OK, it took about 1/2 second for the Scene actions to start being logged, and about 1 second for all of them to run. That seems nominal.
Just what do you mean by "extremely slowly"?
Just an FYI that I'm monitoring the thread.
As Bruce alluded to above, SharpTools is sending the normal on/off commands to the virtual switch and the scene activation takes over from there.
Thank you for all you time on this. I totally understand where you are coming from, but I can assure you something is amiss. While attempting to demonstrate again, in a different view, how this is unique in my case, I stumbled on the issue occurring also from the activator button.
What we are seeing here is that I turned on the scene (goodnight) and only some of the lights changed state. It appears the scene is commanding them off (see "Dining Room Light" in logs, but the dining room light never turned off, as shown in the refreshed page for the scene.
I do not believe this is a z-wave timing issue, because I turn all of those on/off one at a time w/o any issues, as seen in the original video
I guess it could be a data overrun issue, where there isn't enough delay between sending each z-wave command. I would assume a protocol like z-wave to have implemented flow control to avoid this. That is pretty basic in any network protocol. I am making assumptions here, which I don't like to do, and trying to limit this to what I observe for what I have tested to try and get to the bottom of this.
I only have 3 gen 1 z-wave switches left on my network, and none of them are in this scene -- everything is z-wave plus 500 series or newer. I do have two Gen 1 smoke detectors -- I guess I can move them over to my integrations hub (C3) if you believe any of these older devices are the cause to the lost state change commands.
Thanks @josh . What I was wondering is if the SharpTools integration interferes because of the way it performs a refresh after sending the commands. But I have since discovered, for the first time, that the issue does also happen outside of SharpTools.
Why don't you try the metering option in Scene? When commanding a long list of devices in fast sequence, the mesh can be a problem. So see if metering the commands by 70 msec makes a difference or not. This has the hallmarks of a mesh network problem. It's not SharpTools, and it's not the app, and it's not your devices. That leaves your mesh...
How long since you did a Z-Wave repair?
1 Like
I never tried turning that on. I thought it had to do with monitoring the use of the scene. I will give it a try. Thanks.
I played with the setting, and 70ms did the trick. I tried a lower value, and had issues, so I set it to 75ms to be a little safer.
Thank you for responding to this. And I apologize for rehashing this. I did not understand the setting in the other discussions.
If you are interested in the original issue I raised, I can get a screen recording tomorrow.
From a most basic capture, here you can see the scene reports as being set:
and the switch here, shows as off:
This is not specifically what I reported, but it is related.
Did any of you ever figure this out?
I have similar issues every day with incomplete scene activations that show as activated/on.
I have really slowed metering (2000ms), no metering, optimization on/off, rules to repeatedly activate scene, etc.
Still, every day I have to pull scene up for it to force refresh, where it no doubt shows the delta and updates to "off" and then I manually hit activate scene again. Sometimes, several more times. 
1 Like
I believe it has to do with what was said earlier:
When the network or Z-wave radio gets too busy and some commands don't make it to the device (or the device ignores them), the scene will simply assume incorrectly that those devices match the scene. As long as a refresh causes the device to reply with it's actual state, the scene will then reflect that.
They have largely stopped making updates to "Scenes" in favor of the newer "Rooms" app.
My last experience with Rooms, however, seemed to show similar behavior--but that was a year or so ago.
I've toyed with building an app that would have "positive control" (requiring a response from all devices and forcing a refresh and retry if not set properly) but that's a lot of work I've not really dived into yet.
In the mean time, waiting a bit, forcing a refresh, waiting a bit, then sending another activate if the scene becomes "unset" in your app can be a workaround (possibly also looping thru this a few times with a delay between loops-but, that's been criticized as overly complex).
Any idea of what to do this via rule machine, etc.? I'm trying to avoid having to get my phone out every time to nurse it to completion. I have tried adding a rule to hit the endpoint of the scene in an attempt to have it 'load a page' but that didn't seem to work for calling the refresh either. A last resort that might work would be to integrate a tasker profile/action to do the same, externally to HE, and then repeat my trigger for 'activate scene per mode' rule.. but really want to stay internal if at all possible to remove failure opportunities.
Clearly, you must have device and/or mesh issues. First thing to do is to migrate your Scene to Room Lighting, which can import it directly. In Room Lights, you can use the logging to confirm that all device commands have in fact been sent. If they have, then there is a non-responsive device or a mesh issue standing in the way. Such issues would have to be fully resolved for any app to work reliably to turn on these lights.
Another thing you can look at is the All Off app in our public repo. This app will repeat off commands to devices until they actually turn off, or will report failures. This is a useful diagnostic tool to identify those devices that have problems responding reliably to commands -- in this case the off command.
This is completely false. Z-Wave has internal flow control and does send every command.
3 Likes
I am already using the "All Off app", but realized I had added a handful of new devices since originally setting it up and just added those.
I also imported all my scenes into Room Lighting as suggested and even though metering isn't an option (we discussed this before in other related posts due to mesh size), it seems to working significantly better than Scenes. I still usually have a device or so that isn't in correct state, and I trigger the scene / activate room light to complete it.
But more often than not, it's a GE receptacle which I don't think reports its status back to the hub very well (or at all), so I've got a refresh rule that will now run every 3 minutes to hit all those (and similarly, older) devices.
I'll monitor for a few more days, but it's really close to perfect now! Thanks!
I ended up throwing out my GE/Jasco switches because they were a problem on the network in general. If you want to use the GF devices (probably Gen 1 Z-Wave) There is a wave polling app to handle the refresh of these devices. try not to poll to frequently, as it can bring down z-wave performance overall.
1 Like
