“SHORT” question: How do I reliably determine that a light-switch toggle was the result of a HA automation?
I’ve seen the suggestion to use trigger.to_state.context.parent == none, which works when the light turns on via motion, but it doesn’t work on another automation, where I turn the lights off via a time_pattern & motion check. Inspecting the trace.json, I don’t see anything I can key off of.
LONGER explanation:
GOAL (and I’m sure this is common):
Motion detected, lights turn on
Human powers off the switch. He must want it that way.
Do not automate anything with those lights until there has been no motion or further manual control of that switch for some cool down period
Automations I’ve got so far: Motion detected:
if manual_control_timer active, restart timer
else, turn on light
(this gets correctly identified as a HA automation, vs a human interaction)
Every 30 seconds
if no motion has been detected for 20min, and manual_control_timer not active, turn off light
This one is the problem - when this automation turns off the light, it triggers the next automation, bc I don’t know what to key off of to identify it as an automation vs a manual control
ON light switch ToGGLE
if trigger.to_state.context.parent_id == none (a human toggled it, but doesn’t seem reliable), we restart the manual_control_timer
This one restarts the timer regardless, bc parent_id is always none, and gets me stuck in an infinite manual control
Anyways - thank you so much for any advice. Other than that one detection, I think this setup would work pretty sweet
user_id will contain a long alphanumeric string if it was a user (otherwise none).
Search for context.user_id and you’ll find examples of how it is employed.
NOTE
Another technique, that is reliable, uses switches that can report their button actions (pressed, released, etc). Several z-wave, zigbee, and UPB switches support this feature.
Another way is to detect a dimmer switch’s brightness change (i.e. motion turns on the light to a preset level and if that level is changed by the user then it indicates the user wants manual control and the auto-off feature is temporarily disabled).
In all of my testing, user_id is always null (unless I trigger something from the dashboard). Maybe I missed something there.
About the zigbee reporting note: interesting! I didn’t know that. I’m using all wifi switches currently, but even if I was able to detect physical button pushes, I’d still want to group them as manual-actions alongside google-assistant comments.
Yes, there is to my knowledge no possibility to know if someone pushed the physical switch, or an automation triggered it. At least not with Tasmota switches.
You didn’t miss anything, that’s how it works. For user_id to contain a value, the user must login to Home Assistant and control an entity via the Lovelace UI.
If someone manually operates a switch, you can use the techniques I detect it.
It is improper. Disabling/re-enabling automations as part of normal operations is an anti-pattern.
Part of the guide is creating a “node red” user in Home Assistant. With that setup, I see in my history if node red controlled devices automatically versus manually controlled one’s.
I use automations and the Logbook indicates which devices were turned off by an automation using the automation’s name (not a generic Node-Red user). The devices controlled by a user via the UI indicate the user’s name. Devices controlled via their physical switches (manual control) show no identification.
Thanks I figured I was probably missing something easy here because I knew I saw that somewhere in the past. The history screen of the entity definitely doesn’t give as much detail as the logbook.
I have some node red automations that keep track of how a light was controlled by storing variables. I’m not sure if/how variables can be used in Home Assistant automations to possibly keep track of manual versus automatic control of a light though, but just another possible idea here.
I’ve also used your brightness change idea and that seems to be the simplest solution here if its a dimmer.
OP here. As a new HA user, I keep switching up my approach to motion-controlled lights.
I’ve since “refined” my logic to use timers instead of the “for” attribute.
This is the kitchen, for example.
when motion detected:
if timer is idle:
turn on lights
start timer
pause timer (immediately, so it's stuck at 20 minutes, for example)
when motion clear
start timer
when timer completes:
turn off lights
So far, I’ve really been liking the timer approach, as I can see the countdown in the UI, and it is easy to debug/test by manually finishing/starting timers. Other actions can also reset the timer, like opening the fridge, etc, which may be just out of motion-sensor FOV. gracefully handles the “purposefully turn off the light” scenario, as the timer keeps the light off as long as you’re in the room, + 20 minutes (or whatever).
Sometimes I sit at the kitchen island and work, motionless, so I may add an extra feature where the lights occasionally pulse, reminding me to move when the timer is approaching it’s end.
@123 I often look at these history entries as well and i see that an automation has turned them on or off. Is there a way to use this history info in an automation?
I have an AC which i can turn on with a IR remote not connected with HA. But when this happens and the AC is turned on i want to check in HA if an automation triggered the AC or it was manual. Because if it was manually turned on the tons of automation managing the AC should turn off or atleast dont trigger. They only need to trigger when the AC was turned on by HA.
Is there a way to make simple sensors with the history data and check if something was turned on by an automation or manually?
When an automation is triggered, the resulting trigger variable contains a property called context. The value of this property can serve to indicate who triggered the automation. However, it’s not necessarily a definitive indicator (meaning sometimes the trigger’s source can be ambiguous). Here’s a table indicating sources and values. and here’s an example that attempts to determine if the source was a manual operation.
Would you mind opening a new issue on github for this? The one linked unfortunately got auto-closed and nobody who is familiar with that system looked at it.
I did this using esphome’s switch state . For me, when the state of the physical switch is on , there is an automation which turns off the motion activated light automation and again turns them on when the switch state become off again. This way I can keep the lights on when I want by turning the physical switch on .
