Rebuilding HA server - some advice before I start

Ho all,

I’ve been running HAA (Hassio I think) for the past year and a bit on a Proxmox server, in a VM. No issues, all running good. Recently the server had some issues and I have decided it’s time to downscale.

So, I got rid of it and now using raspberry pis for the smaller, light weight servers. (Pi Hole, etc)

I decided to go with an Intel Nuc for Home Assistant (This one in particular )

I have a few questions:

  1. Since my backups (snapshots) are of Hassio - does that mean I can only use Hassio for the reinstall?
  2. If yes to the above, is there a decent way to run it on Linux without it being the main OS?
  3. I hope the above Nuc will be adequate enough for HA?

I am toying with the idea of installing Linux on the Nuc and then a couple of servers (Plex and maybe something else, as well as HA)

Or will it be best to dedicate the whole Nuc to just HA?

Appreciate any input of feedback.

You can run Debian on the NUC and then Supervised installation for HA, but from the little I understand from what you said so far, a Supervised install might not be the best option for you as it requires technical skill to maintain, unless your HA is not critical in which case it is good for learning.
If you want a HA installation you can depend on you should go the HAOS route. You can also go the HA Core route on Debian or Ubuntu, but you will lose the ability for add-ons. Basically any add-on you will have to install and connect from Docker yourself, which can be difficult if you do not have the technical skill, although you could learn.
To summarise if you want a HA installation you can depend upon HAOS is the only reliable route, as anything else is good for tinkering and learning; expect lot’s of downtime, things broken you need to fix, etc. It all depends who you are as a person.

1 Like

Thanks for the reply, much appreciated.

I’m pretty technical (Software developer by profession) so I don’t mind tinkering. If I were to install HA Core; does that mean I can’t restore from my snapshots, correct and would need to start from scratch?

If I was in your shoes I would start from a fresh and clean installation. Too many things changed in HA and like with any other system at some point a fresh install makes more sense.
Since you are a Dev, Core might be a better bet for you as any add-on functionality you can run in Docker, or even you can install Qemu and VMs on Debian.
I suggest you go with Debian, then install docker and then almost anything is possible.

1 Like

Core does not have backup capability, but you can do things on Debian/Ubuntu.
Another way is to run SAMBA and just let another machine connect and copy things. There are many ways you can address backup but not from inside Core.

1 Like

That is WAY overstating things.

I run HA Container and transitioned from HA in a venv a very long time ago.

I definitely can say that “lot’s of downtime, things broken you need to fix” has never been my experience solely based on the non-HAOS installation type. That idea has been promulgated (intentionally?) to not give brand new non-tech savvy users (the “target audience”) the idea that they can actually handle something they aren’t equipped to deal with. And so it saves headaches for the dev team dealing with people who have no idea how to maintain their systems.

If the OP is a minimally competent software guy then figuring HA out isn’t going to be a burden and HA Container loses them nothing if they can figure out Docker.

And the opposite has also been the case as well. There are downsides to running a OS Or Supervised install type - broken supervisor or OS updates is a big one. It doesn’t happen often but when it does it’s sudden and totally without warning.

That will never ever happen with a Container install.

And then there is the “locked down ecosystem” of OS and Supervised where if you do something to your system that HA deems unsatisfactory you get the “unhealthy” label and it literally prevents you from ever updating anything HA related until you fix that issue. It’s more likely that could happen in a Supervised install because you actually have some control over the host OS.

That will never happen with a Container install.

Correct.

However you can still just copy over all of the configuration files from your config directory (including the hidden .storage folder) and all of the config stuff will work as before. You just won’t have any add-on or supervisor config stuff.

I would recommend running HA Container (runs in Docker) over HA Core just simply due to the ease of maintenance. I can’t honestly think of any good reason to use a Core install over a Container install (and remember I used to run like that). You lose nothing at all running HA in docker. And you gain the ability to run other apps in docker as well.

3 Likes

Thanks for this; very useful as well.

I like the idea of just starting fresh again; something I have been wanting to do for a while now. I also do like the idea of running other apps on the Nuc and not just HA. I could also run a Plex server off there that connects to external storage and such things.

I’ll experiment with different setups and see which one suits me more and then take it from there.

Thanks all, I have a better idea now of what is possible and a good route to follow.

I have my system set up on Fedora Linux using podman for containers and systemd for basic process management, and while I’m tinkering now, I expect the setup to low maintenance once I’ve got it how I want it.

I’m planning on posting a detailed guide at some point, but am happy to give suggestions in the meantime. (Same basic things should apply to Debian / Ubuntu, although systemd and podman versions may be more out of date.)

I can’t speak to moving from Hassio backups, but once you have this set up, all future backups are “just use regular backup of whatever sort on one data directory and everything in it, and there you go”.

1 Like