? days without an accident

Social
21

I was at sea for around three weeks recently and just before reaching a distance where the cell signal from the coast would be just too weak for a data link, my HA sent me a notification about an anomaly on the home power usage. I spent the next two days trying to catch a stray 4G signal to ssh in before I gave up. It was terrible. Of course it was something totally harmless I discovered when I got home last week…

I met a guy with a Starlink on his boat at a marina. We talked about how well it worked for him. I need this.

So I go out sailing to relieve stress from an overconnected world, and then I stress out because I’m not connected :frowning:

2 Likes
22

Being that I’m an IT nerd by trade, the last dozen-ish years in consulting, all I do all day long is deal with messes made by people JUST smart enough to make them, but not nearly smart enough to clean them up.

Having said that, stability is probably one of my number one concerns with anything in my house.

Largely, my implementation of Home Assistant is stable as can be. I have left it running without “babysitting or tinkering” for perhaps a month at most, where I did nothing but just use the dashboard to work lights, covers, and various entertainment systems. But that’s only because of the way this question was phrased, so I provide my own color below.

Now, being FORCED to tinker with it due to an issue? The only one I can recollect is the zigbee fiasco with 2023.5. I think there was one other update somewhere that gave me a pretty severe headache, but at least that one was a bit easier solved.

The other thing that gave me a massive headache was the mail and packages integration. Similar to the zigbee issue, the responses were largely “works fine for me, sorry about your luck.”

In my opinion, the system will be ready for “production” not when we “power users” can stop tinkering with it, but rather when the support finally rises to a level that issues get resolved - whether they affect 1 user or 1000 users. Until such time, there will always be people willing to sell their soul to Google or Amazon or Samsung because of stability and supportability - interoperability be damned. Due to this, Home Assistant will just be another “nerd toy” that every wife will rip out of the house before our bodies are even cold.

1 Like
23

Well said. I don’t see it happening. HA is not owned by a monolithic organization which can enforce reliability and support standards. People donate their time to develop what they want, how they want it, and to their own personal standards. I know there is a process to get things rolled in, but I don’t think it’s as rigorous as you’d use for a commercial product. Especially, nobody likes writing documentation, and frankly most developers aren’t good at it anyway. You need someone with a user perspective for that.

24

I wouldn’t be so sure about that. Have you noticed, with what pace NabuCasa is expanding? I remember the time, where only Paulus and Frenck were “the developers”, now NC has 40 employees…

The standards are coming to HA as well, have you seen, how many manufacturers are getting into the “Works with Home Assistant” world? KNX was another big step for that.

And counting only by the numbers, HA is the most important home automation system in the world. Way more than 500.000 installs running, in almost every country of the world. That’s imho right where propietary software like HUE is as well.

I personally believe in open source projects, and if they are managed well, there is a good chance for a lot of money to make. I wouldn’t go that far, that NC is already a well managed company, but that isn’t a must for now. In my eyes, HA is doing well, and is getting better every day. As I said, 40 employees want to get paid, and if they just sit around, they won’t be staying long.

1 Like
25

Oh, I’m aware of the reality of that, and the challenges it brings.

For something people can go and download for free, it’s fantastic. But there’s still the knowledge chasm that has to be crossed somewhere by the user, and that’s what’s difficult for some people.

As for documentation - my CS101 professor said:

“pretend you are writing instructions for your grandmother to use your app”.

Always seemed to work well.

26

From the Signify 2022 financial results:

Signify’s installed base of connected light points increased from 96 million at YE 21 to 114 million at YE 22

Granted, this includes all of their smart lighting products (not only Hue) and it doesn’t say how many hubs are active, but it gives a good guesstimate point. It’s a lot more than 500k. Google Home / Apple Homekit / etc are probably an order of magnitude higher even.

HA is great for geeks like us. And it actually is pretty stable if you know what you’re doing. But as long as it isn’t polished as the aforementioned products, and as long as every update breaks a ton of things and need active maintenance to get it back working, it’s never going to get out of the geek into the mass consumer market area. For 99% of mass consumers, the moment they read ‘flash this image to an SD card’, the HA adventure is over before it even started.

But maybe that’s actually not a bad thing after all.

1 Like
27

Yeah, I read these numbers, I’d say it’s a nice try to keep investors cool. Let’s do some math and see how that turns out. :wink:

96 million light points, that’s the Signify speak for everything that “lights” (the Ambilight for the TV counts not as one, but x times, depending on the number of stripes you have installed).

