I'm really starting to worry, getting scared of my future with HA

About 2 years a go i’ve started small with Home Assistant after a while fiddling around with Domoticz.
I was very happy with it. I did have three MiLight lamps and some 433 MHz switches. So i started building a system, which finally looks much better, than Domoticz and even my wife finally accepted, that time will change and we are going to meet the future.

Home Assistant became a nice hobby. I bought some more things, like Z-wave switches, controllers, more MiLight stuff and so on.

My system did grow to fully sophisticated smart home.

Thermostat, energy readings, automated self-made Jacuzzi, Denon AVR, Pioneer AVR, some WiFi media players, Z-wave motion sensors, OMV NAS readout, Torrent servers readout, Raspberry readout, Broadlink infrared remote and i still forget something, for sure.

I’ve got so excited, that i even got NabuCasa to sponsor, I’m buying coffees to devs on GitHub, Sponsor them on PayPayl.

BUT!

Last three months HA is getting me mad!

I do understand, there has to be movement to next levels, i do understand security patches and so on.
But there are so many changes last time, I can’t keep up.

Z-wave had to be migrated, costed me days to understand, what i have to do.
Denon integration failed and needed to be updated by the devs.
Broadlink remote i still have not working, as i do not have the time to find out, how it works, because is everything, but not easy.
Garmin integration for my smartwatch stopped working, needs update from devs.
And today i found out, the Pioneer AVR stopped also. The integration in core is more than 2 years old and no one updates it.

It is too much, guys, too much at once. I wanted a hobby, where i sometimes have to find out, how something works, get it done and move on to the next goal.

Now I’m spending countless hours fixing broken stuff, searching answers, communicating with devs and so on. While friends are baking some beef outside, I’m starring at the screen wondering, if i will find a solution for the next problem.

And here comes the fear. Like, my MiLight system is supported by HA thru the LimitlessLed integration. That one also is not updated for years and i really start smelling shit, one day it stops working too and i will sit here again, starring at the screen again, in the dark, because the lamps won’t listen to me anymore.

Is this the story, I’ve got involved with? Is this the all mighty Home Assistant, that can do almost anything?
I feel like, i’ve got a second job, and I’m not getting paid for it.

And yes, i understand, that all the devs are volunteers and it is what it is and so on. But please, really, please, can we somehow build in some kind of warranty, that things once integrated will STAY working?

So, sorry for this rant, I’m just really frustrated by the next broken integration and need to ventilate it somewhere. I hope, someone will understand my current state a think about this a little.

15 Likes

First, I do understand your frustration and the burden that resolving these issues can become.

That’s unlikely to happen anytime soon. Like you pointed out, it’s mostly built and supported by volunteers (notwithstanding that Nabu Casa employees draw a salary), especially the integrations. Just because an integration becomes “official” doesn’t mean it will be supported forever (for example, the original author may disappear and no one volunteers to take their place).

I don’t foresee this arrangement changing so if it no longer suits your needs you will probably need to think about choosing an alternative solution. Home Assistant, as a volunteer software project, is not likely to quickly evolve into a product with service/support guarantees.

I don’t need official support, and i do understand, when SOMETIMES something slips away.
But like I said, it is becoming just too much at once.

I really love HA and love fiddling around with some code, just not enough to do that daily… And if i want to keep the support of the family, it will help, when things do not stop working randomly… That’s all…

So, something like: When an integration looks abandoned, it will be fine, when the devs get notice on time and they will get stimulated to update it, before it stops suddenly. :wink:

Problem is if a dev isn’t using an integration and none of the beta testers use it either well then no one will know there is a problem until you the user do. For some things maybe you didn’t read the release notes… maybe you should just roll back till you can reconfigure for the changes.

5 Likes

Originally you asked for an integration warranty. To provide a warranty, someone must be “officially” responsible for maintaining the integration. If you don’t want official support then a warranty is moot.

If an integration “looks abandoned” it’s likely to be abandoned unless someone volunteers to maintain it. No volunteer, no maintenance and the integration may eventually become unusable. No one is “stimulated” to support it unless they have a vested interest in it (or is paid to do it).

I left HA about 2 years ago after frustration on the lack of device support in OZW 1.4. At that time both OA and openHAB projects were in transitions but, with the assistance of the Z-Wave developer at OH, I moved to openHAB.

I am now in the process of moving back here to HA because, in my personal opinion, the controlling developers at OH pay no attention to user needs. The developers expect the users to attempt ato document the project. Here, the developers, who are the experts in the product, write the official documentation.

Currently, in openHAB if you have recent Z-Wave devices, you cannot run their stable release, but must run one of their alpha or beta releases. One advantage of zwavejs is the releases are not tied to the project release schedule.

You will find no project involving humans is perfect. I really understand having a tight schedule. My smart home hobby has been ignored for a while lately too.

