Getting ready to start buying for my first real smart home, and am considering Home Assistant as my main platform, but I do have some concerns about upkeep.
I’ve messed around with HA a bit, and am not scared by the complexity. But from what I read in the forums, I see a commonly repeated phrase about things occasionally breaking on update. This is a border-line deal-breaker for me, because I do not want this to require any kind of regular software upkeep. So, assuming I get my system set up exactly how I want it, and that I’m not looking for any additional features…
Are there any reasons I can’t just stop updating altogether? (I get that not regularly updating can cause problems when you eventually DO update… but why would I ever NEED to update at all??)
Bug fixes to existing integrations you use - for example when a cloud service you use changes their API
Security vulnerability
To be honest, the complaints you see here about things breaking on update are just a tiny proportion of the userbase of HA. Remember happy people don’t post to say it’s still working.
I haven’t had an unforseen breaking change in literally years, at least two, probably a lot longer. Releases include a list of known breaking changes, and as long as you’re not blindly updating (as many of those who post about problems do) then you can prepare for those. Even with those known breaking changes few of them have caused any actual downtime in all the years I’ve been using HA (over 6 now).
Of course, if you want “set and forget” then no open source platform is going to be for you. You’ll want to use a commercial hub that handles all that for you - with all that means.
Based off your and Templeton’s answers, it seems like the stability for set-and-forget seems like it would be possible with updates turned off. Why then, would open-source platforms not be a good fit?
Even a no open source platform can give you a headache.
After more then 10 years of using Fibaro (no open source) i was being fed up.
Every update fixed some things and broke other things.
That’s when i decided to go for HA.
HA is very different in a positive way (at least in my opinion)
Altough i’m struggling as a newbie i’m happy i made that choice.
There are a lot of very smart people in the community who can help you out if you run into a problem.
No home automation platform can be static, unless you’re going to avoid any cloud service and only use Zigbee or Z-Wave devices. Even then you probably only want to use Z-Wave since new Zigbee devices may well need a software update before they’re supported.
Cloud APIs change. You’ll need to update when this happens if you want to keep using that cloud service.
Local APIs can change too. You’ll need to update when this happens if you want to keep using that local service.
Bugs happen. You’ll need to update to fix them.
If you don’t use any cloud services (no weather reports etc), don’t use any local WiFi devices, only use the same Zigbee or Z-Wave devices (no replacing them with different models when they fail), and never expose it to the Internet then you can probably get by with a static HA install.
I update HA at least once a month, sometimes more, but each update typically is a two to three minute affair, including the time taken to skim the release notes, issue the update command, and wait for HA to restart. Sometimes it takes longer as the release notes have something I need to address, in which case it’s maybe a 5 minute job (but often is nothing more than comment out that bit of YAML).
I’m drawn to the customizability and cost effectiveness of HA, and don’t mind putting in work on the front end – my goal is to create an excellent stable smart home that operates intuitively for outsiders. I realize that many in this community see HA as somewhat of a hobby, and that’s not me. (Nothing wrong with it as a hobby! Just not my jam) I see it as a means to an end – I’m still trying to gauge how naïve that perspective really is, though.
From systems I’ve looked at, I think I’ll likely end up with HA or Hubitat since they both offer the kind of automation complexity I expect I’ll need.
The initial “load” on your time will be high, on any platform, while you get it set up and learn how to use it. Once you approach “steady state” the time investment will drop off. It’s never going to be zero, but it’s reasonable to assume that if you’re just keeping on top of updates it’s under 5 minutes a month.
My own setup is stable, the only time the family notices is on the odd occasion something goes wrong (like last night, when the power supply to one of the LED strips failed), or if I’ve been tinkering and not got things quite right.
Sure, pottering about with it is a hobby, but only to the extent that it’s keeping me learning. I’ve often gone months at a time without making any changes beyond updates, followed by a flurry of changes as I’ve added some new sensors/other devices.
Perhaps something to consider is what do you envision happening when (inevitably) something does go wrong, such as a light not turning on when expected.
If you are ok with looking through system logs and forum posts to try to figure out what happened and then tinkering with the system to fix whatever is needed, then you will love Home Assistant.
But if “not your jam” means that process is something you want to avoid, then you may be frustrated with Home Assistant and perhaps a commercial system would be more appropriate for you.
Maybe another something to consider is that you always have a backup for the critical needs.
Indoor lightning, heating, doorlocks…etc.
Don’t become totally independend from your system.
There isn’t a system that never fails !!
I have arrived at a quasi-set-and-forget solution by using mqtt devices only and use HA solely as a Frontend. Specifically, I
ran every single device on mqtt : write gateways if it doesn’t support it out-of-thebox
ran my automation logic from a separate system that listens/posts to the mqtt bus
had HA gone completely offline - I use cloudflare zeroTrust to sign in to the HA instance
maybe once in 3 years, the zigbee2mqtt/HA integration breaks : z2m still posts to the homeassistant/# topics, but HA would not read it. That time I had to reset the z2m integration on HA.
I had to write tons of mqtt gateways, but by moving HA out of the centre of my solution, I have achieved higher flexibility, stability over time, and graceful degradation.