I didn’t change anything and everything returned to normal at 10:00 PM, phew
What exactly does your scene do? (the old scene before the update)
What type of trigger does that scene use?
It’s not impossible that there’s a scene update bug ![]()
An easy solution to fix that would be to simply restart Gladys (but it’d be nice to understand where the bug comes from)
Yes, with the current Zigbee2mqtt integration that’s normal, I agree it’s not great.
@AlexTrovato is working on improving the configuration of this integration.
Good to hear it’s back ![]()
Next time, make a separate topic on the forum, it’s complicated to mix subjects ^^
Here’s my current scene :
The old scene did exactly the same thing, but I just added the state feedback of the shutters’ position (and therefore the Telegram message that goes with it); otherwise the behavior wasn’t changed.
I restarted my NUC yesterday (without any particular effect) — I don’t know if that has the same effect as restarting Gladys?
By any chance didn’t you modify the scene right in the 10 minutes after sunset?
Because, in itself, modifying a scene that’s already started will only change the next scene and not the scene currently running!
(But still, if the sunset had already started, you wouldn’t have 2 executions of the scene)
I’ll try to reproduce the bug (I’ve reread the code — it seems fine to me though)
Yes, restarting the nuc is the same thing ![]()
I don’t remember at all when I modified my scene ![]()
But even if I had made that change then, I shouldn’t have had the bug only at that time, right? I made the change 10/15 days ago
Ok, super weird that you made a change 10/15 days ago, saw nothing for 10/15 days, and then, without touching anything, a message appears out of nowhere? ^^
That surprises me a lot
I agree that sometimes there are convoluted bugs, but are you sure about that?
Maybe the work to display an execution history of scenes should come sooner than expected
Is there a feature request for that?
So I do think there’s a problem, because I just noticed something a bit more « precise ».
Going back through my Telegram messages, I notice that this double-message behavior only appeared from the moment I disabled the scene 10 days ago. Even with the scene disabled I was receiving the message « Living room shutters closed ».
So I scrolled back further and before that I was only receiving the expected message (the one with the shutter position percentages).
I don’t know if that helps you.
Not sure a feature request was written, right? ![]()
Ok, that helps to understand the behavior (not normal for a scene to continue running after being disabled), however when rereading the code and doing some local tests, everything works fine…
You never changed the trigger? (still sunset?)
If my memory serves me well, I started right at the very beginning by setting a time (which I don’t remember anymore, of course
)
Then I made a change to set it to sunset, because a fixed time, with the changing length of the days, didn’t work.
Ok, that helps, but unfortunately I just tried to reproduce the bug in every possible way and it’s impossible ![]()
Changing a scene does indeed cancel the previous triggers for that scene… I’m a bit stumped here
That seems crazy to me, could you reproduce this bug?
Very good question… I can run some tests to see, yes; however, I’ll try to observe over several days so it won’t be a quick response ^^
No worries, the main thing is really to find a way to reproduce the bug!
It’s definitely possible that there’s a bug in this part of Gladys (it’s quite a complex part), and I imagine it must be a particular scenario (like creating a scene with a trigger, removing/adding a new trigger while the scene is disabled, something like that), we need to find it ![]()
I write this message with no small amount of embarrassment… ^^
I found the problem, and of course it was on my side.
When I migrated my setup to my NUC, I had left my RPi on its rack and had cut its power. Except the Pi was on when I went to check… (a child or maybe even me not paying attention, I don’t know).
Which meant my old Gladys was running, with its scenes…!!!
I wonder if the problem with my Zigbee devices also came from that; I’ll be able to run tests now that the RPi is out of commission!!
Sorry for my confusing messages ![]()
Glad you found it!
.
.
.
.
Can we laugh about it? ![]()
Don’t hold back at all ![]()
![]()
![]()
Hehe
It happens to everyone
Good that you found it — that’s what matters
I’m closing the thread!