1 Like

You deserve a :heart: but I didn’t want it to be misconstrued as totally agreeing with you.
I too feel like this has been a second job but I agree with others, HA is built on the work of others, there is no “product” per se, other than Nabu Casa, which does what it advertises, not support custom components.

I recently started deleting things I added a while ago that I don’t need anymore, maybe cut some of the excess out of your system to remember the simple pleasures of things turning on when you want them to. Media control seems to be a little “extra credit” still.

4 Likes

Important discussion, thx. But hoping to add a chuckle…

8 Likes

Just some quick thoughts, you don’t have to upgrade always to the latest and gratest. Once you got a stable system you should stamp that as your main.

And maybe setup a second HA instance as development/test. Update there, check if things still work. If not, post about it and schedule some time for yourself to tinker. That way you got less headaches about automations not working, sitting in the dark with no lightbulbs able to turn on, you create time to sit with your friends and eat some of that delicious beef and once you got spare time, or bored, HA time :slight_smile:

You’ve reached a certain point where you should evaluate the stuff you got and if it’s going to last. Mainstream things, or stuff from large corporations (not always) tend to have a lot of attention here, thus always someone developing uppon it. Got some cheap ass weird sensor from a dusty corner from AliExpress warehouse and happen to come with a weird integration? Meh… maybe best to let it dust some more unless willing to tinker with it a lot.

That’s unfortunately the reality with HA. It’s grown so fast in such a small amount of time, it is just impossible to keep everything stable as to where other platforms have a much slower release cycle, giving everyone more time to upgrade/update other codes. Developers of integrations sometimes scratch behind their ears like damn, another release and something is deprecated so gotta fix that too now… lol

9 Likes

I think this issue has at least two sides. One HA project could do better and one that is impossible to solve

There are too many breaking changes that affects the integrations. The API between HA and integrations has got to stop deprecation hell and define and document a committed API. On the Foswiki project we had many discussions on this but we were a majority in the core team that agreed that plugin API was holy. This ensures that integrations will keep on working even when not actively maintained - with respect to the interface to HA.

The other side is that an integration has at least two interfaces. One to HA and one to the product it interfaces. And this is the part that HA does not control. Corporations have short horizons. They sell products and promise more than they intend to keep. Do you love your Logitech Harmony? Remember when they closed their API? Yes they opened it again. But now they announced that the product goes end of life. It is only a matter of a few years and you have a door stop.

What can we users do about it?

Well the first is obvious. Shout loud when the devs making breaking changes with no backwards compatibility. It helps when there are many voices shouting and it is not the same that has to try and represent the users.

The second is probably more important.

Choose your products wisely

Go with products with standard interfaces like Zigbee and Zwave

Prioritize local control. Products that can only be integrated via the manufacturers cloud will break. You can expect 5 years from product introduction and then the product is likely to stop working or no longer as useful API. Corporate greed

Learn a little DIY. A device that runs ESPHome is not likely to break. Or like me, I write my own ESP software and communicate via MQTT. MQTT is stable. ESPHome is community owned and not depending on cloud. Even if abandonned, it will work as long as you have wifi.

If you want a motorized curtain, do not buy one that depends on cloud. Buy a dumb system with wires for open and close and control it with a locally controlled device.

Keep things simple, and local. Stay away from the all too smart products from the stores from companies that come and disappear.

I use an integration that integrates my Miele washing machine to HA so HA can announce “washing machine cycle completed”. I give that 2 years, then it will be broken. I am prepared for it and I can live with it because it is more fun than it is essential.

Your lights, your heating, your curtains, they MUST work. Keep these local and use either DIY or open common standards like Zigbee/Zwave or relay driven.

And the good thing is that nothing is more fun and satisfying than a good DIY solution. That is a great hobby.

20 Likes

Fully agree and understand. Every time HA fails I feel let down and every time I update I know I’m in for a rough ride because of it is break the rest of the family (and I) get mad.
2021.6 broke Sonos integration for us so all scripts using TTS breaks (we use TTS to give feedback for most automations and since sonos.restore is now broken, TTS announcements fail and thus the rest of the scripts) . Hopefully it will be a quick fix, but until then our “smart” house is dumber that smartphone without Internet access.

I wish it was a way to run a parallel installation of HA that could contain a couple of test-script for my specific installation, but with Z-Wave USB-sticks that’s not possible I guess.

1 Like

I felt like you at some point and it’s disgusting, I ended up hating that I began the HA route. However nowadays I spend roughly 1-2 hours a month in maintenance, fixing broken things, which, for me, more than makes up for the benefit of having a smart home.

With all humility, a couple of tips you might already know that saved my relationship with home assistant.

If everything is working fine, do not update, period. Only update when it’s mandatory because something broke or because there’s some new integration you need. Otherwise, just don’t update, it’s not worth it.

Stop building up more things and focus in making what you already have realiable.

