uh, raspbian…
Yes but it isn’t ‘Debian’ - there are significant differences and Raspbian isn’t supported (as I am repeatedly told). For instance, apparmour is not included in the kernel.
I honestly wouldn’t have thought that it would have existed. Then again I didn’t know there was an Ubuntu distro for it either. I thought there was just “Raspbian” that would work on the Pi.
Before this gets derailed into a discussion that I don’t want to be in, please just wait until there is an announcement.
Neither did I until a couple of weeks ago!
i can’t speak for you, but i never had 1 problem running addons on supervised on linux
at least the addons i use or i used.
i just use ubuntu and raspbian as base os.
I’ve kept pretty quiet on this thread. 10 months ago was running SmartThings (for ZWave) and a few other platforms (TP-Link, Honeywell primarily) that was all integrated with the old Stringify. Worked pretty well and was EASY - I just quickly setup the integrations in Stringify and was building ‘flows’.
Stringify was a great little platform while it lasted. Dead simple, fairly powerful, tons of integrations - only thing it lacked was ability to update/edit outside the app.
When Stringify was shuttered last summer I looked for another platform and found HA. FWIW, I’m currently the quasi-IT guy / guru at our small business and know enough to have setup our 40+ PC network, SonicWall firewall, NAS Drive, etc. Also handle our website and code .JS, HTML and .PHP primarily. I know enough to be dangerous and had ZERO knowledge of Linux/Ubuntu/RPi’s etc.
Like many NOOBs, I Googled a bunch, found these forums, found some YouTube videos (JuanMTech and others) and set about my way. Ordered a RPi3+, followed along on a video tutorial and got HA .86 (or so) up and running. The setup was somewhat intimidating (as I had no Linux / RPi experience), but was able to get it working. Don’t ask me what sudo does or doesn’t do, what version of python I need or don’t need, what the difference between debian, raspian is, etc. I honestly don’t know and don’t care. I want my home automation platform to WORK and be EASY. I have a full time job and don’t need another one, this is a HOBBY, I don’t get paid to tinker w/ HA (my wife feels the same, “great make automations but don’t take an entire Saturday away from the family doing it”).
After 3-4 months in I looked at getting some cameras integrated via ffmpeg platform and whoops - the Pi came to almost a complete stop. Clearly it wasn’t powerful enough to handle.
2 weeks later I’d procured an i5-6260 NUC w/ 4GB RAM and 128SSD from eBay and used JuanMTech’s tutorial to install Supervised HA / generic Linux install (I found this out in this thread, b/c [and this is the key point] - I didn’t care what method I used - it’s a hobby - I just want easy and functional). I had zero clue I’d just switched a whole different install method. I ‘Etchered’ a drive and got on with it.
It’s been a bit of a pain to learn / figure out how to upgrade Docker, upgrade Ubuntu and which commands I needed to use, how to get in via SSH, etc.
Going forward, I need to know that HA is stable, functional and that if my install method does need to change that there are clear & easy to follow tutorials for NOOBs (and ppl like me) to follow. Most of the videos/tutorials currently are from gracious USERS/CUSTOMERS of the platform - few of these are from the devs (understandable w/ pace of development).
Key to long term, commercially viable success is to make it easy for non-power users. If power users get under the hood and hack away on ProxMox or run 37 other containers to do whatever else - great - but the path of least resistance needs to be kept for NOOBs who just want easy - but who also need something more powerful than a RPi3+ or RPi4. Unsure if HA is going to black-box type hubs eventually that are out of the box configured, auto-updating and such - or if it wants to remain a well cobbled-together Github project.
How does a brand-new, NOOB user go from zero to functional - what’s their best path forward, how long does it take, how painful is it? Windows users don’t want to edit the registry, they want it to just work. Chromebook users don’t want to edit the kernel - they want quick and easy. If HA is focused on getting more users this is critical. If it wants to be commercially viable it’s imperative. If it wants to be a tinkering platform, enough revenue to pay a few on-staff devs, then maybe not so much.
Regardless of the path forward, I personally enjoy and love using HA - it’s a bit more ‘cumbersome’ than I’d like it to be, but given current alternatives it is what it is and I’ll deal with it and certainly appreciate the hard work of devs, users and all contributors to make it better in the mean time.
TLDR; 90%+ of users just want EASY and FUNCTIONAL. I’m in this group. I’m willing to fight thru the upgrades/breaking changes, but I have better things to do that learn Linux / Docker / Venvs / ProxMox / etc. I just want to use the platform and have it work - otherwise, my wife is going to leave me b/c I’ll be spending more time ‘fixing HA’ than getting my @ss off the couch to flip the light switch myself…
Are you aware that you can run Home Assistant on a NUC without going the supervised route? After reading your post, it seems like you aren’t. Either way, any beginner can run home assistant with HassOS+Supervisor+HomeAssistant Core on a variety of hardware platforms. It seems like that’s all you would need. If you had gone that route, you wouldn’t have needed to learn linux or docker.
