Hi all,
I am not sure if this is the right topic, if not feel free to move this questions.
I am planning to start my own home automation project with openHAB and try to plan a potential future setup.
As I understand the current solution, openHAB works as a single Server that aggregated ALL data from all devices and services I want to automate.
Imho this has the potential problem, if the server is out of service for some reasons that all devices and services in my home will be out of service.
Is there a way to setup a system, in a way, that I have a distributed network of servers (all specified to do one use case) and only have an admin panel that sends instructions to the other servers.
Every server should have itâs own automation logic, so that if one of the other systems fails, the majority of services still functions.
There are several options to solve your issue:
MQTT, Remote openHAB binding and at least, keep your lights and other stuff controllable by physical switches outside openHAB.
Thanks for the quick responses.
I havnât thought about implementation specific solutions yet, since I did not know, if there is already a solution.
I think remote openHAB will be the easiest solution - did not find it in the binding list.
On this node: is there a way to install without frontend or disable the frontend after installation?
And yes, of course everything will also be controllable outsite openHAB.
No. In any well-designed home automation, at least any critical system (e.g. HVAC, lighting) should be built to work standalone (i.e. when the server is down), eventually in some degraded but acceptable mode.
Thatâll give you sufficient time to replace or reanimate your central system should that one die.
With proper preparation (spare parts on site, disk mirroring, backups in place) thatâll then just take you a couple of hours at most.
Looking at your intended setup, you already will be having local control on your HomeMatic and Somfy controllers. Devices hanging off those will continue to work as long as those local controllers keep working even when your OH is down. OH is NOT a SPOF when it comes to critical functionality.
A truely distributed setup on the other hand (involving multiple OH servers which is how most here interpreted your question) is a different league.
That will create a completely different but very hard to handle type of problem: coordinating states among multiple systems. And itâs systems that are designed to run standalone in the first place, i.e. as the (only in âtheirâ universe) central unit.
Expect a lot of friction in trying to implement any failsafe distributed logic on top of that.
Believe it or not, but the overall reliability of distributed setups is known to be worse in effect because of the greatly increased complexity.
Long story short no you donât need multiple OH servers and trying to reliably build any such âmeshâ is a waste of time. Your time is better invested in planning for disaster recovery.
How my system works is very complicated and I have different solutions and different ways to control things as a backup.
Example My aircons/HVAC
The sensibo to google took about an hour to setup and just works. I have had very little issues with it.
Now that we have a sensibo binding it is very easy to setup automations in openHAB like.
When the temp outside reaches a SETPOINT and it is THIS TIME OF THE DAY then DO THIS
Another Example is my IFAN03 for celling fan and lights
Thanks for your explanation.
The setup picture was just an example - currently I do not have anything installed.
I just bought a ZigBee USB radio stick and a thermostat to get started and the basic understanding.
I was âjust thinking bigâ, what can be achived with a few (critical) systems, in a way that each system is independent from the rest of the âmeshâ.
My planned minimal valuable product is indeed a single server solution with everything able to still work as if no smart solution would be present (at first propably lighting and heating)
I appreciate why this would concern you, and itâs a common question for newcomers. When we lack confidence in something, we tend to be very concerned about failure scenarios (particularly with technology).
If it helps to build your confidence, Iâve been running openHAB on a Raspberry Pi for just over three years, and Iâve had exactly one failure. That was only a few months in, and it was due to a power outage corrupting my SD card. Iâve since installed a UPS and havenât had any downtime that wasnât my own fault. Many other OH users would say something along these lines.
So, thatâs all I think you really need: a UPS and regular backups.
Interestingly, many people go the other way. They want OH to be the sole brain in their home automation, because they donât want third parties knowing when they turn their lights on or off. I leave my TP-Link Kasa devices connected to their cloud so that I can use the Kasa app in a pinch, but others deny them Internet access.
This makes a good example.
Choices;
A thermostat that does its own local thing with the heating system, but can also be interrogated and adjusted by a supervisor system (openHAB for us)
or
A thermostat that relies on some intermediate system to control the heating.
As you see above, the general recommended approach is towards independent sub-systems, and openHAB taking the supervisory role. Adding intelligence and convenience, but not essential.
I will think about it a bit more the comming days, and how I can achive a setup where critical can and will work independent as Singe-Responsibility-Unit.
Needs a lot more figuring out what I want to do on my site.
You all have given invaluable input to this. Thanks