I would like to run a separate HA instance on a separate Raspberry Pi for testing purposes. They would be on the same network.
I read already that this is not a problem generally as long as I have separate host names (Settings>System>Network).
My question is if I take a copy of my productive system and put it on my test system will this create issues? For example if my devices are connected to both instances at the same time.
I would disable all automations on the test instance so they don’t collide with my productive automations. Is there anything else that can cause havoc?
Or would it be better to put a fresh empty system on my test instance?
As long as you don’t have devices tied to specific hardware on your first HA instance (e.g. Zigbee dongle with ZHA, USB-attached Rflink, …) it should work.
As an example when developing a new package (sensor templates, automations), I do this on my test system. Most of the time, I’ll set up simulations for the inputs using input booleans / numbers and sensor / binary templates and same for the outputs. This way I can test the logic without actually affecting any of the real devices. So in this use case, I’d move the package I’m developing over.
Are you running them in one network? I understand - two different names / two different ip’s? How to set the second instance up - from scratch / from backup with change in IP? Please let me know, I want to migrate my ha green to ha yellow and use yellow as master + green as test. Thx!!
I set them up from scratch since they were different systems for different houses. Initially I managed the configuration independently, over time I realized I was mostly copying and replicating the same configuration over and over again and it was error prone. Now I generate 90% of the config for each system from a build script. Anyways…
In your situation backing up and restoring is likely the fast approach; whether you can run both systems at the same time depends on the integrations you use. For example zwave and zigbee use dongles and these dongles can only be plugged into one server at a time.
Another approach would be to use the test system as a scratch pad and start with an empty config. Then when developing a new blueprint, automation, etc. You develop and test it there and create templates / input_numbers, etc. to simulate the device inputs and outputs. If you start using Home Assistant packages; it’s very easy to copy packages from one server to the other.
