well when you used supervisor it will be contained inside HASS…
whats is “the manual installs”?
I’m talking about when you install HA directly onto your own OS, not using their VM, docker or supervisor.
If you use this, https://www.home-assistant.io/docs/installation/docker/, you have HA core running within Docker.
What you’ll lose is this
and all its fancy addons
i get it, but why don’t use docker if you want to use only HA core?
It’s a supported method use HA core with docker, check link below.
1 Like
So is not using docker. Just drop it, you have done enough to confuse this thread so far.
2 Likes
Why do you prefer JSON to YAML? You can’t write comments in JSON.
What are the differences between image/machine type NUC vs generic x86-64?
Isn’t NUC just a standard Intel CPU in a very small form factor?
2 Likes
I run suprvised on a RPi, So I’d be left with either:
- Buy a NUC or way more powerful hardware than a RPi and run with something like ProxMox
- Run it on core and loose Addons, meaning I have to manual install and configure everything I need
- Buy a second RPi for the other containers and things I already run and dedicate an RPi to HASSIO
From my point of view everything costs both money and time, but is clear that if I can save on these, running supervised on docker, it’s because someone else is spending for me, and basically for free.
So please, consider switching to a reasonably payed model where you can afford to maintain that install.
Always thankful to anyone involved in this little great masterpiece: “Home Assistant”
1 Like
My explanation is bad, what I’m trying to say is that if people will use just the core without the supervisor or addons (or maybe use alternative solutions to have the addons) the community that is using the official addons will be smaller and much less testing for official addons.
Just one example:
I currently use the Deconz as an addon, few month ago there was a problem with the addon implementation, I could report the bug, participate in testing the solutions. While I waited for this to be resolved I’ve found out that the Deconz is a available as a standalone docker image so I used that until a solution was ready and switched back to the official addon.
If I did not switch back to the official deconz addon than I can’t contribute to the testing of the official addon, bugs I found are directly related to the Deconz docker creator and not to the addon. This is why I switched back to the addon to test the solution so that others can use it
hi,
you have Python 3 parsers and minifiers that support single and multi line comments. JSON 5 supports comments.
btw, json was originally defined as data protocol soi bad or good idea was that you know what you are parsing ahead of time.
I do not prefer JSON, if it is up to me, I would use key = value pairs as INI in py 3 to setup HA.
Py 3 config parsers are flexible machines. You can customize it to no end, create a custom parsers, which I did in past. Again, I unfortunately work with Yaml in Azure, and Kuberneties among other platforms and honestly it is nightmare. I could line up hundred of colleagues to vouch for that.
I wish yaml could disappear for ever as it is incredible bad format to work with. Most of users of HA at the beginning had so much issues with initial configuration of HA in linux including me because I would forget to indent some crap somewhere. next thing the whole thing would choke on the bootstrap with errors that never explicitly point to YAML. Simply, crap idea stays crap idea no matter what.
People make fun of configparser and INI because it reminds them and me of Windows nightmares. I can tell you that there is nothing from Python object prospective so far that I could not stuff in INI and express it in string format. After that you can objectify it with ast module and viola you have, dict, list, complex template substitutions, anything you want.
best
1 Like
You could install the right image for your RPi from here and still have addons (it’s the install for beginners). The downside could be if you need access to the OS it won’t be possible. But if you don’t need it and everything else you run on your RPi is on docker, then you could install portainer addon and manage them from Home Assistant…
So no need to buy anything else
Well, yes, I run nothing that can’t be “dockerized” after all. Thanks for the input about portainer, it’s actually my best bet
Above thread is a good discussion about the supervisor, and strategically if there are existing tools we can leverage for the container management responsibilities. Kubernetes certainly an option, one that industry is pretty much fully aligned on, and @zaneclaes makes some excellent points to continue the discussion.
2 Likes
Not questioning the decision as we don’t know the whole story, but my first thought was “so Pascal has been working on this in his own time for most of the last 3.5 years, yet now that he’s a full-time employee of Nabu Casa working on Home Assistant he doesn’t have the time to do it anymore…?”
Yes, bur you can’t easily disable updates to supervisor which is what we will need to stay on an old version if this depreciation becomes reality.
What’s the best VM solution to use for Home Assistant using low spec hardware? I’ve got a Intel NUC Pentium Silver J5005 with 8GB RAM. I’ve tried VirtualBox, but seems to be quite resource intensive. I think I’ve probably got resource for one more VM such as Squeeezeserver (which will get rid of another RPi). But I wonder if there is something better?
Are there any other VM solutions that would be less resource intensive and work with Home Assistant? i.e. Proxmox?
Cheers
Will
Logical conclusion is that it’s likely the dev and maintenance workload has increased as well, and that the requirements have not remained constant nor proportionally constant to time available
I wonder if anyone’s using the Linux KVM module with the qcow files
Search this thread for Proxmox. There’s a couple of scripts by Whiskers’ in his github repo you can use to install it as a LXC container in proxmox, or as a vm in proxmox
2 Likes