Hacker News Reader

Comment by mongol | Hacker News Reader

I gave up on Home Assistant, it required too much tweaking of configs between releases. I took inspiration from https://github.com/stapelberg/regelwerk instead

View on HN

I don't know where all the frustration with HA is coming from. I've only started with it last year and so far the experience is amazing. Not only that everything just works with it, it's also so simple to create fun automations, and recently I've even built a little watering system with a Pi Pico, and integrating it with HA to show graphs and controls was a breeze.

Also, literally can't recall one time I had to update a config when upgrading HA. It's the one piece of software I'm never worried when pressing "update".

I've only recently (less than a month ago) started to use HA and I have polar opposite impression from you.

I gave up on updating because the 2 times i updated the HA Core package and the OS most of my integrations broke. The first time i bothered changing the config. The second time I decided to rollback and keep everything as is.

Maybe the issue is that you used the "Core" installation method which isn't a fully managed installation and only really there for "advanced" use cases. It has this warning above the installation steps:

> This is an advanced installation process, and some steps might differ on your system. Considering the nature of this installation type, we assume you can handle subtle differences between this document and the system configuration you are using. When in doubt, please consider one of the other installation methods, as they might be a better fit instead.

I use the container installation method and it's usually as simple as updating the version number in Docker Compose and restarting. I'd imagine the full OS image is even easier and more stable.

I use what they recommend: I installed Home Assistant Operating System on a Raspberry PI 3. Which still means there's an OS and there's HA Core.

> Home Assistant Operating System: Minimal Operating System optimized to power Home Assistant. It comes with Supervisor to manage Home Assistant Core and Add-ons. Recommended installation method.

I tried on purpose to do everything as 'basic' as possible because I don't want to spend time making sure it keeps working.

no idea why it's recommended to install manually; I'd recommend docker every time. currently running on docker and upgrading was indeed just pulling new images (didn't even have to tweak the compose file, just docker compose pull). There's been a few helpful UI messages about outdated integrations in the config (e.g. workday detection moved to UI config, used to be yaml).

Iirc Home Assistant Operating System does use Docker. It's a bare bones Linux distro built with buildroot that basically just runs Docker containers

I don't think the problem is how to update: it's easy enough in the interface, it just works. The problem is whether after the upgrade you suddenly don't see the heater anymore, or another device stops working.

Ah ok, I misunderstood then. I presumed if you used the OS installation method then the upgrade was done from within the web UI, but I maybe have misunderstood that too.

yep, through the UI. the update went fine, the system afterwards didn't work properly though.. :(

> I'd imagine the full OS image is even easier and more stable.

It notifies you when there's an upgrade, you just click "Install" in the GUI.

That said, every time I update it the disk image grows by like a full gig. Haven't figured out where that's going so I just don't update it anymore. It's running on a minimal headless computer and doesn't have enough space to keep doing that.

Yeah like others suggested, maybe it has to do with the installation method. I'm using a Home Assistant Blue. It all came configured, I'm clicking "update" a couple of times a month and never had the slightest issue.

I think it really depends on when you entered the ecosystem and the maturity of the integrations you're using. HA has some contribution standards but you can get plenty of "I made this in my freentime and upstreamed it" integrations that end up going through major renovations. Sometimes vendors take notice and also weigh in.

Additionally, if you got in ~5 years ago when it was pure yaml, it seems a couple times a year there's breaking changes where more yaml is ripped out and manual reconfig is necessary (it'd be nice if it just migrated the config for you)

> I don't know where all the frustration with HA is coming from

Because it's breaking all the time. They have a big surface, and integrations are changing fast, this is natural. But at the same time they are also eager to change and break things on their own software, and this all accumulates to a messy experience for many people.

It’s not natural. I use both hA and Openhab with about 20 integrations/addons and HA breaks more (hello zwave}. Openhab almost never breaks on an addon level. If it breaks at all it’s in the core and patched quickly.

How HA has handled ZWave is my only complaint with HA.

If you only had it for a year, you may not have seen it yet. I had lots of configs setup, then I did not have it running for a year or so. Trying to make it run again on the latest version was frustrating. I want to set up and forget to the largest extent possible.

I've been an HA used for quite a while. I used to maintain my own venv and religiously update my yaml with each release to keep up with the deprecations and changes. A year or two ago I finally decided to learn about Docker by moving my HA install into a docker-compose setup (and migrate my zwave config to zwavejs), and since it felt easier to rollback in cases of breakage, I have subsequently been updating whenever I feel like it and "just see." Surprisingly, while I get a lot of warnings in the logs, I find that my fairly complex setup (zwave, zigbee, mqtt, tasmota, ESPHome, custom 433 RF over GPIO, custom NRF24L01 over GPIO, BLE, webhooks, iOS, and a custom Alexa integration via fauxmo) continues to work flawlessly.

I realize that I'll eventually have to deal with the tech debt that I'm accumulating, but it's been really surprising how good hass has gotten at auto-migrating configs and dealing with neglectful users such as myself.

Once I switched to a docker-based setup, this has been the polar opposite of my experience. It really took away the pain of a manual setup, dealing with python etc.

My update process has been the following for about a year, with zero issues:

- Create a backup - Update all HACS add-ons - Update the docker image and restart

I've never had to restore a backup. I do wait about a week before updating to see if there are some really breaking issues, but so far I've experienced none.

I haven't really touched my YAML in a long while, especially not for anything release-related.

Case in point: I just did the update that TFA mentions, and is was as simple as running `docker-compose pull && docker-compose up -d`. This is on an install that has 20+ different integrations, 30-40 "devices" (some of them are virtual) and a lot of entities.

This has been the case for the last 2-3 years at least.

> it required too much tweaking of configs between releases

I’ve been using it for a long time, including one switch from rPI to a Proxmox VM. I had to manually change the config once or twice, and that was several years ago before they decided to clean up their userfriendlyness.

Reply on news.ycombinator.com