I run 5 home assistants connected to 5 zwave sticks that publishes/listens to mqtt and one “main” one that uses mqtt to discover devices. Works pretty well.
Hard for me to understand your setup.
You are running 5 HA at different platforms? all connected to its own zwave?
One main HA central listening to all?
Yes. I have a “main” HA using only mqtt. Then for each integration i have a different HA that works as a zwave-to-mqtt / zigbee-to-mqtt / samsung-smart-tv-to-mqtt / etc bridge. This way i can update only one part of my home automation system without breaking everything. For example i have multiple zwave sticks and each runs attached to one home assistant instance. I can upgrade one at a time. Or i can just upgrade the main ha for newer interface stuff. Some “glue” is done via node red by re-routing some data to mqtt topics or routing from one topic to another. I now plan to add another “main” home assistant only for automations. This way i can update ui without breaking changes affecting anything else.
I had 2 seconds of full downtime last year, and a couple of hours in total of partial downtime where only a couple of devices did not work while upgrading one of the “bridges”.
All my sticks are connected over the network with socat/ser2net and the home assistant instaces run in docker (auto restart, balance on 2 swarm nodes, etc). I also run two instances of the “main UI” in case i need to work on one and add new stuff the old one works with the old config in the meantime.
This way home automation always works for my wife and kids and I can tinker with a small part of it without bringing the whole system down.
I call it HAHA - high availability home automation
1 Like