Integrate everything as local as possible. Avoid as many clouds as possible… If something can be controlled locally, do it, no matter the effort.

EDIT:

and for those situations where you can’t avoid a cloud, it’d be nice to have an alternative. For example, alexa integrations break every few months. I have both the alexa_media_player and alexa_remote2. I have all automations set with both integrations. When one craps out, I just disable those automations and enable the others, quick and easy. Then I can calmly wait for the integration to get fixed.

Just my 2 cents.

Good luck!

4 Likes

I really think that this a bad strategy. One day something will break, even if you are not updating anything. Most likely because the vendor of one of your devices changes their API. In that case you’ll want to be able to update quickly if that is an important device, integration our automation. When you are then updating to a HA version many versions newer that the one you are using it is very likely that other things will break too. Then you will be stuck at your screen with multiple (intertwined) issues at an unforeseen moment. A moment that you maybe better spend grilling beef and drinking beer. I believe you need to have a conscious update strategy for yourself. I do not know which is best, but what works for me is:

  • Update once per month and plan some time for it (usually an hour is more than sufficient)
  • Skip the x.1, x.2, x.3 versions and take the x.4 or higher versions
  • When doing the update:
  • Read the blog-post (especially the ‘breaking changes’ section)
  • Make a full snapshot
  • Update everything (HA, OS, Custom Integrations, add-ons)
  • Test it!
  • If everything works, you’re done!
  • If it doesn’t work, use your planned time to try to fix it.
  • If you cannot manage to fix it within the planned time, revert back to your snapshot and plan some extra time later to make the new versions fully operational

Just my 2 cents…

10 Likes

Occasionally I have some problems with new HA rollouts, but for the most part each upgrade goes well. And I’ve got a pretty sophisticated mix of Homekit, z-wave, zigbee, sonoff, google assistant, a number of integration modules and the list goes on.

I do take the motto that “if it works, don’t fix it” so I haven’t upgraded to the new z-wave module, the old one does the job for now. I will at some point tackle that endeavor when I have time.

I generally don’t upgrade to a new version of HA right away, usually wait for the bugs to get worked out in the first few releases.

I always do a full snapshot before each upgrade. And if it’s really a major change, I will do a snapshot of my VM so I can just revert the whole system back.

Since my HA is part of my home security system I do have to make sure it’s working 99.999% of the time. If an upgrade fails I revert back until I have time to troubleshoot.

@Plevuus posted good steps, pretty much my method too.

2 Likes

I do agree with @Plevuus, try updating regularly, but read ‘breaking changes’ thoroughly!

Not updating until you need it might cause the break down of a lot of other things…and than you have to back trace all breaking changes since the last time you updated and try to figure why nothing seems to work anymore…

Not doing it myself, but some people have a ‘shadow’ system, on which they try new software before deploying it, and try to fix it if it doesn’t work :thinking:

3 Likes

I concur…also I run HA on a virtual machine, of which I take a snapshot before every major update :wink:

3 Likes

I still preffer to sit down for a whole day once a year than doing many little things very very often, but that’s just personal of course. I sit one or two straight days a year and spent the rest of the year grilling beef and drinking beer.

Try both and let us know I’d say :smile: I got fed up of updating, fixing, updating, fixing…

2 Likes

Perhaps nabu casa should put a pause on hiring and buy some of that equipment they need in order to keep obscure integrations working.

Or start a donation drive. I’ d be happy to donate, say, a device a year to solve the problem of devs not having a device.

4 Likes

Problem is OP is saying that that is already happening to him when trying to keep his system updated. There’s no need to wait 6 months to update for that to happen, updating regularly brings those problems too.

I’d rather do the job once or twice a year and forget about it for a looong time…

That’s why I suggested keep it local as much as possible. If it’s not possible, have an alternative for every automation.

I was in his situation and that solved the problem for me. I do understand your point though. From a technical point of view, your suggestions are way better, of course. From a psychological point of view, perhaps it’s not.

From a psychological standpoint is much better to do a lot of work 2 days a year than spending your days fixing little things week after week after week. If your system fails once every 4 months, you don’t get this “whaaaat??!! again??!! wtf?!! :rage: :face_with_symbols_over_mouth: :rage:”. When that happens every week, even if those are minor things, it ends up building up in your mind.

Do not update I say! :smiley:

1 Like

That was one thought I had in moving from HA to OH but other factors negated that theory, at least in this instance.

The code here is tested before it is released. Although, in theory, the longer release cycle of OH with alpha, beta, and stable release4s SHOULD be more stable, in the case of OH there is a last minute rush to add things just before a beta or stable release, negating much of the expected stability.

In both cases developers of integrations (addons) miss warnings of upcoming breakage. In the case of OH developers are sometimes not told and find out after things have broken. I work(ed) quite closely with the zwave developer at OH and, many times we found out avoidable breakage after the fact.

1 Like