Yes, after reading thru all of this post I’m now aware that I am able to run HassOS on my NUC - though (from what I read) I’d need a M2 SSD <-> USB adapter and then to use Etcher to write the HassOS image to the SSD. From my understanding it would ‘limit’ me to running only HassOS on the NUC - though still allow me to use the ‘built-in’ add-ons (Samba, DuckDNS, etc.) w/out having to manually install. I also realize that this method is likely the best one for ‘NOOBs’ to take - as they’re likely not going to want to mess w/ running Docker and other containers anyhow…
I’ve not seen (though I haven’t looked that hard) a tutorial (video or otherwise) that walks a user through these steps - though (having used HA for 9-10 months now) I’m sure I could figure it out w/out too much difficulty. Being a middle-ground user (not a NOOB and not an expert [and having a fairly powerful system]) I still do much prefer the way it’s installed now (Supervised + generic Linux) as it gives me flexibility to do other things w/ the system in the future if I choose…
Definitely get there’s no one-size-fits-all install solution and not an ‘easy’ solve.
Same with me. I have had no issues with any addon in the core or community repos and I have a couple of other non HA repos as well running as addons.
Well… agree on your side, and I know if you are windows user and never touch Unix HASS can be very difficult, the documentation is so huge that in order to understand one will required to know where to ask. However its better than having to use something without no support at all. Furthermore learning basic Unix should be easy. Docker is actually make it even easier. I believe soon someone will create a script where on can just run and HASS and its addons will be installed.
I like the idea of HassOS - only OS running on hardware dedicated to home automation. It just seems like more reliable solution. But Hass under Ubuntu or whatever is more flaxible, that’s for sure.
Raspian is a OS Debian based optimized for Raspberry Pi. A raspberry pi 4B with 4GB RAM run Raspbian, Docker, Home Assistant and Kodi without problem. So it should be considered
yeah… and adding boot from SSD make it even better
Yeah. I use it with SSD
Would it make more sense to streamline the add-on container process for HA Core (Docker, no Supervisor) rather than make Supervisor work everywhere?
I’m looking at the Home Assistant Docker Hub, and of the 209 repos, many are already duplicate builds of add-on containers for multiple platforms.
So what if the add-ons which were officially supported had containers which were configured to plug into Docker and just work with HA? Whether someone had HA Core or Home Assistant (with the supported renamed Hassio – sorry, I sincerely can’t keep the naming convention straight yet), the add-on container would be pre-built to connect to the HA network, etc.
From reading(ish) through most of the 900+ comments, it seems like most generic Linux installers (including myself) care about having add-ons integrate simply, but would be fine if that simplicity didn’t require Supervisor, or the extra support burden of making Supervisor work everywhere.
Community add-ons could somehow be flagged as beta, exist in a separate repo (I recall Helm and Kubernetes having a process through which community submissions can eventually graduate to official status), or get linked to the plug-in developers’ repos (kinda like Discourse does).
Supervisor is great, and I doubt this would fulfill everyone’s desires, but even relative-Docker-noobs could probably handle managing add-ons through Portainer if the add-on containers were built generically enough to just plug into any functional Docker environment, but customized enough to connect to HA, and just go. Or maybe a docker-compose route (which Portainer is bad at, currently being limited to v2.0). Of course, advanced Docker users could still choose their own adventure.
I dunno, I’m probably missing things, and caveat emptor regarding thinking out loud. It just seems like Docker was built to solve exactly this kind of problem, and maybe leaning in to Docker everywhere would get more people closer to happiness than Supervisor everywhere.
The use of addons from supervisor does some nice things, already discussed:
- directly from HA UI you configure it
- directrly from HA UI you can see logs
- they can directly interact with HA itself
- directrly from HA UI you can go through the addon UI, so there is no need to use a reverse proxy or make a subdomains
- you can snapshot HA state, and all the addons for easy restore
It’s already that, but the problem is not addons side and not even the supervisor itself, it’s how supervisor is managed on generic linux to work.
I know there are nice things about supervisor. I said as much. I like nice things. But I don’t think there’s anything about the hold which implies everyone is going to get everything they want. Things are on the chopping block. The question has partly become what’s nice vs. what’s necessary.
Maybe “Pareto principle” or 80:20 is a better way of looking at it. Streamlining the Docker containers for add-ons has the potential to deliver high leverage returns.
I’m only offering a perspective I hadn’t seen (or maybe did and forgot 700 replies later) in this thread, and I don’t think that’s all there is to it. At the same time, not all of the nice things in your list would necessarily be sacrificed with a thoughtful container strategy.
We all know the reason.
this could be true, but we saw there wasn’t probably a real strategy.
True, but add-ons promote window shopping, at least for me.
While I’m typing, kudos to @ludeeus for HACS.