Google Home says "Sorry, something went wrong" when activating a scene

I created a new scene that includes a Template Switch. When activating the scene via google voice command everything shuts on/off as it should but google returns “Sorry, something went wrong for (scene name)”. This only happens with scenes that includes a Template Switch.

Same goes when trying to control the Template switch individually, it all works as it should but google still replies with “Sorry, something went wrong”.

Nothing is broken but it’s just annoying to hear that reply. every single time. Anyone know how to solve this issue?

1 Like

This is still happening (for me, at least), and it happens on any scene I call for.
None of mine include a template switch.
The scene always runs anyway, and about 5 seconds later the GoogAssistant says “something went wrong.”
It’s as if Google is expecting an acknowlegement of some kind that never arrives, or arrives indicating error.
I’m surprised that there aren’t more posts like this, unless it’s something that only a few of them do.

1 Like

I have this issue when my scene contains either a Nest thermostat or a harmony remote (turn on only - off works fine). What i did was remove them from the scene, and then add an automation with the following trigger:

  - platform: event
    event_type: call_service
      domain: scene
      service: turn_on
        entity_id: scene.morning

Anything you add to this automation is effectively part of the scene, but Google isn’t waiting for some sort of reply.

1 Like

Another member suggested having Google activate a short wrapper script which triggers the desired scene.
I’ve found that solution to be effective.

I belive it’s effective because the script can report “completed” to Google almost immediately.
Google seems to expect the “completed” response within no more than a couple seconds from commands it passes to HA, and it seems to timeout (and complain “something went wrong”) if the response doesn’t arrive fast enough.

1 Like

That works too. My problem with scripts is that they can’t be assigned to a room, and if a “scene” isn’t assigned to a room, it only works for the user who set up the HA integration. No idea why. But if i want voice match and my partner to be able to call the scene, i need to use my solution above.

I like your solution, but am not sure if I fully understand the relationship of the automation and scene. That is, I’m unclear on what triggers it (i.e what is exposed to Google that Google ‘see’ and invokes, the scene or the automation).

Or is scene.morning an empty scene that Google sees (so it ‘runs’ fast enough for Google) and the hollow scene triggers the automation to do the actual work?

I put everything that works in a scene, in the scene. You call the scene from Google. The automation triggers automatically when the scene is activated, and i put things like service calls and problematic devices in there. I suppose you could leave the scene empty and put everything in the automation, yeah.

Ah, yes. That makes sense.
In my case, the lights are mostly Insteon and they can take up to 10 seconds to all finish responding.
(I guess the ‘Inst’ in ‘Insteon’ doesn’t stand for ‘instantaneous’)

So if I have a scene with, e.g. just lamps, and there are about 5 or more Insteon’s, the scene will take too long to run for Google.
What light-control technology do you use? (I am going to guess it’s not Insteon :wink:

Lutron caseta and a handful of wemo wall plugs. All work instantly (no cloud involved). My slow problematic stuff is Nest, Harmony and Tesla