I’m currently on the latest release of Home Assistant0.85.1. I’m experiencing weird things with my stuff. Lights and switches are turning themselves on and off. I can’t figure out what is causing it. From what I gather it seems like only Z-wave stuff are affected. Stuff happens without me physically pressing any thing. It can happen while i’m doing nothing at all. I’ve checked my automations and I can’t really see that any of them would cause this. Some of the affected devices are not even included by name (or inside a group) in the automation file. It usually happens when I’m at home but it has happened once as far as I can tell while I was outside. There are currently only two people who have access to my home assistant. Myself and my wife. I am using SSL (https). I don’t think my password has leaked nor is it particularly weak… I think.
I was wondering if any of you guys have ever experienced anything similar. Any idea on where I could start looking? Below is what I’m currently using. I’m trying to give as much details as possible here. Please ask if I left something out. I hope I’ve attached all relevant files.
Host:
I’m running hassio inside a docker under ubuntu on a virtual machine (VMWare). Stuff were behaving weird back when I was running things off a rpi3 as well. I’ve had these problems for a long time!
Here is a list of the devices that are usually affected by this random on/off bug. Listed by Entity Id (if that helps). All these devices are z-wave.
light.movie_room_fan_level
light.movie_room_ceiling_lights_level
switch.living_room_stereo_power
switch.movie_room_power
switch.tv_room_power
switch.tv_room_htpc
switch.garage_power_plug
I know this might be asking a lot. But I’ve pretty much tried everything I can think of. So super grateful if anyone has a couple of minutes to give me some pointers here.
Also, analyze home-assistant.log. It’s there for a reason. If HA is doing this, there will be evidence in the log. Make sure DEBUG is enabled (which is the default.)
With Z-Wave devices, it’s possible they can communicate directly. E.g., if you have a Z-Wave motion sensor, it’s possible a Z-Wave light or switch is associated with it and being directly turned on or off by the motion sensor. I’ve seen people get bit by this before.
You may have inadvertently created associations between Zwave devices where one device directly triggers another. Those may not necessarily show up in the logs as triggers. This may take some time to track down but you can go through the Zwave configuration menu under “Node group associations”.
I have several intentional associations for when I always want two or more devices to react as one. For instance, I always want the counter lights to come on when the kitchen lights come on. The signal to turn the counter lights comes directly from the kitchen lights switch not from HA. In fact the counter lights come on milliseconds before the kitchen lights ramp up.
Exactly. If a light changes (and the log shows that it has a new state), but nothing in the log before that shows HA changing it (e.g., from an automation, script, etc.) then chances are it changed via another means (e.g., another z-wave device changed it via an association, a ghost flipped the switch manually, … ) Not saying the log will answer the question, but hopefully it will provide more data to analyze that might help narrow things down.
The OZW_Log.txt file can also be analyzed, although that’s a bit harder. Still, more data that might help provide clues.
Great Idea. I’ve looked through 'em all now. It took a bit of time. You might not be able to answer this but i’ll give it a go and see if you have any clue. A part from The “Aeotec ZW090 Z-Stick Gen5 EU (Node: 1 Complete)” which is the controller itself for z-wave, existed it almost every “lifeline” group association, which I guess is normal as the controller wants to stay informed of any triggered association. I’m only guessing though. The following existed that i didn’t add myself.
Node 40: MuiltiSensor 6.
Group: LifeLine
In this group: Unknown node (255: (0 ? 255.0 : 255))
That last part looks weird. Unknown node…? I wonder how I go about to remove this as the node is not selectable in the list. Any idea?
Other than that nothing else looked weird. Most of my group associations are empty as I usually execute most of my stuff through an automation.
I just checked pastebin, are you sure your spaces are correct in the automation file? It seems like some of your triggers are located under action? YAML can be picky about the spaces sometimes.
I don’t have the Multisensor so I don’t have any reference. Since this is a device that could easily be a trigger device for other lights or switches that could easily be in other rooms or odd locations, you might look at this as a suspect. I would spend times with the logs and see what happens at the same time as an odd light change before declaring it the problem. I don’t know how disruptive it might be in your situation, but you could cover it up, take it down, remove power etc. to see if the problem goes away. If it is the problem, remove node, add node and rename it to the old name should correct things.
Thanks for taking your time to have a look. I am not certain about anything really. When I do a “check config” everything checks out. Also, the last time when I experience these weird on/off problems I check all of the automations on the start page of Home Assistant to see if any of them were triggered. There is a field that says “Last triggered: x hours ago” or something like that. None of the listed automations on the start page of home assistant had been triggered. I’m not sure if you can really trust that simply by checking though.
The reason why stuff is organized a bit weird and not structured the same way in every automation is because some of them I made by using the automation guide. I didn’t like it though so I edited the ones after the first couple of entries by hand.
I would also look at your history, to see if you can spot a specific pattern. Is it the same time everyday? Is it completely random? Also, are your zwave devices connected to some cloud service (IFTTT, Stringify, or the manufacturer’s cloud)…if so, check to see if there is some kind of schedule or flow running within the app.
Thought that was the trick, but it did it again just after I finished. But it would still be nice to have a report that tells me which devices to look at. I have 103 devices to go through to check for associations. So a report would be nice.
Awesome. That seems like a perfect match with my problem because I had a 255 entry on my multisensor. I figured out how to remove this broadcast id. So all I can do now is just wait a couple of hours/days and see if the issue is solved.
How did you manage to remove the broadcast ID from the reporting group association? Did you end up excluding the device from the network and include it again?
) Not saying the log will answer the question, but hopefully it will provide more data to analyze that might help narrow things down.