  • Let’s just assume, we have around 15 to 20 light points in an average home (and that’s conservative).
    • 4x livingroom
    • 2x hallway
    • 4x TV
    • 2x bedroom
    • 2x ambient lights around the house
    • 4x garden and garage
    • 1x bathroom
  • Let’s further assume, not only private homes, but small and midsize offices use HUE as well. It’s just a guess, but 33% of all the million light points could go into that category. Makes around 120 for an average office, small to mid-size, some bigger, some smaller
    • 2x per one employees desk makes 100 with 50 employees
    • 10x hallway
    • 4x toilets

That brings us to:

  • 33% out of 96.000.000 = 31.680.000
  • 67% out of 96.000.000 = 64.320.000
  • 31.680.000 / 120 = 264.000
  • 64.320.000 / 20 = 3.216.000

Look how that came down to only 3.4 million users. Now taking into account, that if they talk about light points, they are really talking about what they sold, not what really is used, I’d lower that number by 40%, to get to some kind of reliable number of real users. That would be around 2 million.

Now let’s see the HA side. The analytics website gives us 250.000 installations. Here we can assume, that at least two thirds don’t have the statistics on. It is an opt-in, so you’d have to set that checkmark explicitly. I’d even go so far to say three quarters of users don’t have the statistics on. Staying with my guess, we’re talking here about 1mil users. That is not that far from Signify, if we only take HUE as the competitor, numbers might look even better for HA.

Let me make that clear, these are my estimated numbers, I can be totally wrong here, and this is just a brain teaser, not a valid statistical approach. What I want to say is this, HA has imho reached that stepstone, where you can’t ignore it on the market. If HA doesn’t want to play nice with you as a manufacturer, you realistically can see customers move elsewhere. And that’s where I believe is the potential for HA. If handled well, and I have no reason to assume otherwise, HA can make it to the TOP3 of all home automation systems.

And for people that want a plug&play system, they can (and I think a lot already do) always buy the Yellow. Unbox it, give it some power and of it goes. So not any difference from a system like HUE.

28

Nonsense. Yellow, blue, purple (future color?) makes NO difference.

They key you’re missing is that upgrades need to stop breaking things.

None of this does anyone any good if it works out of the box and the first update destroys it.

29

A whole lot of assumptions and made up numbers doesn’t make math any more valid :slight_smile: I think you massively underestimate just how big these ecosystems are within their market segment and just how much HA is lagging behind the big names in the mass consumer market. Any consumer slightly interested in smart lighting will know what Philips Hue is, brand recognition is very important here. Home Assistant ? Huh what ? Can I buy this at Home Depot ?

As long as it isn’t sold as a boxed product at your local hardware store or has the brand recognition of a Google or Apple or Amazon, requires any form of tinkering and - getting back to the topic - provides long term stability rather than regularly self destructing after an update, HA won’t ever be able to compete in the non-geek arena.

1 Like
30

This topic is on the verge from diverging from the original intention by turning into a rant. Read the original post again if you disagree: It’s about your own habits and not how you are affected by updates. The geek and fun side of things — and perhaps things that go wrong due to your own doing.

2 Likes
31

I think it really depends on what integrations you use. Anything cloud connected, the answer is probably no. API changes will break things. If you want to add new unsupported devices, the answer is most likely no too.

But if your smart home is ‘complete’ (haha, yeah I know…) or you use an entirely standardized protocol for everything, like MQTT, then why not ? I am going to find out. My production system is on 2022.11.5 and I decided to not update it anymore at all. I only use one single outfacing integration, MQTT, nothing else. Everything just works fine so far. I do experience some diffuse long term memory leaks in the HA process. They’re small, noticeable as a slowly growing trend over months. I do auto restart the HA process from time to time due to this. The host Pi has an uptime of around 240 days right now.

If at some point I want to update in the future, I would only have to care about breaking changes in the MQTT component and possible issues with database schema updates.

I do have an up to date dev install too though, so I can try out new things.

2 Likes
32

I’m running three instances / three homes with Home Assistant. I can normally leave them unattended for months. I don’t necessarily “check-in” regularly, however I spend a lot of time building and tweaking new functions!

I’ve just set up a brand new irrigation system in Home Assistant, of course that needs a lot of initial attention as devices turn offline, sensors spike, statistics are flowing in, and automations do not yet cover the special cases.

For me key to a reliable HA setup is having a ton of warning-type automations. I normally end up with twice as many automations to detect, handle, and notify upon errors, versus automations just fulfilling immediate use cases.

:bulb: When Home Assistant checks in with you, you do not have to check in with Home Assistant.

4 Likes
33

When I first started back in 2015/2016 I went 9 months without restarting HA or Updating. Then I updated an everything broke. Turns out they completely changed the OS at the time. I then kept up with it regularly. I probably go 2 months at most without updating/restarting when I’m too busy to work on my own setup.

At the time, you could only run HA via venv. Then it switched tot he all in one installer, then resinOS, then hassio, then the current 4 install methods. Which thankfully hasn’t changed and probably won’t change anytime soon.

Anyways, I would expect HA to last for years without restarts. At least I’d expect my setup to do so. The only “gotcha” is supervisor auto-updates and if you’re in the beta channel, that can break things sometimes. However if you’re running in the stable channel, I’d expect it to hum along without any involvement.

3 Likes