I’ve been using the Conbee II stick on a Raspberry Pi 4 for months now - so far without any problems. Since last weekend, however, there have been increased connection failures when using Phoscon, i. This means that the browser suddenly shows that the connection to Phoscon can no longer be established. You are suddenly and abruptly logged out and the start page for logging in is displayed again.
Similarly, the Android HUE Essentials can only access the network etc. very sporadically - the message often appears that the bridge cannot be reached.
What have I already done?
A backup (Phoscon) imported from a month ago → same problem
A complete backup of the Raspberrypi from a month ago → same problem
Changed the SD card and reset it with the same backup mentioned above → same problem
Do you have any ideas? Is the Conbee II stick defective? Where can I find log files if necessary? In the aforementioned problems, the Raspberry Pi including the other services can be reached normally; I therefore rule out a physical connection problem.
If it’s from troubes with USB connections to your stick you could may be find some trace of USB connect/disconnect via dmesg -T | grep USB or dmesg -T | grep tty
If is a power issue may be could you get some trace via dmesg -T | grep voltage
You can read out errors via the Debug View in deCONZ. Under the menu (above) click on “Help”, then on “Debug View”. First select the items “Info”, “Error” and “Error L2” and copy the result into your answer.
That’s exactly the point. When I call up the deCONZ GUI (via sudo -E deCONZ), it is displayed and I can connect to the stick. The graphic with the connections builds up and after about 20 seconds deCONZ ends hard. The message “Invalid machine command” then appears on the terminal. In this respect I am not able to analyze the log output. So is there a way to redirect the log to a file? And if yes, how?
By the way: I have already done a new installation of deCONZ.
In fact, I’m not using any deconz extension and the Pi is connected via LAN.
No, basically I don’t want to use the GUI. The Raspberry Pi should (continue to) only be used as a server (“headless”). The problem is that deconz crashes after a short time (about 30 seconds) after start. This affects both the service deconz and the GUI variant. This behavior has only been occurring for a short time. Until then, the application had been running for months without any problems. There were also no changes in the ZigBee network. Only general system updates were done, although resetting to an old system status still leads to the error. So my guess is that some device within the ZigBee network is triggering the error and causing the software to crash.
In the meantime, my further analysis has shown that the IKEA roller blinds could possibly be causing the problems. If I delete these, the software will not crash. On the other hand, there have been no problems with the blinds for months. How can this be checked?
Addendum: It’s not the fault of the roller blinds from IKEA. The system crashes anyway. Is there really no useful debugging?
After countless tests, crashes etc. I have now found out that the problem is obviously with a specific Philips IRIS lamp. As soon as I integrate these into the network, the entire system crashes reproducibly. In order to rule out that there is a technical defect with exactly this lamp (although even this should probably not lead to the system crashing…), I bought another lamp. And the same problem occurs here too. How can you now make a meaningful debug here so that the problem is finally solved?