?

6

The network configuration is stored on the gateway, and the device database is stored in /home/<User>/.local/share/dresden-elektronik/deCONZ/zll.db (for Raspbian)

1 Like