CubitOom

What type of metadata is on a server attached to posts, comments, votes and such?

Yeah needed it for my monitor. I didn't want to figure out USB passthrough so I just installed Windows on a > 50,000 powered on hours HDD and booted from that. Then once I was done I put it about as far away as I could from my PC.

You are right. But proxmox and many of the other suggestions aren't vms either.

If you are dipping toes into containers with kvm and proxmox already, then perhaps you could jump into the deep end and look at kubernetes (k8s).

Even though you say you don't need production quality. It actually does a lot for you and you just need to learn a single API framework which has really great documentation.

Personally, if I am choosing a new service to host. One of my first metrics in that decision is how well is it documented.

You could also go the simple route and use docker to make containers. However making your own containers is optional as most services have pre built ones that you can use.

You could even use auto scaling to run your cluster with just 1 node if you don't need it to be highly available with a lot of 9s in uptime.

The trickiest thing with K8s is the networking, certs and DNS but there are services you can host to take care of that for you. I use istio for networking, cert-manager for certs and external-dns for DNS.

I would recommend trying out k8s first on a cloud provider like digital ocean or linode. Managing your own k8s control plane on bare metal has its own complications.

Ok, hmm I wonder how much work it would be to implement it using open source models. I think the hardest part would be translating the voice instructions to an API call that HA can use correctly.

Then there is the whole hardware issue to fix too. I do know that some SOCs are getting good at running 7B parameter models locally but the cost is still probably going to be prohibitive.

So what is Home Assistant using for this?

If I were to build it myself I'd probably over complicate it by using multiple llm agents on a local server. Probably use whisper to do the speech to text and then Mistral fine tuned on the Rosetta code dataset to send the API calls to HA. However that wouldnt keep it from always listening to me and trying to interpret what I say into a command for HA. Is that just a prompting issue for whisper or would I need another agent to turn on whisper?

I could maybe get this to run without specialized hardware like a GPU but it would be better to have something for the llms to be a bit more responsive.

Use whatever you want for personal. But I would suggest trying to use containers for hosting if you haven't already. It really blows the idea of needing a stable OS out of the water since you can just declare everything you want in a config file and tear down and spin up with the app you need ready in less than a minute.

You can use Ubuntu still of course in a container. But things get really interesting when you use smaller attack surface distros like Alpine, BusyBox, or even a distroless container.

I just got super excited and then really disappointed in less than a second...Now I know how my wife feels.

It's still two separate passwords so I think it qualifies as 2 factors.

But yes the password manager has one gpg key which only has one passphrase used to decrypt the passwords saved in the password manager. So if that was compromised then so would all passwords

I'm just using my password manager in place of the authenticator app.

So rather than using an app like Google authenticator or Authy to see what the new random sequence is for the MFA, my password manager stores that QR as a string and will display the same random sequence that a normal MFA app would.

They key difference is that my MFA is synced across any device that I have configured my password manager on using the same cryptographic keys and version control history.

So if my phone is dead, lost, or stolen, I can still access my banking account via MFA as normal.

I suppose it brings up the idea of what a "factor" is in how it's used for MFA. If a factor is supposed to be a different device, a different app on the same device as your password manager, or just a different passphrase that's constantly changing.

Thank you, these are all great points.