I have a few scenes for my kitchen/dining room. They worked great until after the 2.2.4.147 update, not they are hit and miss if they work or not. Most of the time, I turn on a scene through SharpTools, and nothing happens, or one of 8 or so lights will change, and then the scene turns off. I have ignore off enabled, so it appears to stop there. Strangely, in the groups and scenes app, it will report a scene is active (but the switch for the scene reports off), while the devices are not in the correct state for the scene.
This is version 2.2.4.147 (upgraded from 2.2.4.145) which should have been a minor update, but it is clear this was a major update. Is there any chance someone can identify the source of all these reported issues with 147 and get it fixed quickly?
There weren't any changes to Scenes in this release, so I'd like to really understand what has changed in some detail.
If you have "Ignore off enabled" there should be no necessary relationship between the state of the Scene and the activation switch being off, since it's off state is ignored. The Scene is activated, and then should stay on. What do you mean by "and then the scene turns off"? On its own?
I'd really like to see logs related to this, and if you'd be willing to roll back to 2.2.4.145 and verify that they work as you expect there, but then not with 2.2.4.147. That would help.
I can't say that there is not something happening that shouldn't, but need your help to find what it is.
Yup. I understand. I will roll-back and see what happens. While attempting to take screen shots of turning on “cleaning” scene and showing it flips back to off automatically, I discovered what it actually appears to be is the performance is just really bad. I have 7 devices in the scene, but some don’t change state for several seconds. I discovered this because I turned on one scene, faster what seemed like several seconds of no response except for one light, I then turned on another scene, and the first scene was still firing as it was firing the 2nd scene. I will play more with it as well to see if the performance is the only issue.
All z-wave except on custom device for my personal hue group device. This probably also explains why one of my older z-wave dimmers (not-smart) would appear to never change state after commanding it on/off (even though the light would turn on/off). Maybe instead of rolling back, I should force a z-wave repair?
It’s not uncommon for nodes to fail during a network wide repair from time to time, as there are massive route changes that happen during this time. The best thing to do is attempt a single node repair on nodes that failed the network wide repair
That fixed it. But I did find a bug in my hue group that I fixed too, which was the cause for the scene turning back off. Between the two it appears to work now.
@armand I’m experiencing something very similar to this. I migrated from a C4 to C7. That went very smoothly. My mesh seems rock solid as network wide repairs generally complete without issue. However some dimmers when driven by scenes get “forgotten.” It’s very strange. Driving those same devices from their device pages or a dashboard works perfectly every time. Network wide repairs and specific device repairs also complete flawlessly.
I upgraded to 2.2.4.158 and upgraded the firmware on all devices I could. The issue persists and it is otherwise the only issue I am having. The upgrade from C4 to C7 is astounding. Performance of everything is greatly improved — especially all my Z-Wave devices. That’s what makes this doubly surprising.
Any help or suggestions I’d greatly appreciate it. Thanks all!
I found I had to perform a refresh device list, and then refresh scene. After that, I recaptured the scene, and it has been working correctly since. My main “goodnight” scene has been much slower than it used to be on the C-3, but I find my hub seems fine now, and performance is acceptable (though slow still) when using certain apps (rule machine and simple lighting apps)
Wow that seems to have corrected most of my scenes woes. Some scenes did indeed suffer from a level of corruption when it comes to the levels that were set. I made a spreadsheet of all the values using the old hub then just manually corrected against it.
Very strange that this aspect was not as smooth as the rest. I did have some strange timing issues and laggy behavior just as you describe. But now almost one week later everything ZWave is working quite smoothly.
Agreed. It's very weird that scenes would do this. Everything it needs to know about a scene should be in the scene Child app's state, so it doesn't make sense.
Now on 2.2.5.124 and still this is not completely resolved. I still have a few devices that do not receive scene related commands. As before the device page and rule machine events fire correctly and all devices respond correctly. But my scenes still seem to be broken.
On further testing this effects new scenes too and not only "old" scenes that came along with the backup I restored from my old hub. So now it's a real puzzle. @support_team perhaps this is something to escalate? I'm not sure I know what else to try. When our sleep scenes do not complete correctly and two lights are left on in an adjacent room due to the misfires the WAF/SAF takes a hard hit!
@support_team I did some further testing. It seems that something is up with scenes in the 2.2.5… branch. I can confirm that at least one device is "forgotten" when a scene is activated from its device page, the "activate scene" button, or otherwise from RM, a dashboard, etc. The problem is experienced with scenes that contain only ZWave or only ZigBee devices or mixed scenes that contain ZigBee, WiFi, ZWave, etc. devices. The scene composition, timing, or size does not have any effect on the results.
Looking in the log I can confirm that when this problem occurs the device that is "forgotten" never is sent any instructions. My suspicion is that scenes do some checking when activated to determine if a device already is at (or near?) the scene's target value. If it is, the device is not sent any commands. This becomes an issue during any chain of actions. If the target device has not yet updated the hub with its changed values or is in the process of changing then this checking process will only see the old value and not send commands to the device. One thing I have noticed is that the problem occurs most frequently when a chain of events are programmed.
For example, consider the very common "going to sleep" rule that shuts off a series of lights, adjusts a few color temperatures, then does that again a few minutes later… In such a rule the same light may appear quite a few times in a row. My going to sleep rule broke when I updated to 2.2.5. That is when I first noticed the issue. I thought I had ZWave mesh problems when migrating to the C-7 but that is not the case. ZigBee and even WiFi connected lights experience exactly the same broken behavior within the going to sleep rule.
I have not had time to dig into the issue, but yes, there are some weird things happening with scenes. When I go to bed, I active a good night scene. And there are no rule or actions that should be turning on lights, but every morning, when my wife leaves for work the lights in the living room are on. As if the scene was “turned off” and the states were inverted.
I will dig into this over the weekend. I wonder if it is related.
@armand@support_team I performed several more tests. I see in the logs that at times scene members actually are being sent the incorrect values in addition to simply being skipped/forgotten. For example, one of my bedside lamps is being sent 55% when in the scene that is run it should be sent 0% and switched off. The scene settings pages confirm that the correct values are stored within the scene. But nevertheless when the scene is activated either no value is sent to some devices (as in they are mysteriously "forgotten") or the incorrect value is sent.
The manner in which scenes are activated do not matter: it could be from a Rule, button, called on its device page, or by clicking "Activate Scene," etc. After a fresh reboot or after the hub has been running for a long time the incorrect behavior still occurs. My mesh is quite robust. Direct interactions with devices are instant and never miss. The issues I am experiencing are purely related to scenes. I am experiencing the exact same issue with new scenes, old scenes, scenes with devices on different networks, and scenes with devices that only are on the same network. This is a real mystery.
I have reached the limit of what I can do as analysis. There do not seem to be any remaining logs or data that I can collect. This is really frustrating at this point and I am not sure what else to do. At the moment any automation that uses a scene is broken.
