Awhile back, I brought up the question about a Z-Wave best practice about whether I should remove or leave in Z-Wave nodes that were only used for the holidays…some said to leave in others suggested removing…
That conversation can be found here:
I ended up leaving the nodes in the system as other households and professional responsibilities came to bear.
I’ve noticed during the last 6 months or so that my Z-Wave network is having issues…lights don’t come on as they should (rules do fire) and there are delays when devices are manually switched on.
What I find interesting, is that the removed Holiday Nodes still show as online, even through they are sitting in boxes on a shelf in my garage.
Do I understand it correct that you switched off the devices, shelfing them but you left the thing definitions in your system?
If the answer is yes, I understand the issues:
Z-Wave is a mesh network. What you see in HABmin map is the last configuration the controller and the remaining devices know in terms of which node “sees” surrounding nodes. That is used to forward messages from nodes the controller does not see directly and vica versa.
Now that you’ve taken some nodes out of the mesh, the information about the whole net is no longer accurate and some messages might not be delivered via the best connection, thus too late and maybe never be delivered if timeouts do happen.
A network heal can take care of the network information and will update the neighbors information the controller and the remaining devices have, improving your network stability. There is a setting in the controller where you can decide when it will happen or you can switch it off entirely. You may want to check this setting. A network heal is needed when the network structure is changed by adding, removing or relocating devices.
A network heal produces a lot of traffic, it is best done when there is not much traffic expected otherwise.
About your reason to physically remove the devices: I guess, basically you need to decide between the hassle with the network consequences the (temporary) removal will cause versus the energy that the devices will consume when left in place.
“Online” means they were fully configured. They stay fully configured even when they do not respond when switched off. The term “online” is misleading here, it does not imply the controller can still communicate with the devices.
Good to know. And I learn new facts
My assumption stems from my own experience for a long time now, having seen the same situation here in my installation with old heating valves that were not properly excluded due to failed hardware.
I thought heals take place daily around 2am? I thought that was a default setting, correct? I have not disabled it. The nodes have been on the shelf for almost 10 months…one would think the heal should have corrected things by now.
Thanks for clearing that up @chris, I could not imagine that the ZWAVE things would act differently from all others. If you say this is a bug, is this something I should post as a GitHub issue?
I would like to say that it doesn’t matter, but I could be wrong! I know people have had problems with zombie nodes causing routing, but I’m not sure if the problem is “real” or not. Certainly, if you unplug a device, it will cause problems in the network immediately. Now, depending on what type of devices you have (ie how old - are they ZWave+) it (in theory!!!) ought to recover itself as there are automatic systems to search out the routes if commands fail.
If a number of devices in the network are a few years old though, this may not work well, and the network would require a heal to restore the routes - assuming all the heal commands go through ok it should restore the mesh to its former glory (and again assuming that there are still enough powered devices to form a good mesh).
However, that’s what I think should happen - what really happens is something that I can’t be sure about as it’s mostly not within the control of the binding.
There is one open already - I’ve just not had the chance to look at it unfortunately.
Are you on the latest version of the binding? There was an older version where the heal wasn’t working correctly.
To add my own experience to the previous question (delete or don’t delete), I’ve always had bad experiences if I don’t remove old nodes. This is on a fairly robust network with 30+ zwave devices ranging from very old (6 years +) to pretty new (several months old). At least for OH zwave, I’ve noticed severe degradation by leaving in zombie nodes.