Splitting OH2 on gateway (HW bindings) and main application engine

Hello,

let me provide some context: I have a small datacenter on the basement, where I run a couple of servers for NAS, backup, content delivery, and some other applications. I’d like to run OH2 there instead of on a RPi upstairs. There is no proper, e.g., Z-Wave signal reception there so I was wondering if I could use a RPi as a “gateway” just forwarding everything to the actual OH2 installation on the datacenter.

Latencies are pure ethernet, so I’m not worried about them in terms of events or alarms triggered by ZWave devices.

Is this a completely crazy and or stupid idea? Makes any sense? Would it imply a lot of coding? Please note that I don’t think OH cloud fits here… and while a ZWave range extender looks like a possibility, I was wondering if a setup like this might make sense on other scenarios, like running full OH2 instances on public cloud providers.

Serial port over IP (socat/ser2net) might be another possibility, exposing the RPi /dev/* to the main OH instance. Just thinking aloud.

thanks,
Ignacio.

Yes.

Search the forum for “ser2net” or “socat” and you will see examples of how others have made it work. I personally never got this to work but didn’t try too hard either as I found a different way to solve my problem (passing the USB dongle to an ESXi VM which, for the curious, required downgrading the USB drivers in 6.5 to the previous version, search the forum of ESXi for details).

Another alternative is to host a slave instance of OH on the remote Pi and use the MQTT Event Bus configuration to sync it with your master OH instance in your data center. I would probably go with this route.

Finally, you might be able to get proper Z-Wave signal reception in your data center with one or two more strategically placed repeaters or mains powered devices. It’s a mesh network so one or two well-placed devices can do wonders for the network. You will want to have at least two mains powered devices with good connectivity to your controller. I learned the hard way that one zwave outlet quickly got overwhelmed when it was inexplicably the only device that could see the controller.

Thanks Rich,

I’ll definitely give it a try. I also have some sensors (not ZWave, but custom RF or wifi-enabled arduinos) that I’d like to feed into the system via MQTT, so I’ll play around with the binding.