OK, hear me out. I have a Hubitat C8 Pro with several Z-Wave devices on it. I literally just plugged a Z-Wave Dongle into my Home Assistant, put Hubitat in add mode, added the dongle to Hubitat's Z-Wave. Instantly, every single Z-wave device became available to Home Assistant instantly, with native Z-wave drivers on Home Assistant (well, ok, several devices needed additional help but it was easy as cake to get them all up and running). No need at all for the Hubitat integration to communicate with Z-wave from Home Assistant. It's all completely native and works in complete cooperation with Hubitat. I could now remove Hubitat and control everything entirely from Home Assistant if I wanted to....
So here's my question: WHY THE HECK CANT I DO THIS WITH HUBITAT??!!! Why can't I connect my second hubitat C-5 directly to the Z-Wave mesh to use as a literally direct failover if the C-8 Pro should crap out? Like, what the literal heck guys?
This seems like an obvious selling point. If you could pair 2 Hubitats together and use Rule Machine so that if Hub A fails, Hub B takes over everything... Instant redundancy. Why the heck can't we do this when it's obviously possible?
Well, I don't know what I've done, and Hubitat isn't happy. It doesn't seem to work as seemlessly as I was hoping, after further playing....
Time to try and recover my mess. Still, I wish Hubitat had some sort of high-availability features for those of us with multiple hubs. My limited playing around makes it seem like this should be doable somehow.
Upon further investigation, it seems like Hubitat doesn't support being demoted to a secondary controller. Additionally, Z-wave documentation seems to say that only the primary controller (Hubitat) can manage the mesh or add/remove devices, but I seem to have successfully removed some devices from the mesh with Home Assistant, but that also made Hubitat quite upset. I think I am done playing for now. I should have probably just stuck to turning lights on and off for tonight but I got intrigued and kept playing around...
You must first recognize that the ZWave protocol has NO specification for redundancy. None, Zero, nada. It has this idea of a primary and a secondary controller but their definition might shock you, as it did me, when I was doing this in 2018'ish.
The concept of Primary and more importantly the use of Secondary was related to portable / handheld remote controls. It could be used to add, remove and control ZWave devices. To Add, a secondary controller sends a message to the primary asking for the next node ID and then uses an Include with the new device. A success is handed back to the primary and the node gets added to the master db on the primary. It's a similar process in reverse for an Exclude. Controlling a device is very straightforward, you'd peck in the Node ID into the handheld and then push the On or Off or... whatever button to do the job. At the time, there was no such thing as a "hub".. not even an imaginary one. By the time I started messing with secondary controllers, the Shift Controller command was on every controller. Even Hubitat in the C-3 days, had a Shift Controller exposed.
The road you're trudging along has long history of disappointed people, hoping they discovered the mythical failover.
Good luck, but don't say you weren't warned. Like the thousands of dreamers before you, I hope you find the answer.
If you have 2 C8s or C8 Pros you can use the cloud backup to create duplicate hubs but you would need to turn the radios off on the failover hub until needed.
Would you need to shutdown and take the failover hub off your network until needed (even after turning off the radios) or can you just leave it powered up and on your network and turn the radios on if the unfortunate event happened?
I am not sure the risk exposure warrants the complexity. Supporting a 1+1 telecom like redundancy scheme is non-trivial. Mirroring state and being able to go active->standby and vice versa is fraught with a lot of detail and error legs to handle all the degenerative cases.
In reality the only failure I have seen hubitat have is some sort of system corruption (and it is very rare) such that Hubitat locks up. It is easily recovered with a WiFi smart plug I have on the Hubitat power supply so I can power cycle hubitat independently and remotely. In two years I have power cycled Hubitat once. So it doesn't appear to be a chronic failure mode.
So retooling ZWave and adding a lot of failover code to hubitat doesn't seem worth it from my perspective. My 2 cents anyway.
With the radios off on the failover hub you could leave both hubs powered up. It would need some rework to be viable again but when the C7 was king I had looked into this capability.
Yeah, it seems that the secondary controller is not being informed about device state changes unless the changes were caused by the secondary controller, which unfortunately makes this not really as viable as I was hoping. While I get native "control" from the secondary controller, if it isn't aware of the current device states when the primary controller is doing the controlling, the use cases become much more limited.
Yeah, as noted by @csteele & @thebearmay there is no concept of failover or redundancy in Zwave.. - Your going to have to play games at the radio level, with two primaries, and just allow one set of radios to be active, and have them impersonate the same 32-bit Home ID (unique to the network) and an 8-bit Node ID (unique to the device within that network)
@thebearmay's approach does this with the restored backup, that includes the radio database and settings... - So this isn't easy, given the current limited radio access that HE gives us around the SiLabs Z-Wave radio silicion.
That all said, this HA/Failover approach, DOES become feasible with Matter, and Multi-Admin - Then both hubs can be admins on the same network, and with some effort, there could be failover (active/passive failover logic) added into the hub. Again, not sure it's worth the effort, and definitely not on the HE near-term term roadmap - But given the limitations around the various Z-protocols, it seems like Matter would be a much easier starting point, given the protocol is designed to support multiple-admins on the network concurrently - You just need some additional handshaking (which could be done with HubMesh) to track heartbeats, who is the active admin, etc.
JMHO on an easier way to approach this (not sure if replace all Z-devices with Matter is something you'll want), but long term, at least your not fighting against the low level details of the protocol itself.
Like the thousands of dreamers before you, I hope you find the answer.
