I began to suspect I was living in a horror movie last night… My zigbee bulbs all started switching on an off at random intervals (all out of step with each other, even the two that I always control as a pair via a Zigbee group). Sometimes it would be just a second or two between switch off/ons, sometimes a minute or more - seemed truly random. It seemed to be just Zigbee devices playing up - I didn’t notice any other HA controlled devices (e.g. Wifi / Z-Wave connected) doing anything.
I use a CC2531 dongle and ZHA. HA OS install - core-2021.6.4/OS 6.0. All the problematic bulbs happen to be Osram - I didn’t have any non-Osram ones plugged in so don’t know if that’s relevant or not.
I’m contemplating a move to Zigbee2MQTT so earlier in the evening I’d been trying to get a second CC2531 dongle setup with Z2M on a different bit of hardware. I couldn’t get it to start so had just abandoned it thinking the dongle may be faulty (but left the dongle plugged in). It seems unlikely this was the cause but I can’t think of any other possible reason for this weird activity. This went on for an hour or two and even carried on after a host restart of my HA box. I really wasn’t in the mood to troubleshoot at the time, so I ended up just cutting the power to the lights and going to bed. When I restored power this morning, things seem to be back to normal.
I checked the logs and nothing stands out, but I’m still not particularly familiar with what is normal or not from zigpy. I’ll upload some log screenshots later (can’t do that from work PC).
In the history, there was no mention of these being switched on or off, or indeed anything else that might have triggered it or seemed relevant. The status of the lights wasn’t being updated in HA at all - history for individual lights didn’t reflect any of this happening.
For additional info, I’m in a rural location with no close neighbours (not helping the “feels like I’m in a horror film” vibe…), so no chance of picking up any local networks.
One other thing… I have been doing a lot of experimenting with local binding recently, trying (and failing) to get my Osram bulbs to reflect status changes from a bound remote back to HA. Not sure if there’s some way for the Zigbee mesh itself to get messed up as a result of this…
I couldn’t determine any pattern at all - seemed really random. I wouldn’t really call it “flickering” as it wasn’t on and off that quickly. Occasionally just one or two seconds between on/off, but more often many seconds or even minutes between.
I can’t rule out electrical issues for sure, but it doesn’t seem likely. The lights were not all on the same circuit, and other things on the relevant circuits didn’t seem to be affected.
Looks like the bulbs loose connectivity. The fact that they go off at that time might be an Osram peculiarity.
Maybe a range issue?
EDIT: Or they actually loose power, and the log shows when power is back and they rejoin the network. Hard to tell without matching the logs with the actual on/off timing
Thanks. Yeah, I should have paid more attention to the timings - my excuse was I just wanted to go to bed at this time!
Range sounds like a possibility. Is it possible that the other CC2531 stick I was experimenting with, while not actually connected to anything, was still causing some interference in the Zigbee network in some way?
I had left the other dongle plugged in to a running Raspberry Pi, so it would have been powered. Will plug it back in and see if the weirdness starts again.
Oooh! Extra stick powered up now. Nothing happened for a while, but I left my office at 17:30 with the lights off, and returned at 17:55 with both on and nothing related in the history! Log viewer didn’t show that time, but while I had the log viewer active a bulb turned off an then on again. It’s around the time of the last few lines in this screenshot.
OK, I think this confirms that it’s the second zigbee stick that’s the issue. I’ve searched the log today for instance of “joined” (I can’t see a similar one for dropping off). There were lots of these in the early hours of the morning when I initially had the other stick plugged in, then almost none for the rest of the day, until just after I’d plugged it back in, when I suddenly got lots of them again. The IDs are mostly for the bulbs that are closest to the second coordinator.
I’ve unplugged it again now, as I think that there’s not a lot more to find. I wonder if this is an issue generally with having a second co-ordinator active, or if it is because maybe there’s an issue with it. If it’s a general issue with having two going, perhaps it wouldn’t be an issue if I forced one of them onto another channel…
